WO2003041292A1

WO2003041292A1 - Interference canceling method, interference canceling device and receiving device

Info

Publication number: WO2003041292A1
Application number: PCT/SE2002/002022
Authority: WO
Inventors: Jonas Karlsson; Masayuki Ariyoshi
Original assignee: Telefonaktiebolaget Lm Ericsson (Publ)
Priority date: 2001-11-09
Filing date: 2002-11-05
Publication date: 2003-05-15
Also published as: JP2003152682A

Abstract

To more accurately reduce multiple access interference by appropriately generating interference replica signals in accordance with the state of the signal being handled in each stage. In a multi-user receiver using a multi-stage interference canceller, searchers 4S1U1-4S3Uk are provided for path detection of the respective multipaths of first through k-th users in each stage, and path information is obtained from the same input reception signals as the input reception signals supplied to the IGU's (interference replica generating units) 5S1U1-5S3Uk of each stage. At the IGU's 5S1U1-5S3Uk, rake combining reception, replica signal generation and the like are performed based on path information supplied from the searchers 4S1U1-4S3Uk. In the first and second stages, interference replica signals of the respective users are generated, and in the third stage, reception information data is obtained by rake combining reception from reception signals with the multiple access interference removed by means of these interference replica signals.

Description

IN TERPERENCE CANCELING METHOD, INTERFERENCE CANCELING DEVICE

AND RECEIVING DEVICE

BACKGROUND OF THE INVENTION Field of the Invention

The present invention relates to a communication technology in communications by a CDMA (Code Division Multiple Access) system for obtaining received information by canceling interference in the received signal.

Description of the Related Art

The CDMA system is a communication system in which the same frequency band is shared by a plurality of channels, the channels being separated by spreading the spectrum with different spreading codes, and has in recent years come into use in radio communications in cellular radio communication systems and the like with the aim of making more effective use of frequencies. In radio communications, the signals sent from each station on the transmitting side arrive at the stations on the receiving side with mutual interference, so that in a radio communication system according to the CDMA system, if the number of channels sharing the same frequency band, in other words, the number of users simultaneously communicating increases, it is difficult to isolate the signal transmitted from a specific station by despreading on the receiving side. The interference due to signals from other stations which affects signals from stations on the transmitting side is generally called multiple access interference (MAI), and in order to increase the capacity of the radio communication system according to the CDMA system, some kind of technology is required to reduce the effects of this multiple access interference.

As such a technology for reducing the effects of multiple access interference, there is the multi-stage interference canceller which is mainly used in a multi-user receiver in a radio base station. The multi-stage interference canceller is composed of multiple stages for canceling multiple access interference due to a plurality of mobile stations on the transmitting side, each stage generating a signal from a respective mobile station as an interference replica signal for respective users, these interference replica signals being used to cancel out the multiple access interference in the received signal. In a cellular radio communication system, a plurality of paths with different transmission wave propagation times are formed according to the topography and the like between each mobile station and the radio base station, and the signal arriving from one mobile station at the radio base station is dispersed into a plurality of signal components which have come via multiple paths of different delay times, so that multi-user receivers generally make use of diversity reception technology known as rake combining reception wherein each dispersed signal component is despread and these are combined at maximum proportion.

Fig. 10 is a drawing showing a structural example of a multi-user receiver using a general multi-stage interference canceller according to the conventional art. This multi-user receiver takes k users (k mobile stations) which are ordered from first to &-th, has a structure provided with a three-stage parallel interference canceller, and as shown in the drawing, comprises an antenna 10, a radio signal processing portion 20, buffers 30S1 and 30S2, k searchers 40Ul-40Uk, k x 3 IGU's (interference replica generation units) 50SlUl-50SlUk, 50S2Ul-50S2Uk and 50S3Ul-50S3Uk, subtractors 60S1 and 60S2, and k * 2 adders 70SlUl-70S2Uk and 70S3Ul-70S3Uk. In the reference numbers, "S" and "U" correspond to the number of the stage (S) to which each constituent element belongs, and the number of the user (U) being handled (for example, a constituent element having appended an "S1U1" on the reference number belongs to the first stage, and handles the signal of the first user).

At the mobile stations of each user on the transmitting side, not shown, a predetermined pilot signal known with respect to the receiving side, a data signal representing various information not known with respect to the receiving side and the like are CDMA-modulated using a spreading code assigned to each user, and the CDMA-modulated signal is high-frequency modulated into a radio signal of the shared frequency band by means of a known carrier wave and transmitted. The antenna 10 receives the radio signals sent from the mobile stations of each user and supplies them to a radio signal processing portion 20. The radio signal processing portion 20 outputs the signals from the antenna 10 as received signals of a base band suitable for processing by the interference cancellers to the buffer 30S1, searchers 40Ul-40Uk and IGU's 50SlUl-50SlUk.

The buffer 30S1 is a delay buffer for delaying the received signals for the time required to generate an interference replica signal in the first stage. The buffer 30S2 is a delay buffer for delaying the received signals from the buffer 30S1 for the time required for interference cancellation by the interference replica signal generated in the first stage, and for generation of the interference replica signal in the second stage.

The searchers 40U1, 40U2, . . ., 40Uk respectively perform despreading of the received signals while shifting the timing of the spreading codes assigned to the first, second, . . ., k-t users, and detect the reception levels at the respective spreading code timings for the signals from first, second, . . ., k-t users contained in the received signals. The above-described pilot signals transmitted from the mobile stations of respective users are examples of signals capable of being used for detection of these reception levels. When using the pilot signals for detection of the reception level, reception levels at the respective spreading code timings are detected by the searchers 40U1, 40U2, . . ., 40Uk for the signals of the portions corresponding to the pilot signals from among the signals of the first, second, . . ., k-ύi users contained in the received signals. The relationship between spreading code timing and reception level obtained in this way indicates the composition of delayed waves coming via the multiple paths between the user and multi-user receiver (mobile station and radio base station), and is generally referred to as the delay profile. The searchers 40U1, 40U2, . . .,

40Uk, for example, find the peaks corresponding to the multipaths in the delay profile by discriminating peaks of a reception level exceeding a predetermined threshold value in the delay profile, and detect the spreading code timings of the thus validated peaks as the respective path timings of the multipaths.

In a radio communication system according to the CDMA system, the path timings of the received signals must be acquired in order to perform despreading and demodulation on the receiving side, and the searchers 40Ul-40Uk correspond to the means for acquiring the path timings. This type of searcher is also generally called a multipath searcher, one being provided with respect to each user in the first stage of the multi-stage interference canceller as indicated by the searchers 40Ul-40Uk in the drawing, for detecting information concerning the radio transmission path through which the radio signal from each user has arrived, and supplying path information based on the thus detected information to the IGU of each stage corresponding to the same user. The path information supplied by the searcher includes the detected path timings and path (peak) number, as well as information such as the order of reception power of each path (the order of size of the peak reception level) and the like as needed, this information being used to assign fingers for performing rake combining reception (multipath diversity reception) at each stage and designating the synchronized reception timings (spreading code timings for reception synchronized with the paths which are rake-combined). The searchers 40U1, 40U2, . . ., 40Uk corresponding to the first, second, . . ., k-t users shown in the drawing supply path information based on the results of the above peak validation with respect to the IGU's 50S1U1, 50S2U1 and 50S3U1, IGU's 50S1U2, 50S2U2 and 50S3U2, . . ., IGU's 50SlUk, 50S2Uk and 50S3Uk respectively corresponding to the same users.

The IGU's 50SlUl-50S3Uk are IGU's (interference replica generating units) for generating respective interference replica signals. These IGU's have the same basic structure with regard to circuitry and the like, but as the input reception signal for generation of each interference replica signal, uses the signal from the prior stage (the IGU of the first stage uses the received signal from the radio signal processing portion 20), and uses the spreading code assigned to the corresponding user for despreading and the like.

The basic structure of such an IGU is shown in Fig. 11. As shown in this drawing, an IGU comprises n rake detector fingers 50FDl-50FDn each having a channel estimator 50a, a despreader 50b and a channel compensator 50c; a rake combiner 50d; a preliminary decider 50e; n replica generating fingers 50FGl-50FGn each having a respreader 50f and a channel decompensator 50g; and an adder 50h. The rake detecting fingers 50FDl-50FDn are fingers for detection for the purpose of rake combining, each executing despreading processes such as despreading procedures. The replica generating fingers 50FGl-50FGn are fingers for generating respective signal replicas (replicas of the signals received on each path) forming an interference replica signal, each executing a respreading process for once again spread modulating the signal respectively obtained by rake combining and preliminary decision processes (details to be described below) at the rake detecting fingers 50FDl-50FDn, the rake combiner 50d and the preliminary decider 50e. In the normal conventional IGU's, the numbers of fingers for detection for rake combining and for replica generation, as in the rake detecting fingers 50FDl-50FDn and the replica generating fingers 50RGl-50RGn as shown in the drawing, are the same, and the operating rake detecting fingers and replica generating fingers have a one-to-one correspondence. The path information from the searchers as described above is supplied to the channel estimator 50a, despreader 50b, channel compensator 50c and channel decompensator 50g in a group of corresponding fingers for the information on each path, and is designated by the assignment of fingers and the synchronized reception timing at each finger. The channel estimator 50a despreads the pilot signal in an input received signal in accordance with a path timing in the path information, and compares the obtained pilot signal with a known pilot signal. Due to this comparison, the channel estimator 50a estimates the channel variation (channel variation due to fading) such as phase rotation or amplitude changes undergone by the received signal on the radio transmission path according to the designated timing, and supplies the results of the estimation to the channel compensator 50c and channel decompensator 50g. The despreader 50b despreads the input received signal (input received signal including pilot signal and data signal) in accordance with the path timing in the above path information, and outputs the result to the channel compensator 50c. The channel compensator 50c performs channel compensation to correct the phase, amplitude and the like of the input received signal despread by the despreader 50b to its state prior to undergoing estimated channel variations based on the estimated results from the channel estimator 50a. As a result, the signal following the channel compensator becomes a signal which has been weighted for maximum proportion combining depending on the received power of the path which has been despread. Additionally, the channel compensator 50c adjusts the output timing of the signal after channel compensation based on the path information, such that the input received signal of each path having undergone despreading and channel compensation is outputted from all rake detecting fingers to the rake combiner 50d at the same timing.

The rake combiner 50d adds the signals outputted from all of the rake detecting fingers. As a result, the received signals from the respective paths are despread and rake-combined. The preliminary decider 50e is a decision means for performing a preliminary deciding operation for deciding on the signal levels after rake combining, and outputs signals in accordance with the results of the decision to the replica generating fingers.

The values indicated by the decision results are temporary received data decision values for generation of interference replica signals, but in the IGU of the final stage, the value is taken as the final decision output of the received information data (as indicated by the dashed line in the drawing).

The preliminary deciding process at the preliminary decider 50e may, if needed, sometimes be performed by a soft decision. Additionally, with regard to the preliminary decision symbol for generating the interference replica signal, a method of improving the performance as an interference canceller receiver by multiplying with a suppression coefficient (a factor smaller than 1) prior to inputting into the replica generating fingers is known, and if this method is used, then an operator for multiplication of the suppression coefficient can be provided as appropriate between the preliminary decider 50e and the replica generating fingers 50FGl-50FGn.

The respreader 5 Of again spread-converts the signal in accordance with the results from the preliminary decider 50e, and outputs the result to the channel decompensator 50g. The channel decompensator 50g performs a channel decompensation (a reverse compensation to return phase rotations or changes in amplitude to the state of the original input reception signal) to undo the channel compensation performed by the channel compensator 50c based on the estimated results from the channel estimator 50a. As a result, the signal after channel decompensation becomes a signal component in accordance with the reception power of the respreading path, and is returned to the level of the original input reception signal. Then, the channel decompensator 50g adjusts the output timing of the channel decompensated signal based on the aforementioned path information, and arranges the signal replicas from the respective replica generating fingers so as to be outputted to the adder 5 Oh at the same timing as when in the original input reception signal. The adder 50h adds the signal replicas from the respective replica generating fingers, and outputs the result as an interference replica signal.

The IGU's 50SlUl-50SlUk, 50S2Ul-50-S2Uk and 50S3Ul-50S3Uk in Fig. 10 all have the basic structure described above. The subtractor 60S1 receives the interference replica signals generated by the IGU's 50SlUl-50SlUk, and subtracts the respective interference replica signals from the signal received via the buffer 30S1. The subtractor 60S2 receives the interference replica signals generated by the IGU's 50S2Ul-50S2Uk, and subtracts the respective interference replica signals from the signal received via the buffers 30S1 and 30S2. The adders 70S2U1, 70S2U2, . . ., 70S2Uk respectively add the interference replica signals generated by the IGU's 50S1U1, 50S1U2, . . ., 50SlUk to the residual signal left after subtraction by the subtractor 60S1, and output them as input reception signals to the IGU's 50S2U1, 50S2U2, . . ., 50S2Uk. The adders 70S3U1,

70S3U2, . . ., 70S3Uk respectively add the interference replica signals generated by the IGU's 50S2U1, 50S2U2, . . ., 50S2Uk with the residual signal after subtraction by the subtractor 60S2, and outputs them as the input reception signals to the IGU's 50S3U1, 50S3U2, . . ., 50S3Uk. In the above-described structure, when a radio signal transmitted from the mobile station of each user is received at the antenna 10, the received signal is supplied through the radio signal processing portion 20 to the buffer 30S1, searchers 40U1-40U2 and IGU's 50SlUl-50SlUk of the first stage. In the searchers 40U1, 40U2, . . ., 40Uk which have received the reception signal, the reception levels of the pilot signals and the like from the first, second, . . ., k-th users contained in the reception signal are respectively detected, and path information of the multipath is obtained for each user. In the IGU's 50S1U1, 50S1U2, . . ., 50SlUk, the path information from the respective searchers 40U1, 40U2, . . ., 40Uk is received, and the despreading process at each rake detecting finger, rake combimng at the rake combiner 50d, the preliminary decision process at the preliminary decider 50e and the respreading process at each replica generating finger are performed in order.

Here, in the respective rake detecting fingers, channel estimation by the channel estimator 50a, despreading by the despreader 50b and channel compensation by the channel compensator 50c are performed in accordance with the respective path timings in the received path information, thereby performing the despreading process for each path of each user.

Additionally, even in corresponding replica generating fingers, despreading by the despreader

5 Of and channel decompensation by the channel decompensator 50g are performed in accordance with the respective path timings in the same path information, thereby performing the respreading process for each path of each user. As a result, the signal from each user in the reception signal undergoes rake combining and is once demodulated, and the respective signal replicas of the multipaths based on the demodulated signals form the respective users are added to the adders 50h of the IGU's 50SlUl-50SlUk. Here, groups of rake detection fingers and replica generating fingers which have not been supplied path information are not activated.

The multipath signal replicas added at the respective adders 5 Oh are outputted to the respective adders 70S2U1, 70S2U2, . . ., 70S2Uk of the second stage as interference replica signals of the respective users generated in the first stage. Additionally, these interference replica signals are subtracted at the subtractor 60S1 from the original reception signal coming via the buffer 30S1. These post-subtraction residual signals become residual signals which have undergone interference cancellation by subtracting the interference replica signals of all users generated in the first stage from the overall reception signal, and these are also outputted to the respective adders 70S2U1, 70S2U2, . . ., 70S2Uk of the second stage.

At the adders 70S2U1, 70S2U2, . . ., 70S2Uk which have received the residual signals from the subtractor 60S1, the residual signals are added to the interference replica signals respectively from the IGU's 50S1U1, 50S1U2, . . ., 50SlUk. As a result, the signals outputted from the adders 70S2U1, 70S2U2, . . ., 70S2Uk become reception signals from the first, second, . . ., k-th users from which the interference replica signals (multiple access interference) from other users respectively generated in the first stage have been removed. The IGU's 50S2U1, 50S2U2, . . ., 50S2Uk of the second stage take these reception signals from which the multiple access interference has been cancelled as the input reception signals, and as in the above-described first stage, generate interference replica signals in accordance with path information from the searchers 40U1, 40U2, . . ., 40Uk. As a result, in the second stage, interference replica signals that are more accurate than those of the first stage are generated, and the thus generated interference replica signals are outputted to the respective adders 70S3U1, 70S3U2, . . ., 70S3Uk and subtractor 60S2 of the third stage. At the subtractor 60S2, the respective interference replica signals are subtracted from the original reception signal coming via the buffers 30S1 and 30S2, and the residual signal obtained by subtracting the interference replica signals of all of the users generated in the second stage from the overall reception signal is outputted to the adders 70S3U1, 70S3U2, . . ., 70S3Uk of the third stage.

At the adders 70S3U1, 70S3U2, . . ., 70S3Uk which have received the residual signal from the adder 60S2, the residual signals are respectively added to the interference replica signals from the IGU's 50S2U1, 50S2U2, . . ., 50S2Uk. As a result, the signals outputted from the adders 70S3U1, 70S3U2, . . ., 70S3Uk become reception signals from the first, second, . . ., k-t users from which the interference replica signals (multiple access interference) of the other users generated in the second stage have been removed. The IGU's 50S3U1, 50S3U2, . . ., 50S3Uk of the third stage take the reception signal from which the multiple access interference has been cancelled as input reception signals, these respectively undergoing a despreading process, rake combining and preliminary decision in accordance with the path information from the searchers 40U1, 40U2, . . ., 40Uk as described above for the first stage and second stage. As a result, in the IGU's 50S3U1, 50S3U2, . . ., 50S3Uk, demodulation is performed by an input reception signal from which the multiple access interference has been cancelled more accurately than in the second stage, whereby the final decision output of the reception information data is performed.

In this way, in the multi-stage interference canceller, interference replica signals which are more accurate in the latter stages than in the former stages are generated for the respective users, and these are subtracted fro the overall reception signal as interference signals. Then, by repeating this process for many stages, a more accurate multiple access interference cancellation is ensured, so that in the final stage, final reception information data is obtained as a multi-user receiver. A multi-user detection receiver having this type of interference canceller capability is disclosed in Mamoru Sawahashi et al., "Pilot symbol-assisted coherent multistage interference canceller using recursive channel estimation for DS-CDMA mobile radio", IEICE Trans. Commun., vol. E79-B, no. 9, September 1996.

Problems to be Solved by the Invention

It should be noted that common rake receivers which perform rake combining reception have a pulse searcher for detecting a reception level of a pilot signal or the like from a user whose signal is to be received from among those buried in the received signal that includes multiple access interference in order to detect the reception level along each path from the user whose signal is to be received. That is, a rake receiver normally has a path searcher as means for path detection of the multipath of a specific user from among the received signals including multiple access interference, and is structured such that a demodulator portion performs a demodulation process such as despreading in accordance with a peak timing or the like of the detected reception level. On the other hand, by once again spread modulating the signal of received information data obtained by the demodulator portion in the rake combining reception as described above, an interference replica signal is generated in each stage to cancel multiple access interference. Accordingly, each stage of the multi-stage interference canceller can be composed of a portion of a demodulator in the rake combiner, a portion for generating an interference replica signal and a portion for canceling multiple access interference.

On the other hand, the above-described conventional multi-user receiver is also a rake receiver of sorts for performing rake combining reception, the rake detection fingers 50FDl-50FDn of the first stage, the rake combiner 50d and the preliminary decider 50e corresponding to the demodulator portion in a common rake receiver. Therefore, in order to form a multi-stage interference canceller, the modulator portion, portion for generating interference replica signals and the portion for canceling multiple access interference can be formed into a plurality of stages. Thus, in the conventional multi-user receiver described above, a multi-stage interference canceller is formed by providing the replica generating fingers 50FGl-50FGn and the adder portion at each stage, and providing subtractorsόOSl and 60S2 as well as adders 70S2Ul-70S2Uk and 70S3Ul-70S3Uk as portions for canceling multiple access interference at each stage.

As a result, the conventional multi-stage interference canceller has a structure as described above, wherein a searcher for detecting the path timings of multipaths and the like is provided for each user in the first stage. The path timings of the multipaths detected by these searchers are used as synchronous reception timings for rake combining reception in the first stage, and are used as path timings of the signal replicas in interference replica signal generation. Additionally, in the second and subsequent stages as well, the same path timings detected in the searchers of the first stage are similarly used for both rake combining reception and interference signal generation, and the processes repeated for canceling the multiple access interference.

However, the input reception signal to the first stage still contains noise and multiple access interference in the baseband when received after having undergone only radio signal processing, so that the SIR (signal-to-interference power ratio) and SNIR

(signal-to-noise-and-interference ratio) are lower than those of the input reception signal to the latter stages. Therefore, in the searchers of the first stage, there are cases in which path detection is performed on the multipaths with the SIR and the like in a worse state than in the latter stages, so that accurate detection results cannot be obtained. Furthermore, since the detection results in a degraded signal state continue to be used in the stages subsequent to the second stage, the processing in all of the stages is affected. For this reason, if inaccurate path information containing errors such as lost path timings or erroneous data is obtained in the searchers, then these errors will affect all of the stages, thereby making it difficult to correctly cancel out multiple access interference. Such circumstances are particularly likely to occur if the channel is in a state of heavy multipath fading, hence reducing the performance of multi-stage interference cancellers.

Additionally, since the interference replica signals generated in each stage are preliminary replica signals to be subtracted from the reception signal as multiple access interference in order to enable subsequent procedures to be performed, the generation of inaccurate interference replica signals has an adverse effect on the subsequent processing (in the latter stages, not only is the signal processing of that user affected, but the signal processing of other users is also adversely influenced). In contrast, in conventional multi-stage interference cancellers, the path information detected and acquired in the searchers of the first stage are used similarly for both rake combining reception and interference replica signal generation in each stage. That is, as described above, the same number of rake detection fingers and replica generation fingers operate in correlation with each other, signal replicas corresponding to all paths in the same path information are uniformly generated at each stage, to generate an interference replica signal combining all of these signal replicas. For this reason, there is a possibility of erroneous signal replicas being included among the generated interference replica signals, and the probability of this happening is especially high in heavy multipath fading conditions. Additionally, these erroneous signal replicas act effectively as noise to adversely affect the subsequent processing.

SUMMARY OF THE INVENTION

The present invention has been made in view of these considerations, and has the object of offering interference canceling technology enabling appropriate generation of interference replica signals according to the state of the signals such as SIR being handled in each stage, and capable of more accurately reducing multiple access interference without inaccurate multipath information detected under poor signal conditions affecting the latter stages.

Additionally, the present invention has the object of achieving multiple access interference cancellation which prevents generation of signal replicas which can act effectively as noise in each stage, and eliminating the adverse influences from previous stages on subsequent processing.

In order to achieve this object, in the interference canceling method according to the present invention, a reception signal containing signals from a plurality of transmitting stations is received, replica signals corresponding to signals from the respective transmitting stations are generated, and the generated replica signals are used to perform an interference canceling procedure on a signal from a transmitting station other than said transmitting stations, wherein path information on the transmission paths through which the signals from said transmitting stations have arrived is obtained from signals which have undergone said canceling procedure at least once, and said replica signals are generated based on the obtained path information. In the interference canceling method according to the present invention, it is possible to make it so that at the starting time of said canceling procedure, path information is obtained from said reception signal, and said replica signals are respectively generated based on this obtained path information, and subsequent to a predetermined time after said starting time, path information is obtained from signals which have undergone said canceling procedure at least once, and said replica signals are respectively generated based on this obtained path information.

In another interference canceling method according to the present invention, a reception signal containing signals from a plurality of transmitting stations is received, and in a plurality of stages, replica signals corresponding to signals from said transmitting stations are respectively generated based on path information on the transmission paths through which the signals from said transmitting stations have arrived, and interference canceling procedures are sequentially performed on a signal from a transmitting station other than said transmitting stations using the generated replica signals, such that in a first stage, path information is obtained from said reception signal, and said replica signals are respectively generated based on this obtained path information; and in a second stage, path information is obtained from signals which have undergone said canceling procedure in the previous stage, and said replica signals are respectively generated based on this obtained path information.

Additionally, an interference canceling device according to the present invention is an interference canceling device for receiving a reception signal containing signals from a plurality of transmitting stations, and in a plurality of stages, respectively generating replica signals corresponding to the signals from the transmitting stations, and sequentially performing interference canceling procedures on a signal from a transmitting station other than said transmitting stations using the generated replica signals, comprising detecting means provided in each of said plurality of stages, for respectively detecting path information on the transmission paths through which the signals from said transmitting stations have arrived based on said reception signal or signals which have undergone said canceling procedure in a previous stage; and generating means provided in each of said plurality of stages, for respectively generating said replica signals based on the path information detected by said detecting means provided in the same stage.

Another interference canceling device according to the present invention is an interference canceling device for receiving a reception signal containing signals from a plurality of transmitting stations, and in a plurality of stages, respectively generating replica signals corresponding to the signals from the transmitting stations, and sequentially performing interference canceling procedures on a signal from a transmitting station other than said transmitting stations using the generated replica signals, comprising detecting means for detecting path information on the transmission paths through which the signals from said transmitting station have arrived based on supplied signals; generating means provided in each of said plurality of stages, for respectively generating said replica signals based on path information detected by said detecting means; and selecting means for selecting and supplying to said detecting means said reception signal or signals which have undergone said canceling procedure in any one of the stages.

In the above-described interference canceling devices, the generating means can be means for demodulating said reception signal or signals which have undergone said canceling procedure in a previous stage by means of a rake combining procedure, and generating said replica signals from the demodulated signals; and said detecting means can separately detect first path information for the purpose of said rake combining procedure and second path information for the purpose of generating said replica signals, and supply these to said generating means. In this case, the detecting means can be such as to set first and second threshold values, and respectively detect said first and second path information based on said first and second threshold values. Alternatively, the detecting means can be such as to set first and second threshold values, detect said first path information based on said first threshold value, and detect said second path information based on said second threshold value from among said first path information which has been detected.

Additionally, in the above-described interference canceling device, said generating means can be means for demodulating said reception signal or signals which have undergone said canceling procedure in a previous stage and generating said replica signals from the demodulated signals; said detecting means can detect path information for the purpose of said rake combining procedure and supply them to said generating means; and said generating means can perform said rake combining procedure based on path information supplied from said detecting means, select paths for generating replica signals from among the rake combining paths in said rake combining procedure, and generate said replica signals based on these selected paths. In the above-described interference canceling device, said detecting means can be such as to detect path information based on correlation values between said reception signal or signals which have undergone said canceling procedure, and spreading codes used for modulation of signals in said transmitting stations.

Furthermore, in the above-described interference canceling device, said detecting means can be such as to supply information indicating paths in the detected path information to said detecting means provided in latter stages, and determine the range of signals for which path information is detected based on said information supplied from said detecting means provided in previous stages.

Additionally, in a receiving device according to the present invention, information transmitted by said transmitting stations can be obtained from said reception signal from which the interference has been canceled by interference canceling devices as described above.

According to the present invention, path detection is performed in each stage, so that imprecise multipath information which has been detected under poor signal conditions will not affect the latter stages, and appropriate interference replica signals can be generated in accordance with the conditions of the signals handled in each stage such as the SIR or the like, thereby enabling the multiple access interference to be more accurately reduced.

Additionally, since the paths for generating signal replicas are set independently of the paths for rake combining, the generation of signal replicas which can effectively act as noise in each stage is prevented, thus enabling the detrimental influence of preceding stages to be eliminated from subsequent processing and achieving the cancellation of multiple access interference by means of appropriate processing.

BRIEF DESCRIPTION OF THE DRAWINGS Fig. 1 is a diagram showing a basic structural example of a multi-user receiver using a multi-stage interference canceller according to a first embodiment of the present invention. Fig. 2 is a diagram showing a structural example of a searcher for executing a conventional searcher algorithm.

Fig. 3 is a diagram showing the structure of a searcher according to a first structural embodiment capable of being employed in the basic structural example of Fig. 1. Fig. 4 is a diagram showing the structure of an IGU capable of being employed for the searcher structure of Fig. 3.

Fig. 5 is a diagram showing the structure of a searcher according to a second structural embodiment capable of being employed in the basic structural example of Fig. 1.

Fig. 6 is a diagram showing the structure of an IGU according to a third structural embodiment capable of being employed in the basic structural example of Fig. 1.

Fig. 7 is a diagram showing a structural example of a multi-user receiver according to an embodiment wherein the search window is changed in the basic structural example of Fig. 1.

Fig. 8 is a diagram showing an example of search window changes in the multi-user receiver of Fig. 7.

Fig. 9 is a diagram showing a structural example of a multi-user receiver using a multi-stage interference canceller according to the second embodiment of the present invention.

Fig. 10 is a diagram showing a structural example of a multi-user receiver using a normal multi-stage interference canceller according to the conventional art.

Fig. 11 is a diagram showing an example of the basic structure of a conventional IGU.

DESCRIPTION OF THE PREFERRED EMBODIMENTS <First Embodiment

A. Basic Structural Embodiment

(1) Structure

Herebelow, an embodiment of the present invention shall be described with reference to the drawings. Fig. 1 is a drawing showing the basic structure of a multi-user receiver using a multi-stage interference canceller according to a first embodiment of the present invention. The present embodiment is a form in which each stage of the multi-stage interference canceller has a path searcher function, and Fig. 1 shows the basic structure of a multi-user receiver for the case where a 3 -stage parallel interference canceller is used as an example of a basic structure applying this form. The present multi-user receiver is a receiving device for a plurality of users used in a radio base station or the like of a cellular radio communication system according to the CDMA system, and is capable of accommodating k users (k mobile stations) numbered from first to A;-th. As a constituent element for receiving signals from these k users and removing interference in the received signals to obtain received information, the present multi-user receiver, as shown in the drawing, comprises an antenna 1, a radio signal processing portion 2, buffers 3S1 and 3S2, k x 3 searchers 4SlUl-4SlUk, 4S2Ul-4S2Uk and 4S3Ul-4S3Uk, k x 3 IGU's (interference replica generation units) 5SlUl-5SlUk, 5S2Ul-5S2Uk and 5S3Ul-5S3Uk, subtracters 6S1 and 6S2, and k x 2 adders 7S2Ul-7S2Uk and 7S3Ul-7S3Uk.

Of these constituents, the first stage of the multi-stage interference canceller is composed of the buffer 3S1, searchers 4SlUl-4SlUk and IGU's 5SlUl-5SlUk corresponding to the respective users (handling the channels of the respective users) and a sutractor 6S1. The second stage is composed of the buffer 3S2, searchers 4S2Ul-4S2Uk, IGU's 5S2Ul-5S2Uk and adders 7S2Ul-7S2Uk corresponding to the respective users and a sutractor 6S2. The third stage is composed of searchers 4S3Ul-4S3Uk, IGU's 5S3Ul-5S3Uk and adders 7S2Ul-7S2Uk corresponding to the respective users and a sutractor 6S2. In the reference numbers, the letters "S" and "U" correspond respectively to the number of the stage (S) to which that element belongs and the number of the user (S) being processed (for example, an element containing "S1U1" would belong to the first stage and would process the signals relating to a first user).

The antenna 1 is a high frequency antenna fro receiving radio signals. At the mobile stations (not shown) of the respective users which are the transmitting side stations, predetermined pilot signals known to the receiving side, data signals representing various types of information not known to the receiving side and the like are CDMA modulated using a spreading code assigned to each user, and these CDMA modulated signals are high frequency modulated to radio signals in a common frequency band by means of a predetermined carrier wave and transmitted. The antenna 1 receives the radio signals transmitted from the mobile stations of the respective users and supplies them to the radio signal processing portion 2.

The radio signal processing portion 2 is composed of a designated amplifier, local oscillator, mixer, AID converter and the like, and outputs the signals supplied from the antenna 1 as reception signals in a base band appropriate for processing by interference cancellers to the buffer 3S1, searchers 4S1U1, 4S1U2, . . ., 4SlUk, and the IGU's 5S1U1,

5S1U2, . . ., 5SlUk. At the A/D converters and the like of this radio signal processing portion 2, the signals from the antenna are sampled at a predetermined oversampling rate, and the aforementioned reception signals are outputted in a digital signal format having a resolution of at least the chip rate of the spreading code (e.g. an integer multiple of the chip rate).

The buffer 3S1 is a buffer for storing received signals from the radio signal processing portion 2, and sending the received signals out after a delay equivalent to the time required for interference replica signal generation in the first stage. The buffer 3S2 is a buffer for storing the received signals from the buffer 2S1, and sending the received signals out after a delay equivalent to the time required for interference cancellation by the interference replica signals generated in the first stage and generation of the interference replica signals of the second stage. These buffers 3S1 and 3S2 are delay means for adjusting the inter-stage transmission time of the original received signals which have not undergone processing such as despreading, and ensure that the received signals are inputted to the subtracters 6S1 and 6S2 to be explained below at the same timings as the interference replica signals.

The searchers 4SlUl-4S3Uk are detecting means for path detection of multipaths such as by capturing the path timings, each acquiring a delay profile from respective input reception signals to perform path detection, and acquiring path information concerning the radio transmission path through which the radio signal from each user has come. Here, as the input reception signals, the searchers 4SlUl-4SlUk of the first stage use the reception signal from the radio signal processing portion 2, the searchers 4S2U1, 4S2U2, . . ., 4S2Uk of the second stage use the respective output signals from the adders 7S2U1, 7S2U2, . . ., 7S2Uk to be described below, and the searchers 4S3U1, 4S3U2, . . ., 4S3Uk of the third stage use the respective output signals from the adders 7S3U1, 7S3U2, . . ., 7S3Uk to be described below. The searchers 4S1U1-4S3U1, 4S1U2-4S3U2, . . ., 4SlUk-4S3Uk perform a despreading process on their input reception signals while shifting the timing of the spreading codes assigned respectively to the first, second, . . ., k- users, and detect the reception levels of the signals from the first, second, . . ., k-th users contained in the input reception signal by means of a correlation value between the input reception signal and a spreading code at the respective spreading code timings. As a result, a delay profile indicating the relationship between the spreading code timings and the reception levels is obtained, and the searchers 4SlUl-4S3Uk validate the reception level peaks which exceed a predetermined threshold value within the acquired delay profile, detecting the spreading code timings of the validated peaks to be the path timings of the multipaths.

The above-described pilot signals sent from the mobile stations of the respective users are an example of a signal capable of being used to detect the reception level in the searchers 4SlUl-4S3Uk. When using the pilot signals for detection of the reception level, of the signals from the first, second, . . ., &-th users contained in the reception signal, the reception levels of the signals at the portion corresponding to the pilot signals are respectively detected at respective spreading code timings by the searchers 4S1U1-4S3U1, 4S1U2-4S3U2, . . ., 4SlUk-4S3Uk. In this case, since the pilot signals are signals that are known by the multi-user receiver which is on the receiving side, the detection of the reception level may employ a detection process which allows for increased gain by means of cumulative addition over a predetermined interval (corresponding to a certain number of symbols or slots). Additionally, the peak validation of the reception level by comparison with the threshold value is just one example of a process for validating peaks corresponding to the multipaths in the delay profile, and other peak validation methods can be used as needed.

As a form for the searchers 4SlUl-4S3Uk in the multi-user receiver, it is possible to employ various structures in accordance with the feature of the present embodiment in which they are provided on each stage, and the structure of the IGU's 5SlUl-5S3Uk described below will change according to the structure employed, but it is possible to use existing searcher structures. Therefore, as a basic structure, the description shall be continued under the assumption that currently existing searchers will be used, and this shall be followed by structural examples of searchers capable of being employed in accordance with the features of the present embodiment.

The algorithms of existing searchers are capable of being divided into portions for delay profile calculation and peak validation. A structural example of a searcher for performing such an algorithm is shown in Fig. 2. The illustrated searcher has a structure for despreading the input reception signal by means of a matched filter 4MF, performing a delay profile calculation in the delay profile calculating portion 4DPC using the despread signal, and performing peak validation in the preliminary candidate path determining portion 4PCPS, SIR calculating portion 4SIRC, threshold value setting portion 4TS and path selecting portion 4PS.

The matched filter 4MF despreads an input reception signal while shifting the timing of the spreading code across a predetermined search window (predetermined interval on the delay time axis) to obtain a delay profile, and outputs the despread signal to the delay profile calculating portion 4DPC. The delay profile calculating portion 4DPC uses the despread signal from the matched filter 4MF to sequentially perform coherent accumulation, absolute square computation and power accumulation (non-coherent accumulation), thus to perform delay profile calculation.

Here, during coherent accumulation, a correlation value (a vector value such as voltage) due to the despreading process is sequentially acquired as an IDP (initial delay profile) each predetermined period (for example, a predetermined symbol length) from the beginning of the search window, and the IDP's of a plurality of periods contained in each standard number of consecutive symbols or slots are coherent accumulated (vector added) with each corresponding element in each period. Thus, by coherent accumulation of the results despread over a plurality of periods, an AccIDP (accumulated IDP) with increased SN ration is obtained. In the absolute square computation, the square of the absolute value of the AccIDP obtained in the coherent accumulation is calculated, thereby obtaining an RDP

(real delay profile) which is a representation of the AccIDP in the power plane. In the power accumulation process, the RDP's which are sequentially obtained by absolute square computation are integrated over a standard number (regardless of phase), to obtain an averaged PDP (power delay profile: a delay profile representing the reception level in the power plane).

The PDP obtained in this way is used to perform peak validation, the peak validation process being composed of steps of preliminary candidate path determination, SIR calculation, threshold value setting and path selection, and these steps are executed by the structure in the latter stages of the delay profile calculating portion 4DPC. The preliminary candidate path determining portion 4PCPS selects a fixed number of peaks having a maximum value in the PDP, thereby to determine paths corresponding to the spreading code timings of the selected peaks as preliminary candidate paths for peak validation. The SIR calculating portion 4SIRC calculates the interference power for use in path determination by averaging the PDP elements aside from the peaks of the preliminary candidate paths determined by the preliminary candidate path determining portion 4PCPS, and calculates the SIR at respective points in time with the peak reception level of each preliminary candidate path as S (signal power) and the calculated interference power as I (interference power). The threshold value setting portion 4TS receives the SIR and interference power values calculated by the SIR calculating portion 4SIRC, determines a TF (threshold factor) as a function of that SIR, and sets a threshold value which is TF times the interference power level. The path selecting portion 4PS compares the peak reception levels of the preliminary candidate paths from the prelimmary candidate path determining portion 4PCPS with the threshold value set at the threshold value setting portion 4TS, and selects preliminary candidate paths of peak reception levels exceeding the threshold value as effective candidate paths. As a result, the spreading code timings (delay times) of the effective candidate paths and the number of these candidate paths are outputted as detection results (path information) for the searchers from the path selecting portion 4PS, and the spreading code timings of the candidate paths are made to designate synchronous reception timings for rake combining reception.

With regard to algorithms for searchers with this type of structure, reference may be made to the disclosure of International Publication No. WO 00/04648 entitled "Adaptive Path

Selection Threshold Setting for DS-CDMA Receiver".

With regard to the searchers 4SlUl-4S3Uk of Fig. 1, it is possible to use existing searcher structures such as the searchers shown in Fig. 2. That is, the structure of existing searcher portions is provided in each stage, and as mentioned above, the searchers

4SlUl-4SlUk of the first stage take the received signals from the radio signal processing portion 2 as the input reception signals, the searchers 4S2U1, 4S2U2, . . ., 4S2Uk of the second stage respectively take the output signals from the adders 7S2U1, 7S2U2, . . ., 7S2Uk as the input reception signals, and the searchers 4S3U1, 4S3U2, . . ., 4S3Uk of the third stage take the output signals from the adders 7S3U1, 7S3U2, . . ., 7S3Uk as the input reception signals. Furthermore, the searchers 4S1U, 4S1U2, . . ., 4SlUk of the first stage output path information respectively to the IGU's 5S1U1, 5S1U2, . . ., 5SlUk of the same first stage, the searchers 4S12U, 4S2U2, . . ., 4S2Uk of the second stage output path information respectively to the IGU's 5S2U1, 5S2U2, . . ., 5S2Uk of the same second stage, and the searchers 4S3U, 4S3U2, . . ., 4S3Uk of the fourth stage output path information respectively to the IGU's 5S3U1, 5S3U2, . . ., 5S3Uk of the same third stage.

The IGU's 5SlUl-5SlUk are IGU's (interference replica generating units) for generating respective interference replica signals. With regard to these IGU's 5SlUl-5S3Uk, it is possible to employ various structures in accordance with the form of the searchers 4SlUl-4S3Uk, but the description here shall assume the case where existing IGU's are used in compliance with the existing searchers described above, and other structures shall be described later. With regard to the existing IGU's, it is possible, for example, to use those such as shown in Fig. 11. That is, as shown in Fig. 11, the structures of existing IGU portions are provided in each stage, with the IGU's 5SlUl-5SlUk of the first stage taking the received signals from the radio signal processing portion 2 as the input reception signals, the IGU's 5S2U1, 5S2U2, . . ., 5S2Uk of the second stage taking the respective output signals of the adders 7S2U1, 7S2U2, . . ., 7S2Uk as the input reception signals, and the IGU's 5S3U1, 5S3U2, . . ., 5S3Uk of the third stage taking the respective output signals of the adders 7S3U1,

7S3U2, . . ., 7S3Uk as the input reception signals. Furthermore, the IGU's 5S1U1, 5S1U2, . . ., 5SlUk of the first stage respectively receive path information from the searchers 4S1U1, 4S1U2, . . ., 4SlUk of the same first stage, the IGU's 5S2U1, 5S2U2, . . ., 5S2Uk of the second stage respectively receive path information from the searchers 4S2U1, 4S2U2, . . ., 4S2Uk of the same second stage, and the IGU's 5S3U1, 5S3U2, . . ., 5S3Uk of the third stage respectively receive path information from the searchers 4S3U1, 4S3U2, . . ., 43 Uk of the same third stage. As a result, the multi-user receiver has a structure such that the searchers and IGU's corresponding to the respective users in each stage use the same input reception signals (input reception signals with the same signal state such as SIR), so that the IGU's are supplied with path information from searchers in the same stage as themselves.

The subtracters 6S1 and 6S2 are computing means for generating residual signals after cancellation of interference from the reception signal and interference replica signals, for canceling the interference replica signals generated in each stage from the received signals. Therefore, in the third stage (the final stage from which the final received information data is outputted) in which interference signals are not generated, there are no constituent elements corresponding to the computing means. The subtractor 6S1 receives all interference replica signals generated in the IGU's 5SlUl-5SlUk of the first stage, and subtracts these interference replica signals from the reception signal received through the buffer 3S1. The subtracter 6S2 receives all interference replica signals generated in the IGU's 5S2Ul-5S2Uk of the second stage, and subtracts these interference replica signals from the reception signal received through the buffers 3S1 and 3S2.

The adders 7S2Ul-7S2Uk and 7S3Ul-7S3Uk are computing means for generating the input reception signal for the channel of each user from the interference replica signals generated in the previous stage and the residual signal left after interference cancellation. Therefore, in the first stage which does not have a previous stage and using the original reception signal as the input reception signal for each user channel, there are no constituent elements corresponding to the computing means. The adders 7S2U1, 7S2U2, . . ., 7S2Uk respectively add together the interference replica signals generated in the IGU's 5S1U1, 5S1U2, . . ., 5SlUk and the residual signals left after subtraction at the subtractor 6S1, and output the results as the input reception signals to the searcher 4S2U1 and IGU 5S2U1, searcher 4S2U2 and IGU 5S2U2, . . ., searcher 4S2Uk and IGU 5S2Uk. The adders 7S3U1, 7S3U2, . . ., 7S3Uk respectively add together the interference replica signals generated by the

IGU's 5S2U1, 5S2U2, . . ., 5S2Uk and the residual signals left after subtraction at the subtractor 6S2, and output the results as the input reception signals to the searcher 4S3U1 and IGU 5S3U1, searcher 4S3U2 and IGU 5S3U2, . . ., searcher 4S3Uk and IGU 5S3Uk.

(2) Operations

Next, the operations due to the above-described structure shall be described. When the radio signals transmitted from the mobile stations of the respective users are received by the antenna 1, the reception signals are supplied through the radio signal processing portion 2 to the buffer 3S1, searchers 4S1U1-4S1U2 and IGU's 5SlUl-5SlUk of the first stage. At the searchers 4S1U1, 4S1U2, . . ., 4SlUk which have received the reception signal, the reception levels of pilot signals or the like respectively from the first, second, . . ., Ar-th users contained in the received reception signal are detected to perform path detection of the multipaths, thereby obtaining path information for the respective users. When the above-described structure of Fig. 2 is used for the searchers 4SlUl-4SlUk, the reception signal is despread by the matched filter 4MF, the delay profile is calculated by the delay profile calculating portion 4DPC, and path information is obtained by peak validation at the preliminary candidate path determining portion 4PCPS, the SIR calculating portion 4SIRC, the threshold value setting portion 4TS and the path selecting portion 4PS. Then, the path information obtained by the searchers 4S1U1, 4S1U2, . . ., 4SlUk is supplied respectively to only the IGU's 5S1U1, 5S1U2, . . ., 5SlUk of the first stage, and replica signal generation is performed based on the path information supplied to the respective IGU's 5S1U1, 5S1U2, . . ., 5SlUk. When the above-described structure of Fig. 11 is used for the IGU's 5SlUl-5SlUk, the despreading process at each rake detection finger, the rake combining at the rake combiner 50d, the preliminary decision process at the preliminary decider 50e and the respreading process at each replica generating finger are sequentially performed, after which the signal replicas of the respective multipaths based on the signals from the respective users which have undergone rake combining and have been temporarily demodulated are added by the adder 50h to form the interference replica signal.

The interference replica signals of the respective users generated by the IGU's 5S1U1, 5S1U2, . . ., 5SlUk are outputted as interference replica signals from the first stage to the adders 7S2U1, 7S2U2, . . ., 7S2Uk of the second stage. Additionally, these interference replica signals are subtracted from the original reception signal coming via the buffer 3S1 at the subtractor 6S1, and the interference-cancelled residual signal with all of the interference replica signals generated in the first stage subtracted from the overall reception signal is outputted to the respective adders 7S2U1, 7S2Us, . . ., 7S2Uk of the second stage.

At the adders 7S2U1, 7S2U2, . . ., 7S2Uk which have received the residual signals from the subtractor 6S1, the residual signals are added to the interference replica signals from the IGU's 5S1U1, 5S1U2, . . ., 5SlUk. As a result, the signals outputted for the adders 7S2U1, 7S2U2, . . ., 7S2Uk become the reception signals from the first, second, . . ., k- users with the interference replica signals (multiple access interference) from other users generated in the first stage cancelled, and these are supplied to the searcher 4S2U1 and IGU

5S2U1, searcher 4S2U2 and IGU 5S2U2, . . ., searcher 4S2Uk and IGU5S2Uk of the second stage.

In the searchers 4S2Ul-4S2Uk which have received these reception signals, the multipath information is obtained by the same procedure as the above-described searchers

4SlUl-4SlUk. However, the input reception signals used at this time are signals in which multiple access interference has once been cancelled by the interference replica signals from the first stage. Consequently, path detection of the multipaths can be performed with the SIR and SNIR of the input reception signals in a higher state than in the first stage, due to which detection results more accurate than in the first stage can be expected. Then, the IGU's 5S2U1, 5S2U2, . . ., 5S2Uk use the same input reception signals as the searchers 4S2U1, 4S2U2, . . ., 4S2Uk based on the path information from the detection results to generate interference replica signals.

As a result, in the second stage, interference replica signals are generated based on path information which is more accurate than in the first stage, thereby obtaining interference replica signals which are more accurate than in the case where input reception signals in which multiple access interference has been cancelled only once are used in the IGU's of the second stage. Additionally, since path detection of the multipaths is newly performed on the searchers 4S2Ul-4S2Uk using the same input reception signals as for the IGU's 5S2Ul-5S2Uk, independently of the first stage, so that the detection results of the searchers 4SlUl-4SlUk of the first stage will not affect the path detection of the searchers 4S2Ul-4S2Uk of the second stage.

The interference replica signals generated by the IGU's 5S2U1, 5S2U2, . . ., 5S2Uk in this way are respectively outputted to the adders 7S3U1, 7S3U2, . . ., 7S3Uk of the third stage and the subtractor 6S2. At the subtractor 6S2, the interference replica signals are subtracted from the original reception signal coming via the buffers 3S1 and 3S2, and the residual signal with the interference replica signals of all users generated in the second stage subtracted from the overall reception signal is outputted to the adders 70S3U1, 70S3U2, . . ., 70S3Uk of the third stage.

At the adders 7S3U1, 7S3U2, . . ., 7S3Uk which have received the residual signals from the subtractor 6S2, the residual signal is added to the interference replica signals respectively from the IGU's 5S2U1, 5S2U2, . . ., 5S2Uk. As a result, the signals outputted from the adders 7S3U1, 7S3U2, . . ., 7S3Uk respectively become reception signals from the first, second, . . ., k-th users with the interference replica signals (multiple access interference) of other users generated in the second stage cancelled, and these are supplied to the searcher 4S3U1 and IGU 5S3U1, searcher 4S3U2 and IGU 5S3U2, . . ., searcher 4S3Uk and IGU

5S3Uk ofthe third stage.

In the searchers 4S3Ul-43Suk which have received the reception signal, multipath information is obtained by the same procedure as in the above-descrived searchers 4SlUl-4SlUk and 4S2U1-4S2U However, the input reception signal used at this time is a signal which has undergone cancellation of multiple access interference twice through processing at the first stage and second stage as well as the adders 7S3U1-7S3U Therefore, path detection of the multipaths can be performed with the SIR or the like of the input reception signals in a higher state than the first stage and second stage, so that detection results which are more accurate than the first and second stages can be expected. Then, based on the path information due to the detection results of the IGU's 5S3U1, 5S3U2, . . ., 5S3Uk, a despreading process, rake combining and preliminary decision process are performed using the same input reception signals as in the searchers 4S3U1, 4S3U2, . . ., 4S3Uk.

As a result, in the third stage, a despreading process, rake combining and preliminary decision process can be performed based on path information that is more accurate than in the first and second stages, and more accurate preliminary decision results can be obtained than in the case where input reception signals which have only undergone multiple access interference cancellation twice are used in the IGU's of the third stage. Additionally, since path detection of the multipaths is performed anew at the searchers 4S3Ul-4S3Uk using the same input reception signals as the IGU's 5S3Ul-5S3Uk independently of the first and second stages, the detection results of the searchers 4SlUl-4SlUk of the first stage and the searchers 4S2Ul-4S2Uk of the second stage themselves will not affect the processing in the third stage.

The preliminary decision results obtained by the IGU's 5S3Ul-5S3Uk in this way are taken as the final decision outputs by the present multiuser receiver. That is, the preliminary decision results (hard decision results or soft decision results) of the IGU's 5S3U1,

5S3U2, . . ., 5S3Uk in the final third stage are outputted as demodulated reception information from the first, second, . . ., k-th users.

In the present multi-user receiver, the processing by both the searchers and IGU's as described above is performed in parallel at each stage, such that the searchers in the latter stages perform path detection for multipaths with a better signal state than the former stages. Additionally, since the detection results in each stage are used in only the IGU's of that stage, the detection results in the former stages will not affect the path detection in latter stages. Furthermore, since the searchers of the respective stages detect multipath information from the same input reception signals as the IGU's to which they themselves supply the path information, the information such as the detected path timing is more suited to the processing in the IGU's. As a result, the path detection of the multipaths and the generation of interference signals is suitably performed in each stage, thereby enabling multiple access interference to be more accurately reduced.

B. Structural Embodiments of Searchers and IGU's (1) First Structural Embodiment Structure of Searcher Next, the various structural embodiments which can be employed for the searchers and IGU's in the above-described basic structure shall be explained. Fig. 3 is a diagram showing the structure of a searcher in a first structural embodiment. As shown in this drawing, the searcher according to the present structural embodiment comprises a matched filter 4a, a delay profile calculating portion 4b, a preliminary candidate path determining portion 4c, an SIR calculating portion 4d, a threshold value calculating portion 4e, a peak validating portion 4f and a peak validating portion 4g.

The matched filter 4a despreads input reception signals while shifting the timing of the spreading code assigned to the corresponding user over a predetermined search window (a predetermined interval on the delay time axis) to obtain a delay profile, and outputs the despread signals to the delay profile calculating portion 4b. The delay profile calculating portion 4b detects the reception levels of pilot signals of the like contained in the input reception signals by means of the despread signals from the matched filter 4a at each spreading code timing to obtain delay profiles, and supplies the obtained delay profiles to the preliminary candidate path determining portion 4c and SIR calculating portion 4d. In the calculating process for obtaining delay profiles in the delay profile calculating portion 4b, coherent accumulation or power accumulation as in the above-described delay profile calculating portion 4DPC, or optimization of these accumulation processes responsive to fading variations or the like can be appropriately employed as needed.

The preliminary candidate path determining portion 4c selects a certain number of reception level peaks exhibiting a maximum value in the delay profile from the delay profile calculating portion 4b, and determines paths corresponding to the spreading code timings of the selected peaks as preliminary candidate paths for peak validation. This preliminary candidate path determining portion 4c supplies information giving notification of the determined preliminary candidate paths to the SIR calculating portion 4d, and supplies spreading code timings and peak reception levels of the determined preliminary candidate paths and to both the peak validating portion 4f and the peak validating portion 4g. The SIR calculating portion 4d averages the elements of the reception level aside from the peaks of the preliminary candidate paths determined by the preliminary candidate path determining portion 4c in the delay profiles from the delay profile calculating portion 4b, thereby to compute the interference power for use in determining the paths. Then, the SIR is computed with the peak reception levels of the respective preliminary candidate paths as S (signal power) and the computed interference power as I (interference power), and this SIR is supplied to the threshold value calculating portion 4e.

The threshold value calculating portion 4e separately computes a rake threshold value ThRAKE for determining the paths to use for performing rake combining reception and a replica threshold value ThREP for determining the paths for generating signal replicas.

That is, based on the SIR from the SIR calculating portion 4d, the threshold value calculating portion 4e sets a peak reception level threshold value (rake threshold value ThRAKE) for determining path timings for the rake detecting fingers of the IGU, and separately sets a peak reception level threshold value (replica threshold value ThREP) for determining the path timings for the replica generating fingers of the IGU. In the present structural embodiment, as an example of a structural embodiment for calculating the replica threshold value ThREP at the threshold value calculating portion 4e a form is employed wherein basically a high value is set for a low SIR and a low value is set for a high SIR.

For example when the SIR is low, it is preferable to increase the signal power by rake combimng of signals from a large number of paths, but if signal replicas corresponding to too many paths are generated, then there is a considerable chance that erroneous signal replicas will be contained in the interference replica signal. Therefore, with respect to a low SIR, the threshold value calculating portion 4e sets a higher threshold value than the rake threshold value ThRAKE for the replica threshold value ThREP. On the other hand, when the SIR is sufficiently high, the precision of the interference replica signals can be improved by generating signal replicas corresponding to a large number of paths. Therefore, with respect to SIR's that are sufficiently high, the threshold value calculating portion 4e sets a threshold value lower than the rake threshold value ThRAKE as the replica threshold value ThREP. In this way, the optimum standards for selection of paths for rake combining and paths for replica generation are not always the same, so that the threshold value calculating portion 4e performs calculations for setting the rake threshold value ThRAKe and replica threshold value ThREP to mutually independent threshold values as optimum values according to the SIR or the like. Then, the rake threshold value ThRAKE is supplied to the peak validating portion 4f, and the replica threshold value ThREP is supplied to the peak validating portion 4g.

The peak validating portion 4f receives spreading code timings and peak reception levels for the respective preliminary candidate paths from the prelimmary candidate path determining portion 4c, and determines whether or not the peak reception levels of these preliminary candidate paths exceed the rake threshold value ThRAKE from the threshold value calculating portion 4e. Based on the results of this determination, the peak validating portion 4f validates the spreading code timings of the preak reception levels exceeding the rake threshold value ThRAKE as the path timings of the rake combining path, and outputs the candidate path information including the path timings and number of rake combining paths (number of peak reception levels exceeding the rake threshold value ThRAKE) to the IGU (rake detecting fingers). In the candidate path information, it is possible to include other information such as reception power order (order of size of peak reception level) of the respective rake combining paths as needed.

The peak validating portion 4g receives spreading code timings and peak reception levels of the respective preliminary candidate paths from the preliminary candidate path determining portion 4c, and determines whether or not the peak reception levels of these preliminary candidate paths exceed the replica threshold value ThREP from the threshold value calculating portion 4e. Based on the results of this determination, the peak validating portion 4g validates the spreading code timings of the respective peak reception levels exceeding the replica threshold value ThREP as the path timings of the signal replica generation paths, and outputs replica path information including the path timings and number of signal replica generation paths (number of peak reception levels exceeding the replica threshold value ThREP) to the IGU (replica generating finger). The path information for the replicas can also include other information as needed such as reception power order (order of size of peak reception level) of the respective signal replica generation paths. Structure of IGU

On the other hand, as a structure for an IGU corresponding to this type of searcher structure, it is possible to employ that which is shown in Fig. 4. The IGU of Fig. 4 comprises n rake detecting fingers 5FDl-5FDn each having a channel estimator 5a, a despreader 5b and a channel compensator 5c, a rake combiner 5d, a preliminary decider 5e, m replica generating fingers 5FGl-5FGm each having a despreader 5f and a channel decompensator 5g, and an adder 5h. The rake detecting fingers 5FDl-5FDn are fingers for performing detection for rake combining, each performing a despreading process such as despreading procedures. The replica generating fingers 5FGl-5FGm are fingers for generating the individual signal replicas (of each path) forming an interference replica signal, each performing respreading processes for once again spread modulating the signals obtained through the rake combining and preliminary decision process (details described below) at the rake detecting fingers 5FDl-5FDn, the rake combiner 5d and the preliminary decider 5e.

In this IGU, there is no need for the number n of rake detecting fingers and the number m of replica generating fingers to be the same, and there is not necessarily a one-to-one correspondence between the operating rake detecting fingers and replica generating fingers. This is due to the fact that, as shown in the drawing, the rake detecting fingers 5FDl-5FDn operate based on candidate path information, whereas the replica generating fingers 5FGl-5FGm operate based on replica path information which is independent of the candidate path information. That is, the above-described candidate path information from the searchers is supplied to the channel estimator 5a, despreader 5b and channel compensator 5c of a single rake detecting finger for each set of information relating to each rake combimng path, whereas the above-described replica path information is supplied to the channel decompensator 5g of a single replica generating finger for each set of information relating to each replica generation path. As a result, the assignment of rake detecting fingers and despreading timings at the rake detecting fingers are performed separately from the assignment of replica generating fingers and signal replica transmission timings at the replica generating fingers. The rake detecting fingers which are not supplied with candidate path information are not activated, and the replica generating fingers which are not supplied with replica path information are likewise not activated.

The channel estimator 5a despreads the pilot signals in the input reception signals in accordance with path timings in the candidate path information, and compares the resulting pilot signals with known pilot signals. By means of this comparison, the channel estimator 5 a estimates channel variations (channel variations due to fading) such as phase rotation or amplitude changes undergone by the reception signals during radio transmission on the rake combining paths of designated timings, and supplies the estimation results to the channel compensator 5c and channel decompensator 5g.

Here, the channel estimator 5a supplies estimation results to a channel decompensator 5g of a replica generating finger having as the replica generation path the same path as the rake combining path of the rake detecting finger to which that channel estimator 5a belongs. For example, in the assignment of rake detecting fingers (supply of information relating to each rake combining path), rake combimng paths are assigned to the rake detecting fingers 5FD1, 5FD2, 5FD3, . . . in the order of reception power, and in the assignment of replica generating fingers (supply of information relating to each replica generation path), replica generation paths are assigned to the replica generating fingers 5FG1,

5FG2, 5FG3, . . . in the order of reception power. Alternatively, in the assignment of rake detecting fingers, the rake combining paths are assigned to the rake detecting fingers 5FD1, 5FD2, 5FD3, . . . in the order of shortness (or length) of the delay times of the path timings, and in the assignment of replica generating fingers, the replica generation paths are assigned to the replica generating fingers 5FG1, 5FG2, 5FG3, . . . in the order of shortness (or length) of the delay times of the path timings. By appropriately performing fingers assignment in this way, the estimation results from the channel estimators 5a of the respective rake combimng fingers are supplied to the channel decompensators 5g of replica generating fingers in which, for example, the finger numbers in the reference numbers are the same.

The despreader 5b despreads an input reception signal (input reception signal including the pilot signal, data signal and the like) in accordance with the path timing in the candidate path information, and outputs the result to the channel compensator 5c. The channel compensator 5c performs channel compensation for correcting the phase, amplitude and the like of the input reception signal despread by the despreader 5b to the assumed original state prior to undergoing channel variations based on estimation results from the channel estimator 5a. As a result, the channel compensated signal becomes a signal which has been weighted for maximum ratio combining in accordance with the reception powers of the despread rake combining paths. Additionally, the channel compensator 5c adjusts the output timing of the channel compensated signals based on the above-described candidate path information, so that the input reception signals for the respective combining paths which have undergone despreading and channel compensation are outputted from all rake detecting fingers to the rake combiner 5d at the same time.

The rake combiner 5d adds all of the signals outputted from the respective rake detecting fingers. As a result, the reception signals which have arrived via the respective rake combining paths are respectively despread and rake-combined. The preliminary decider 5e is a decision means for performing a preliminary deciding process for hard decision of the signal levels after rake combining, and outputs a signal in accordance with the decision result to the replica generating finger. While the value indicated by this decision result is a temporary received data decided value for use in interference replica signal generation, in the IGU of the final stage, the value is made the final decision output of the received information data (as indicated by the dashed lines in the drawing).

If required, the preliminary decision process at the preliminary decider 5e can be performed as a soft decision. Additionally, with regard to the preliminary decision symbols for generating interference replica signals, a method is known for improving the performance as an interference canceller receiver by multiplying a suppression factor (a coefficient less than 1) before inputting to the replica generating fingers. This method can also be used in the IGU (IGU's of Fig. 4 and of Fig. 6 to be described below) of the present embodiment, and for example, it is possible to provide a processor for multiplication of the suppression factor as appropriate between the preliminary decider 5e and the replica generating fingers 5FGl-5FGn, so that a signal which has been multiplied by the suppression factor is inputted to the replica generating fingers 5FG1 -5FGn.

The despreader 5f once again spread modulates the signal in accordance with the decision result from the preliminary decider 5e, and outputs this to the channel decompensator 5g. the channel decompensator 5g performs a channel decompensation (reverse compensation to return the input reception signal to the original state with the phase rotated and amplitude changed) to undo the channel compensation performed by the channel compensator 5c based on the estimation results from the channel estimator 5a.

Here, the estimation results from the above-described channel estimator 5 a are supplied to a channel decompensator 5g of a replica generating finger having a replica generation path which is the same as the rake combimng path as described above. Therefore, when there are more replica generation paths than rake combining paths, there will be some channel decompensators 5g which are not supplied with channel estimation results from the channel estimator 5 a operating based on the candidate path information. Therefore, with regard to the replica generation paths which are to be processed by the channel decompensator 5g, channel estimation using this replica path information is performed as needed, to supply channel estimation results to the channel decompensator 5g. For example, replica path information of replica generation paths which do not have a corresponding rake combimng path based on the candidate path information and replica path information obtained by the searchers is sent to the channel estimators 5a of the rake detecting fingers (channel estimators which do not supply candidate path information), and these channel estimators 5a perform channel estimation based on the relevant replica generation path and supply the estimation results to the channel decompensator 5g performing channel decompensation for that replica generation path.

In this way, channel estimation results are sent to all of the channel decompensators 5g to which replica path information is supplied, and the channel decompenators 5g perform channel decompensation with respect to each replica generation path (with regard to replica generation paths without corresponding rake combining paths, the channel compensation performed by the channel compensator 5 c is not undone, but a similar form of reverse compensation for returning the phase rotation and amplitude change to that of the original input reception signal is performed). By means of this channel decompensation, the respread signals of the respective replica generation paths are divided into signal components according to the reception powers of the respective replica generation paths, and are returned to their levels in the original input reception signal. Then, the channel decompensator 5g adjusts the output timing of the channel decompensated signal based on the replica path information, so that the signal replicas from the respective replica generating fingers are outputted to the adder 5h at the same timing as in the original input reception signal.

The adder 5h adds the signal replicas from the respective replica generating fingers, and outputs the result as an interference replica signal.

Operations In the multiuser receiver of Fig. 1, if searchers with the structure of Fig. 3 are used for the searchers 4SlUl-4S3Uk, and IGU's with the structure of Fig. 4 are used for the IGU's 5SlUl-5S3Uk, then the overall operations will progress in parallel as in the case of the basic structure described above, but the path detection operations at the respective searchers and the interference generation operations at the respective IGU's in each stage will take a different form from the above-described operations.

That is, when an input reception signal is supplied to the searchers in each stage, the input reception signal is sequentially despread in the matched filter 4a, and the despread signals over the range of the search window are supplied to the delay profile calculating portion 4b. At the delay profile calculating portion 4b, delay profiles are computed from the supplied despread signals, and supplied to the preliminary candidate path determining portion 4c and SIR calculating portion 4d. At the preliminary candidate path determining portion 4c which has received such a delay profile, preliminary candidate paths are determined, and the SIR calculating portion 4d is notified of each preliminary candidate path, the spreading code timings and peak reception levels of the respective preliminary candidate paths being sent to both the peak validating portion 4f and the peak validating portion 4g. At the SIR calculating portion 4d which has been notified of the temporary candidate paths, the SIR is computed based on these preliminary candidate paths and the delay profiles from the delay profile calculating portions 4b, and the result supplied to the threshold value calculating portion 4e.

Then, at the threshold value calculating portion 4e, the determining threshold value for the rake combining path and the determining threshold value for the replica generation path are separately calculated based on the SIR from the SIR calculating portion 4d. As a result, a rake threshold value ThRAKE and replica threshold value ThREP which are mutually independent are determined, the rake threshold value ThRAKE being supplied to the peak validating portion 4f and the replica threshold value ThREP being supplied to the peak validating portion 4g.

At the peak validating portion 4f, the peak reception level of each preliminary candidate path supplied from the preliminary candidate path determining portion 4c is compared with the rake threshold value ThRAKE from the threshold value calculating portion 4e, and the spreading code timings of peak reception levels exceeding the rake threshold value ThRAKE are determined as path timings of the rake combining path. Based on these determination results, the peak validating portion 4f supplies candidate path information including path timings of the rake combining paths and the number of rake combining paths to the rake detecting fingers of the IGU.

On the other hand, at the peak validating portion 4g, the peak reception levels of the preliminary candidate paths supplied from the preliminary candidate path determining portion 4c are compared with the replica threshold values ThREP from the threshold value calculating portion 4e, and the spreading code timings of peak reception levels exceeding the replica threshold value ThREP are determined as path timings of the replica generation path. Based on these determination results, the peak validating portion 4g supplies replica path information including path timings of the replica generation paths and the number of replica generation paths to the replica generating fingers of the IGU.

As a result, at the peak validating portion 4g, replica generation paths are selected with a threshold value independent of the above-described rake threshold value ThRAKE as the criterion for determination, and the path information of the selected replica generation path is supplied to the IGU as replica path information separate from the above-described candidate path information. That is, path information for rake combining and path information for replica generation are supplied to the IGU's independently, and processing is begun at the IGU's using the same input reception signals as the input reception signals on which the path information has been obtained.

First, at the rake detecting fingers which have received the candidate path information, channel estimation by the channel estimator 5a, despreading by the despreader

5b and channel compensation by the channel compensator 5c are performed in accordance with the respective path timings in the candidate path information, and despreading processes are performed for each rake combining path. Then, the signals outputted from the respective rake detecting fingers are combined by the rake combiner 5d, and a preliminary decision process is performed by the preliminary decider 5e by means of a combined signal.

Here, one of the rake detecting fingers handles the despreading process relating to one of the rake combining paths in the candidate path information. Therefore, the number of operating rake detecting fingers is equal to the number of rake combining paths designated by the candidate path information, so that reception signals corresponding to a multipath of that number are rake-combined, and demodulated by the preliminary decision process at the preliminary decider 5e. As a result, the received information data is once demodulated, and a signal indicating this received information data is supplied from the preliminary decider 5e to the replica generating fingers.

When the signals from the preliminary decider 5e are received at the replica generating finger side, a respreading process is performed at the replica generating fingers to which replica path information has been supplied from the above-described peak validating portion 4g. That is, at the replica generating fingers, despreading by the despreader 5f and channel decompensation by the channel decompensator 5g are performed, thereby to execute the despreading process for each replica generation path. Then, signal replicas which have undergone the respreading process are outputted from the replica generating fingers in accordance with the respective path timings in the replica path information, added together in the adder 5h, and outputted as an interference replica signal.

Here, each replica generating finger handles the respreading process for one of the replica generation paths in replica path information. Hence, the number of operating replica generating fingers is equal to the number of replica generation paths designated by the replica path information, and signal replicas corresponding to that number of mutlipaths are added in the adder 5h, then outputted from the IGU as an interference replica signal.

Thus, according to the present embodiment, the candidate path information and replica path information are obtained on the basis of mutually different threshold values, and assignment of fingers and designation of path timings and the like are performed independently in the rake detecting fingers and replica generating fingers. That is, the path information for rake combining and the path information for replica generation are generated based on different standards, so that the rake detecting fingers and replica generating fingers are controlled independently of each other. Accordingly, it is possible to optimize the standards for path information generation for rake combining and path information generation for replica generation according to the SIR, SNIR and the like, so as to adaptively control the multipath selection for rake combining and path setting for replica generation.

If, for example, an erroneous signal replica is generated, there is a detrimental effect on the subsequent processing, so that it is preferable not to generate a signal replica by selecting a path with a weak reception power (low SIR or SNIR) as the replica path. In response thereto, in the present structural embodiment, if the SIR is low, then a replica threshold value ThREP which is higher than the rake threshold value ThRAKE is set by the above-described threshold value calculating portion 4e, and the path timings are determined by the above-described peak validating portion 4g by means of this replica threshold value ThREP. Hence, only paths in a relatively good state of reception are selected as replica generation paths, thereby enabling generation of erroneous signal replicas to be prevented. Furthermore, while paths with low reception power will result in low quality in the channel estimation results, signal replicas of paths which can have a detrimental impact due to the channel estimation results will also be eliminated. On the other hand, if the SIR is sufficiently high, then the replica threshold value ThREP is set lower than the rake threshold value ThRAKE by the threshold value calculating portion 4e, and the path timing is determined at the peak validating portion 4g by the replica threshold value ThREP. Therefore, a relatively large number of replica generation paths will be selected, generating signal replicas corresponding to more paths, and increasing the precision of the interference replica signal.

Additionally, since the signal state such as SIR and the like becomes better in progressing toward the latter stages in a multi-stage interference canceller, the replica threshold value ThREP in the present structural embodiment will be set to values which become gradually lower (in response to increasing SIR) in progressing to the latter stages. Therefore, at the initial stages where the SIR is low, a small number of interference replica signals due to the few signal replicas which are definitely not errors are removed, and in advancing to the latter stages where the SIR is higher, more interference replica signals due to a larger number of signal replicas are removed due to the increasing accuracy. As a result, an appropriate interference cancellation which makes effective use of the inherent capabilities of a multi-stage interference canceller can be achieved, thereby improving the performance of the multi-user receiver.

This type of approach is similar to interference cancellers wherein the weighting coefficient or suppression coefficient are adjusted for each path. However, in the present structural embodiment, the generation of the interference replica signals is itself controlled by the detection result outputs of the searchers, thus making it possible to generate only appropriate signal replicas in accordance with the input reception signals to the IGU's, consequently avoiding the generation of unneeded signal replicas.

(2) Second Structural Embodiment The structure of a searcher according to a second structural embodiment is shown in

Fig. 5. This searcher is a simplified version of the searcher of the above-described first structural embodiment, comprising a matched filter 4a, delay profile calculating portion 4b, SIR calculating portion 4d and threshold value calculating portion 4e which are like those of the searcher of Fig. 3, and comprising a preliminary candidate path determining portion 4c', a peak validating portion 4f ' and a peak validating portion 4g' which differ from those of Fig. 3 in the signal exchange format and the like.

The preliminary candidate path determining portion 4c', as with the above-described preliminary candidate path determining portion 4c, determines preliminary candidate paths and notifies the SIR calculating portion 4d of the determined preliminary candidate paths, but only supplies the spreading code timings and peak reception levels of the determined preliminary candidate paths to the peak validating portion 4f . The peak validating portion 4f ' uses the spreading code timings and peak reception levels of the preliminary candidate paths from the preliminary candidate path determining portion 4c' along with the rake threshold value ThRAKE from the threshold value calculating portion 4e to perform a determination like that of the above-described peak validating portion 4f, and outputs candidate path information to the IGU (rake detecting finger), but also supplies the candidate path information to the peak validating portion 4g'. At this time, the peak validating portion 4f also supplies the peak validating portion 4g' with peak reception levels exceeding the rake threshold value ThRAKE in the candidate path information.

The peak validating portion 4g' determines whether or not the peak reception levels in the candidate path information from the peak validating portion 4f ' exceeds the replica threshold value ThREP from the threshold value calculating portion 4e. Based on the results of this determination, the peak validating portion 4g' makes the spreading code timings of the peak reception levels exceeding the replica threshold value ThREP as the path timings for signal replica generation paths, and outputs to the IGU's (replica generating fingers) replica path information including these path timings and the number of signal replica generation paths (number of peak reception levels exceeding the replica threshold value ThREP). That is, the peak validating portion 4g' selects peaks indicating replica generation paths from among those selected by the peak validating portion 4f as peaks indicating rake combining paths. As a result, replica generation paths are further selected out from paths which have once been selected as multipaths (albeit for rake combining), thereby more reliably preventing selection of inappropriate replica generation paths, and is particularly effective in cases of a poor state of reception (when the SIR, SNIR or SN ratio are low). Additionally, whereas the above-described peak validating portion 4g, as in the peak validating portion 4f, takes the peak reception levels of the preliminary candidate paths as the object of determination, the peak validating portion 4g' takes only those peak reception levels received from the peak validating portion 4f ' as the object of determination, thus enabling the amount of computation for obtaining replica path information to be reduced and enabling simplification and reduction of circuitry and the like needed for such computation. As in the case of the above-described peak validating portions 4f and 4g, the candidate path information and replica path information outputted from the peak validating portions 4f and 4g' may also include other information such as the reception power order of the rake combining paths and the reception power order of the signal replica generation paths as needed.

Due to this type of structure, the searcher of Fig. 5 supplies the same type of candidate path information and replica path information as the searcher of the above-described first structural embodiment to the IGU's. Therefore, when the searcher of

Fig. 5 is used as the searchers 4SlUl-4S3Uk in the multi-user receiver of Fig. 1, it is possible to use the above-described IGU of Fig. 4 for the IGU's 5SlUl-5S3Uk. As for the operations in this case, the determinations at the peak validating portion 4g' are performed based on the candidate path information from the peak validating portion 4f ' as described above in each searcher of each stage. Then, the candidate path information and the replica path information from the searchers is supplied to the IGU's, and as in the case of the above-described first structural embodiment, the rake detecting fingers and replica generating fingers are controlled independently of each other. As a result, it is possible to adaptively control both the multipath selection for rake combining and path setting for replica generation as described above, thereby enabling an appropriate interference cancellation to be performed.

In the present structural embodiment, since the peak validating portion 4g' selects the replica generation paths from the rake combining paths selected from the peak validating portion 4f, the number of replica generation paths will be less than or equal to the number of rake combining paths. Therefore, in the case where the above-described IGU of Fig. 4 is used for the IGU's 5SlUl-5S3Uk, channel estimation results from the channel estimator 5a operating based on candidate path information of corresponding rake combining paths are supplied to all of the channel decompensators 5g to which replica path information is supplied, so that channel decompensation is performed respectively in these channel decompensators 5g. Consequently, there is no need for operation of channel estimators to which candidate path information is not supplied as is the case in the above-described first structural embodiment.

(3) Third Structural Embodiment

The structure of an IGU according to a third structural embodiment is shown in Fig. 6. In the present structural embodiment, some of the functions which were performed by the searchers in the above-described second structural embodiment are provided on the IGU side. As shown in the drawing, the IGU of the present structural embodiment comprises n rake detecting fingers 5FDl-5FDn each having a channel estimator 5a, a despreader 5b and a channel compensator 5c', a rake combiner 5d, a preliminary decider 5e, m replica generating fingers 5FGl-5FGm each having a respreader 5f and a channel decompensator 5g', an adder 5h and a replica path control portion 5i.

Of these constituent elements, the channel estimator 5a, the despreader 5b, the rake combiner 5d, the preliminary decider 5e, the respreader 5f and the adder 5h have the same functions as those in the IGU of Fig. 4. The channel compensator 5c', in addition to having the same functions as the channel compensator 5c in the IGU of Fig. 4, also outputs a channel compensated signal to the replica path control portion 5i.

The replica path control portion 5i receives candidate path information supplied to the rake detecting fingers and channel compensated signals outputted from the rake detecting fingers (channel compensators 5c'), and supplies information based thereon corresponding to the above-described replica path information to the replica generating fingers. Since the channel compensated signals from the rake detecting fingers correspond to detection signals obtained by detecting the reception signal components of the respective rake combining paths in the candidate path information, the replica path control portion 5i determines whether or not to employ the rake combining paths based on the powers of these detection signals as replica generation paths. That is, with regard to the rake combining paths with low detected signal power, these are not employed as replica generation paths because they cannot be expected to generate effective signal replicas. In contrast, with regard to the rake combining paths with a high detected signal power, these are employed as replica generation paths since they can be expected to generate effective signal replicas. For example, when the signal power after channel compensation exceeds the power corresponding to the above-described replica threshold value ThREP, these rake combining paths are employed as replica generation paths. The replica path control portion 5i in this way selects a replica generation path from a rake combining path in the candidate path information. Then, it extracts candidate path information such as path timings from the selected replica generation paths, and supplies these to the replica generating fingers as replica path information.

The channel decompensator 5g' performs channel decompensation similar to the above-described channel decompensator 5g with respect to the signals respread by the respreader 5f. Due to this channel decompensation, the respread signals for each replica generation path are separated into signal components in accordance with the reception powers of the respective replica generation paths, and are returned to their levels in the original input reception signal. Then, the channel decompensator 5g' adjusts the output timings of the channel decompensated signals based on the replica path information supplied from the replica path control portion 5i, so that the signal replicas from the respective replica generating fingers will be outputted to the adder 5h at the same timings as in the original input reception signal.

In the present structural embodiment, the replica path control portion 5i selects replica generation paths from among the rake combining paths in the candidate path information, so that the number of replica generation paths is equal to or less than the number of rake combining paths. Therefore, all of the channel decompensators 5g' to which replica path information is supplied from the replica path control portion 5i are supplied with channel estimation results from the channel estimators 5a operating based on candidate path information of corresponding rake combining paths, so that the channel decompensation at these channel decompensators 5g' is performed respectively.

The IGU according to the above-described structure requires only the rake combimng path information as path information from the searchers when generating interference replica signals from the input reception signal. Accordingly, the IGU of Fig. 6 can have the same interface as the conventional IGU of Fig. 11 or the like, and can be used in conjunction with conventional searchers as shown in Fig. 2 in the multi-user receiver of Fig. 1. Additionally, it can be used in the multi-user receiver of Fig. 1 in conjunction with a searcher structured so as to supply only candidate path information such as by removing the peak validating portion 4g from the searcher of Fig. 3 or removing the peak validating portion 4g' from the searcher of Fig. 5.

When such a searcher is used with the IGU of Fig. 6 in the multi-user receiver of Fig.

1 (as the IGU's 5S1U1-IGU 5S3Uk), replica generating information is generated as described above by the replica path control portion 5i from the candidate path information and channel compensated signals of each path in each IGU of each stage. That is, replica generation paths are selected out from among the paths which have once been selected as rake combining paths as in the above-described second structural embodiment, and finger assignments and designations of path timings with respect to the replica generating fingers are performed by means of path information which is different from the candidate path information. As a result, even in the case where searchers supplying only candidate path information are used, it is possible to prevent selection of inappropriate replica generation paths such as the case of poor reception conditions, thus enabling the generation of erroneous signal replicas to be avoided and obtaining accurate interference replica signals.

In the IGU of Fig. 6, the replica path control portion 5i determines replica generation paths with the channel compensated signals as inputs, but there is no restriction to such, and the determination can be performed using as the input any other signal enabling evaluation of the reception conditions of the respective paths. For example, despread signals which have not undergone channel compensation can be supplied from the despreader 5b to the replica path control portion 5i, and the replica generation paths determined on the basis of power or the like of the signals from the despreader 5b (this is more effective for cases in which the error due to channel compensation can be expected to be large).

According to the above-described structural embodiments of the searcher and IGU, it is possible to separately generate candidate path information for rake combimng and replica path information for replica generation as explained above, and to adaptively control the multipath selection for rake combining and path setting for replica generation independent of each other. Additionally, since such adaptive control can be performed for each user at each stage, it is possible to generate interference replica signals which are more accurate than in the case of the basic structural embodiment described above, thereby to better improve the performance of the multi-user receiver.

C. Changing of Search Windows (1) Structure

The path detection resolution in a search window is determined by the oversampling rate for detecting the levels of delay profiles. When holding the circuit size and processing times of the searchers constant with respect to search windows, there is generally a trade-off relationship between the oversampling rate in the search window and the width of the search window. For example, when performing high-speed oversampling at four-times speed (making a fourfold increase in the path detection resolution), then the level detection interval of delay profiles will become 1/4, so that the detectable search window width will be quartered. For this reason, in order to perform high-precision path detection without changing the width of the search window, it is necessary to expand the circuit size of the searchers or the make the processing time longer.

In contrast, the multi-stage interference canceller of the present embodiment has searchers in each stage, and the structure is such that accurate path detection can be expected in the searchers of the latter stages with a high SIR. According to this structure, by appropriately changing the oversampling rate, width and position of the search windows in the searchers of the respective stages, it is possible to perform path detection at high precision while keeping the circuit size and processing time of the searchers constant.

Fig. 7 is a drawing showing a structural example of a multi-user receiver for performing such changes of the search windows. This multi-user receiver employs the searchers 4SlUl'-4S3Uk' shown in the drawing to which have been added further functions in the multi-user receiver according to the structure described above in "A. Basic Structural

Embodiment" and "B. Structural Embodiments of Searchers and IGU's". That is, the searchers 4SlUl'-4S3Uk have the same functions as the above-described searchers 4SlUl-4S3Uk (searchers of Fig. 2, Fig. 3 and Fig. 5), and in addition thereto have a function of changing the search window. The change of the search window is performed by controlling the timing of the sampling whereby the input reception signals are taken in the searchers 4SlUl'-4S3Uk'.

In a normal multi-stage interference canceller, a the reception signal being processed is supplied at the same oversampling rate as at the time of reception in which the radio signal was first oversampled with respect to the IGU's of the respective stages. The structural example of Fig. 7 also complies with this structural example, and input reception signals are supplied to the searchers 4SlUl'-4S3Uk' and the IGU's 5SlUl-5S3Uk at the same oversampling rate as at reception. Therefore, the searchers 4SlUl'-4S3Uk' respectively change the oversampling rate in the search window by appropriately changing the sampling rate for taking in the input reception signals. Additionally, the position and width of the search window are changed by appropriately setting the starting point, ending point and resuming point of the sampling for taking in the input reception signal. Since the oversampling rate inside the search windows cannot be made higher than the rate corresponding to the oversampling rate at the time of reception (oversampling rate at the radio signal processing portion 2), they are appropriately selected within the range up to that rate.

Additionally, the searchers 4S1U1', 4S1U2', . . ., 4SlUk' of the first stage supply the respective path timings of the rake combining paths in the respectively acquired path information to the searchers 4S2U1 ', 4S2U2', . . ., 4S2Uk' of the second stage. The searchers 4S2U1', 4S2U2', . . ., 4S2Uk' of the second stage supply the respective path timings of the rake combining paths in the respectively acquired path information to the searchers 4S3U1', 4S3U2', . . ., 4S3Uk' of the third stage. Additionally, the searchers 4S2U1', 4S2U2', . . ., 4S2Uk' of the second stage and the searchers 4S3U1', 4S3U2', . . ., 4S3Uk' of the third stage respectively set predetermined intervals around the respective path timings (for example, intervals corresponding to a predetermined number of chips around the path timings) which have been supplied on the delay time axis as search windows, and control the oversampling rate so as to perform path detection at detection resolutions that are appropriate to those search windows.

(2) Operations

In this type of structure, for example, when a signal reception operation from a first user is begun, the searcher 4S1U1' of the first stage sets the oversampling rate inside the search window to a low oversampling rate (for example, a rate equal to the chip rate), and performs delay profile calculations and the like by sampling the input reception signal by this oversampling rate. As a result, a delay profile with a low detection resolution over a wide search window is obtained, and path timings of rake combining paths detected from this delay profile are supplied to the searcher 4S2U1' of the second stage.

The searcher 4S2U1' of the second stage which has received these path timings sets predetermined intervals around the respective path timings which have been obtained as search windows. Then, when input reception signals corresponding to intervals inside these search windows are supplied, delay profile calculations and the like are performed by sampling the input reception signals at an oversampling rate of at least the oversampling rate of the searcher 4S1U1' of the first stage. As a result, at the searcher 4S2U1', delay profiles of a detection resolution of at least that of the first stage are obtained with respect to only the area around the path timings detected in the searcher 4S1U1' of the first stage, and precise rake combining paths and replica generation paths which are more accurate than those of the first stage are detected on the basis of these delay profiles.

For example, in the searcher 4S1U1' of the first stage, delay profiles as shown in Fig. 8 (the delay profiles on the delay time axis at the upper portion of the drawing) are obtained at a low detection resolution over a wide search window Wl, and the peak timings of PI, P2 and P3 are detected as the path timings of the rake combining paths. Then, the searcher 4S2U1' of the second stage is supplied with information giving notice of the path timings PI, P2 and P3, and at the searcher 4S2U1 ', path detection is performed at a high detection resolution only in the vicinity of the path timings PI , P2 and P3.

Now, assuming that the searcher 4S2U1 ' of the second stage performs path detection at an oversampling rate which is 4 times that of the searcher 4S1U1' of the first stage, if the searcher 4S2U1' samples the input reception signal directly without using the path timing information from the searcher 4S1U1', then the sampling window of the searcher 4S2U1' would be as indicated by the search window W2 in Fig. 8, so that path detection will not be able to be performed over the same interval as the search window Wl unless four times the processing is performed (as indicated by the dashed line window in the drawing). On the other hand, if the searcher 4S2U1' makes use of the path timing information from the searcher 4S1U1' as described above, then the search windows W3 will be set around the path timings PI, P2 an dP3 as indicated by the single-dotted chain lines in the drawings. As a result, it is possible to divide a window corresponding to the width of the search window W2 as indicated by the double-dotted chain line in the drawing, so as to enable high-precision detection to be performed at four times the oversampling rate with respect to only the portions containing the peak timings.

The path timings of rake combining paths detected by the searcher 4S2U1' in this way are supplied to the searcher 4S3U1 ' of the third stage. The searcher 4S3U1 ' of the third stage which has received these path timings sets predetermined intervals around the received path timings as search windows. Then, when input reception signals corresponding to the intervals inside these search windows are supplied, the input reception signals are sampled at an oversampling rate of at least the oversampling rate of the searcher 4S2U1' of the second stage to perform delay profile calculation or the like. As a result, in the searcher 4S3U1', delay profiles are obtained at a detection resolution of at least that of the second stage for only the areas around the path timings detected by the searcher 4S2U1' of the second stage, and rake combining paths of a precision which is higher than those of the first stage and second stage can be detected based on these delay profiles.

The path detection at the searchers 4S1U1'-4S3U1' corresponding to the first user is performed as described above, and the generation of interference replica signals and the like is performed by the IGU's 5S1U1-5S3U1 which similarly correspond to the first user based on these path detection results. Additionally, these operations are performed independent of each other in response to signals arriving from each user in the constituent elements corresponding to each user. That is, when a signal reception operation for a second user is begun, the path detection at the searchers 4S1U2'-4S3U2' is performed in the same operational format as described above (separately from searchers corresponding to other users), and the generation of interference replica signals and the like is performed by the IGU's 5S1U2-5S3U2 and the like corresponding to the second user based on these path detection results. Similarly, with regard to the signal reception operations for the third, fourth, . . ., k-h users, the path detection at the searchers 4S1U3'-4S3U3', searchers 4S1U4'-4S3U4', . . ., searchers 4SlUk'-4S3Uk' is performed (separately from searchers corresponding to other users) by an operational format which is the same as described above, and the generation of interference replica signals and the like by the IGU's 5S1U3-5S3U3, IGU's 5S1U4-5S3U4, . . ., IGU's 5SlUk-5S3Uk and the like corresponding to the third, fourth, . . ., k- users are performed based on the respective path detection results. As a result, with each reception of a signal from the respective users, wide and low-resolution search windows with a low oversampling rate are used in the initial stages, whereas in the latter stages, small search windows with a high oversampling rate are used as appropriate based on information from the previous stage. That is, in the beginning stages which have a low SIR, the approximate path timings are captured over a wide range at a low resolution, whereas in the latter stages having a high SIR, the path timings are captured at a high resolution in the vicinity of the approximate path timings. Accordingly, due to this format in which the search windows are changed, it is possible to detect pat timings with better precision at the latter stages without substantially shortening the intervals over which path detection is performed, thus enabling an appropriate interference cancellation to be achieved in the multi-stage interference canceller.

Additionally, since path detection is performed by restricting the range to only the areas around the path timings detected in the preceding stages, it is possible to hold the quantity of data handled and the quantity of computational processing to a constant value or less while increasing the oversampling rate of the searchers over those of the previous stage. Therefore, even if searchers of a constant circuit size are used and the processing times at the searchers are kept constant, it is still possible to perform high-precision path detection at a high resolution in the latter stages.

In the above-described format, the search window is changed based as the path timings of the rake combining paths in the path information, but it is also possible to appropriately change the search windows based on other information (other information relating to the multipaths obtained in the preceding stages).

<Second Embodiment A. Structure

Next, a second embodiment of the present invention shall be described. Fig. 9 is a drawing showing a structural example of a multi-user receiver using a multi-stage interference canceller according to the second embodiment of the present invention. The present embodiment is one wherein effective path searcher functions are achieved with a simple structure in a multi-stage interference canceller, and Fig. 9 shows a structure of a multi-user receiver for the case where a 3 -stage parallel interference canceller arrangement is used as an example of a structure to which this format is applied.

The present multi-user receiver, like the above-described first embodiment, is a receiving device for handling a plurality of users used in the radio base stations and the like of cellular radio communication systems under the CDMA system, and receives from k users labeled from first to k-th. As constituent elements for receiving signals from these k users and canceling the interference in the received signals to obtain reception information, the present multi-user receiver, as shown in the drawing, comprises an antenna 1, a radio signal processing portion 2, buffers 3S1 and 3S2, k searchers 4Ul-4Uk, k * 3 IGU's 5SlUl'-5SlUk', 5S2Ul'-5S2Uk' and 5S3Ul'-5S3Uk', subtracters 6S1 and 6S2, k x 2 adders 7S2Ul-7S2Uk and 7S3Ul-7S3Uk, and k input control switches 8Ul-8Uk. Of these constituent elements, the antenna 1, radio signal processing portion 2, buffers 3S1 and 3S2, subtracters 6S1 and 6S2, and the adder 7S2Ul-7S2Uk and 7S3Ul-7S3Uk have the same functions as the constituent elements labeled with the same reference numbers in the above-described first embodiment.

The searchers 4U1, 4U2, . . ., 4Uk respectively take signals supplied through the input control switches 8U1, 8U2, . . ., 8Uk to be described below as input reception signals, and detect the path timings of multipaths according to the reception signals from the first, second, . . ., k-th users. As illustrated in the drawing, only a single searcher 4U1, 4U2, . . ., 4Uk is provided for each user in the first stage in the present embodiment, these supplying path information to IGU's of the respective stages corresponding to the same user. As for the structure of the searchers 4Ul-4Uk, it is possible for example to use structures according to any of Fig. 2, Fig. 3 and Fig. 5 described above, but the structure must be such as to supply the obtained path information to the IGU's of all stages corresponding to the same user.

The IGU's 5SlUl'-5S3Uk' are IGU's (interference replica generating units) for respectively generating interference replica signals, and the IGU's 5S1U1'-5S3U1', IGU's 5S1U2'-5S3U2', . . ., IGU's 5SlUk'-5S3Uk' respectively use path information from the searchers 4U1, 4U2, . . ., 4Uk. As for the structures of the IGU's 5SlUl'-5S3Uk' themselves, it is possible to use structures according to Fig. 4, Fig. 6 or Fig. 11 or the like as described above, but an arrangement which complies with the structure of the searchers 4Ul-4Uk must be selected. For example, if the structure of Fig. 2 is employed in the searchers 4Ul-4Uk, then IGU's 5SlUl'-5S3Uk' with structures according to Fig. 6 or Fig. 11 should be employed, whereas if the structures of Fig. 3 or Fig. 5 are used for the searchers 4Ul-4Uk, then the IGU's 5S1U1 '-5S3Uk' with the structure of Fig. 4 should be employed.

The input control switches 8U1, 8U2, . . ., 8Uk are respectively switching means for selectively switching the input reception signals supplied respectively to the searchers 4U1, 4U2, . . ., 4Uk, and are provided in front of the searchers 4U1, 4U2, . . ., 4Uk. These input control switches 8Ul-8Uk are arranged so as to be supplied with the reception signals from the radio signal processing portion 2, and the input reception signals to the IGU's in the latter stages (second and subsequent stages) corresponding respectively to the same users, so that the input control switches 8Ul-8Uk can select between these reception signals and the input reception signal as the input reception signals to send out to the searchers 4Ul-4Uk. In the illustrated structural example, the input control switches 8Ul-8Uk are arranged so as to select either the reception signals from the radio signal processing portion 2 or the input reception signals to the IGU's of the third stage as the input reception signals to the searchers 4Ul-4Uk. The signal selection format for these input control switches 8Ul-8Uk is such as to select the input reception signals which are most suitable for path detection of multipaths at each point in time; the specifics shall be described in the following explanation of the operations.

B. Operations

In a structure as described above, when the antenna receives a radio signal transmitted from the mobile stations of the respective users, the received signal is supplied through the radio signal processing portion 2 to the buffer 3S1 of the first stage the IGU's 5SlUl'-5SlUk' and the input control switches 8Ul-8Uk. At the time of activation of the overall communication system including the present multi-user receiver or the like, when the present multi-user receiver begins operation, the input control switches 8Ul-8Uk send out the reception signals from the radio signal processing portion 2 as they are directly to the searchers 4Ul-4Uk as input reception signals.

At the searchers 4Ul-4Uk which have received input reception signals from the input control switches 8Ul-8Uk, despreading, delay profile calculation, threshold value calculation, peak validation and the like are performed in the same way as the searchers described in the first embodiment above, path information which is in accordance with the searcher structure(only rake combining path information, or both candidate path information and rake combining path information) is obtained. The path information obtained by the searchers 4U1, 4U2, . . ., 4Uk, is supplied to the IGU's 5S1U1'-5S3U1', IGU's 5S1U2'-5S3U2', . . ., IGU's 5SlUk'-5S3Uk' corresponding to respectively the same users.

At the IGU's 5SlUl '-5SlUk' which have received the path information from the searchers 4Ul-4Uk, interference replica signals are generated from the reception signals in accordance with the path information, and outputted to the subtractor 6S1 and adders

7S2U1-7S2UL At the subtractor 6S1, the interference replica signals are subtracted from the reception signals, the residual signal left after subtraction is outputted to the adders 7S2Ul-7S2Uk.

At the adders 7S2U1, 7S2U2, . . ., 7S2Uk which have received the residual signal from the subtractor 6S1, the residual signal is added to the interference replica signals from the IGU's 5S1U1', 5S1U2', . . ., 5SlUk'. As a result, input reception signals from with the interference replica signals (multiple access interference) of other users generated in the first stage have been removed are outputted to the IGU's 5S2Ul'-5S2Uk\ At the IGU's 5S2Ul'-5S2Uk', these input reception signals are used to generate interference replica signals in accordance with path information supplied from the searchers 4Ul-4Uk, and outputted to the subtractor 6S2 and adders 7S3Ul-7S3uk. At the subtractor 6S2, these interference replica signals are subtracted from the original reception signal, and the post-subtraction residual signal is outputted to the adder 7S3Ul-7S3Uk.

At the adders 7S3U1, 7S3U2, . . ., 7S3Uk which have received the residual signal from the subtractor 6S2, the residual signal is added to the interference replica signals respectively from the IGU's 5S2U1', 5S2U2', . . ., 5S2Uk'. As a result, input reception signals with the interference replica signals of other users generated in the second stage removed are outputted form the adders 7S3Ul-7S3Uk to the IGU's 5S3Ul'-5S3Uk'. At the

IGU's 5S3Ul'-5S3Uk', these input reception signals are used to perform a despreading process, rake combining and preliminary decision in accordance with path information supplied from the searchers 4Ul-4Uk, and the decision results are outputted as reception information data from the first to k-th users.

Subsequently, interference replica signal generation and the like based on the same path information for each user is performed in each stage, so as to repeat the reception information data acquisition from reception signals with the multiple access interference removed. Additionally, when the stages operate in steady state and the reception information data from the third stage reach a steady state obtained in a predetermined state, the input control switches 8Ul-8Uk switch the input reception signals sent to the searchers 4Ul-4Uk to the input reception signals to the IGU's 5S3U1 '-5S3Uk' side.

Here, whether or not a steady state has been reached is determined, for example, by monitoring whether or not the state of the reception information data outputted from the third stage and the input reception signals inputted to the IGU's 5S3Ul'-5S3Uk' have reached a predetermined desirable state. In the case where the state of the reception information data is monitored, it is possible, for example, to provide measuring devices for measuring the BER (bit error rate) of the reception information data outputted from the IGU's 5S3U1',

5S3U2', . . ., 5S3Uk' of the third stage and to continually watch the BER of reception information data for each user by these measuring devices. Then, when the measuring devices have begun obtaining the predetermined desirable BER as the measurement result (upon achieving a stable state), the input control switches are advised thereof (of the input control switches 8Ul-8Uk, the input control switches corresponding to the same user as the user whose BER is monitored by each measuring device), and the input control switches which have received such reports switch the input reception signals sent to the searchers to the input reception signals from the third stage.

Due to the switching operation at the input control switches 8Ul-8Uk, the same input reception signals as used in the IGU's 5S3Ul '-5S3Uk' of the third stage are supplied to the searchers 4Ul-4Uk. These input reception signals are signals which have undergone multiple access interference cancellation twice by means of the processing at the first stage and second stage and the adders 7S3Ul'-7S3Uk'. Therefore, at the searchers 4Ul-4Uk, path detection of the multipaths can be performed with the SIR of the input reception signals in a higher state (a cleaner signal state) than when the reception signals were received from the radio signal processing portion 2, thus enabling accurate path information to be obtained.

This accurate path information is supplied from the searchers 4U1, 4U2, . . ., 4Uk to the IGU's 5S1U1 '-5S3U1', IGU's 5S1U2'-5S3U2', . . ., IGU's 5SlUk'-5S3Uk'. As a result, at the IGU's of each stage, interference replica signals are generated based on path information which is more accurate than the path information supplied before. Accordingly, multiple access interference is appropriately removed from the reception signals by more accurate interference replica signals, thereby resulting in more accurate reception information data.

In this way, according to the present embodiment, path information is first obtained from the reception signal at the time of reception, but thereafter, the condition shifts to one of acquiring path information from reception signals with the multiple access interference removed, thereby enabling interference cancellation by more accurate path information. Then, since only a single searcher is provided for each user and the acquisition of path information is achieved by means of switching the input reception signals supplied to the searchers, so that the number of searchers can be reduced from that of the first embodiment above. Accordingly, it is possible to appropriately remove interference signals with a more simple structure than the above-described first embodiment, enabling the performance of multi-user receivers to be readily improved.

C. Modification Example (1) Selection of Input Reception Signals

In the above-described multi-user receiver, the input reception signals to the IGU's of the third stage are supplied to the searchers upon reaching the steady state, but it is possible to supply the input reception signals to the IGU's of the second stage to the searchers. Additionally, it is possible to supply the reception signal and the input reception signals to the IGU's of all stages to the input control switches 8Ul-8Uk, and to have the input control switches 8Ul-8Uk appropriately select the signals to be sent to the searchers 4Ul-4Uk in accordance with the state of reception or the like. For example, it is possible to first transfer from a state of sending out reception signals from the radio signal processing portion 2 to a state of sending out the input reception signals to the IGU's of the second stage, then transferring from there to a state of sending out the input reception signals to the IGU's of the third stage, thus supplying the input reception signals to the IGU's of the latter stages in steps.

(2) Structural Embodiment of Searcher

As described above, in the present embodiment, it is possible to use searchers according to any one of Fig. 2, Fig. 3 and Fig. 5 for the aboveOdescribed first embodiment, but it is possible to separately generate the path information supplied to the IGU's of the respective stages in the searchers 4Ul-4Uk, so that rake combining and replica generation of different multipaths is performed in each stage. For example, when using the structure of Fig. 3 or Fig. 5 as the searchers 4Ul-4Uk, structures corresponding to the threshold value calculating portion 4e, peak validating portion 4f or 4f and peak validating portion 4g or 4g' are provided in each stage, and different rake threshold values ThRAKE and replica threshold values ThREP are set for each stage to perform peak validation, and the candidate path inforamiton and replica path information of each stage obtained in this way can be supplied to the IGU's of each stage. As a result, in the present embodiment as well, it is possible to perform independent adaptive control of multipath selection for rake combining and path selection for replica generation.

(3 ) Structural Embodiment of IGU In the present embodiment, when IGU's 5SlUl'-5S3Uk' according to the structure of Fig. 6 are used, replica generating paths are selected for each stage in accordance with the function of the replica path controlling portion 5i. Therefore, replica generation is controlled independently of rake combimng regardless of the path information from the searchers 4Ul-4Uk (regardless of the structure of the searchers).

Various diverse modifications are possible on the above-described first and second embodiments, the above-described forms for the structure and operations being no more than a single example, and applications and uses being possible according to various forms. While it goes without saying that there can be any number of users capable of being accommodated and number of stages in the multi-stage interference canceller, the following examples can be given as other possible modifications and applications.

Modifications or Applications of the Searchers In the above embodiments, the determining threshold values of the reception level peaks in the path detection by the searchers is determined on the basis of the SIR, but the path detection can just as well be performed by other standards. For example, by selecting a number of spreading code timings in the order of size of the peak level, a number of the paths with higher reception power can be used as rake combining paths or replica generating paths.

Additionally, while the searchers usually repeat the path detection at predetermined timings such as a standard period or the like, it is possible to suitably control the period of path detection in accordance with the stability of detected multipaths or the like, or to change the suitably change the set positions of the search windows at handover or the like. Furthermore, while it is possible to delay the reception signal and to use reception information data signals which have undergone decision as pilot signals, in the case of the above-described first embodiment having searchers in each stage, it is possible to use reception information data signals as pilot signals in the latter stages by supplying the decision results obtained in the former stages to the searchers of the latter stages. While it is possible to supply path information by changing the number of rake detecting fingers depending on whether a high-speed channel with a high transmission rate or a low-speed channel with a low transmission rate, in this case, it is possible to suitably change not only the number of rake detecting fingers but also the number of replica generating fingers according to whether a high-speed channel or a low-speed channel.

Application to Communication Systems

Since the multi-stage interference canceller and multi-user receiver according to the above-described embodiment are capable of performing accurate interference cancellation as mentioned above, it contributes to increased capacity of the commumcation system.

Additionally, since the structure does not presuppose a specific standard communication system, it can be used generally for all commumcation systems according to the CDMA system. Therefore, it can be applied not only to communication systems such as W-CDMA and cdma2000, but also to CDMA communication systems according to other standards.

According to the interference canceller of the above-described embodiment, it is possible to perform appropriate interference cancellation by generating accurate interference replica signals as described above, thereby enabling a more stable and effective interference canceller receiver (receiving device in base stations or the like) to be achieved. Consequently, the system capacity and cell coverage can be increased, making it possible to reduce the transmission power of the mobile stations.

While preferred embodiments of the method and device of the present invention have been illustrated in the accompanying drawings and described in the foregoing detailed description, it shall be understood that the invention is not limited to the embodiment disclosed, but is capable of encompassing numerous rearrangements, modification, and substitutions without departing from the spirit of the invention as set forth and defined in the following claims.

Claims

1. An interference canceling method for receiving a reception signal containing signals from a plurality of transmitting stations, generating replica signals corresponding to signals from the respective transmitting stations, and using the generated replica signals to perform an interference canceling procedure on a signal from a transmitting station other than said transmitting stations, the interference canceling method comprising: obtaining path information on the transmission paths through which the signals from said transmitting stations have arrived from signals which have undergone said canceling procedure at least once, and generating said replica signals based on the obtained path information.

2. An interference canceling method as recited in claim 1, characterized by at the starting time of said canceling procedure, obtaining path information from said reception signal, and respectively generating said replica signals are based on this obtained path information, and subsequent to a predetermined time after said starting time, obtaining path information from signals which have undergone said canceling procedure at least once, and respectively generating said replica signals based on this obtained path information.

3. An interference canceling method for receiving a reception signal containing signals from a plurality of transmitting stations, and in a plurality of stages, respectively generating replica signals corresponding to signals from said transmitting stations based on path information on the transmission paths through which the signals from said transmitting stations have arrived, and sequentially performing interference canceling procedures on a signal from a transmitting station other than said transmitting stations using the generated replica signals, the interference canceling method comprising: in a first stage, obtaining path information from said reception signal, and respectively generating said replica signals based on this obtained path information; and in a second stage, obtaining path information from signals which have undergone said canceling procedure in the previous stage, and respectively generating said replica signals based on this obtained path information.

4. An interference canceling device for receiving a reception signal containing signals from a plurality of transmitting stations, and in a plurality of stages, respectively generating replica signals corresponding to the signals from the transmitting stations, and sequentially performing interference canceling procedures on a signal from a transmitting station other than said transmitting stations using the generated replica signals, the interference canceling device comprising: detecting means provided in each of said plurality of stages, for lespectively detecting path information on the transmission paths through which the signals from said transmitting stations have arrived based on said reception signal or signals which have undergone said canceling procedure in a previous stage; and generating means provided in each of said plurality of stages, for respectively generating said replica signals based on the path information detected by said detecting means provided in the same stage.

5. An interference canceling device for receiving a reception signal containing signals from a plurality of transmitting stations, and in a plurality of stages, respectively generating replica signals corresponding to the signals from the transmitting stations, and sequentially performing interference canceling procedures on a signal from a transmitting station other than said transmitting stations using the generated replica signals, the interference canceling device comprising: detecting means for detecting path information on the transmission paths through which the signals from said transmitting station have arrived based on supplied signals; generating means provided in each of said plurality of stages, for respectively generating said replica signals based on path information detected by said detecting means; and selecting means for selecting and supplying to said detecting means said reception signal or signals which have undergone said canceling procedure in any one of the stages.

6. An interference canceling device as recited in claim 4 or 5, characterized in that: said generating means are means for demodulating said reception signal or signals which have undergone said canceling procedure in a previous stage by means of a rake combining procedure, and generating said replica signals from the demodulated signals; and said detecting means separately detect first path information for the purpose of said rake combining procedure and second path information for the purpose of generating said replica signals, and supply these to said generating means.

7. An interference canceling device as recited in claim 6, characterized in that said detecting means set first and second threshold values, and respectively detect said first and second path information based on said first and second threshold values.

8. An interference canceling device as recited in claim 6, characterized in that: said detecting means set first and second threshold values, detect said first path information based on said first threshold value, and detect said second path information based on said second threshold value from among said first path information which has been detected.

9. An interference canceling device as recited in claim 4 or 5, characterized in that: said generating means are means for demodulating said reception signal or signals which have undergone said canceling procedure in a previous stage and generating said replica signals from the demodulated signals; said detecting means detect path information for the purpose of said rake combining procedure and supply them to said generating means; and said generating means perform said rake combining procedure based on path information supplied from said detecting means, select paths for generating replica signals from among the rake combimng paths in said rake combimng procedure, and generate said replica signals based on these selected paths.

10. An interference canceling device as recited in claim 4 or 5, characterized in that: said detecting means detect path information based on correlation values between said reception signal or signals which have undergone said canceling procedure, and spreading codes used for modulation of signals in said transmitting stations.

11. An interference canceling device as recited in any one of claims 4-10, characterized in that: said detecting means supply information indicating paths in the detected path information to said detecting means provided in latter stages, and determine the range of signals for which path information is detected based on said information supplied from said detecting means provided in previous stages.

12. A receiving device for obtaining information transmitted by said transmitting stations from said reception signal with the interference cancelled by means of an interference canceling device as recited in any one of claims 4-11.