CA2076061A1 - Method of forming a channel estimate for a time-varying radio channel - Google Patents
Method of forming a channel estimate for a time-varying radio channelInfo
- Publication number
- CA2076061A1 CA2076061A1 CA002076061A CA2076061A CA2076061A1 CA 2076061 A1 CA2076061 A1 CA 2076061A1 CA 002076061 A CA002076061 A CA 002076061A CA 2076061 A CA2076061 A CA 2076061A CA 2076061 A1 CA2076061 A1 CA 2076061A1
- Authority
- CA
- Canada
- Prior art keywords
- channel
- estimate
- time
- derivative
- forming
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Abandoned
Links
- 238000000034 method Methods 0.000 title claims description 37
- 238000005562 fading Methods 0.000 claims abstract description 38
- 230000005540 biological transmission Effects 0.000 claims description 12
- 108010076504 Protein Sorting Signals Proteins 0.000 claims description 10
- 238000005070 sampling Methods 0.000 claims description 10
- 230000006870 function Effects 0.000 claims description 5
- 230000003111 delayed effect Effects 0.000 claims description 4
- 230000006978 adaptation Effects 0.000 description 13
- 230000004044 response Effects 0.000 description 6
- 238000010586 diagram Methods 0.000 description 4
- 230000003044 adaptive effect Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 3
- 238000004891 communication Methods 0.000 description 3
- 230000006854 communication Effects 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 230000017105 transposition Effects 0.000 description 3
- 239000013598 vector Substances 0.000 description 3
- 238000007476 Maximum Likelihood Methods 0.000 description 2
- 230000021615 conjugation Effects 0.000 description 2
- 239000006185 dispersion Substances 0.000 description 2
- 241001463014 Chazara briseis Species 0.000 description 1
- 101001052394 Homo sapiens [F-actin]-monooxygenase MICAL1 Proteins 0.000 description 1
- 102100024306 [F-actin]-monooxygenase MICAL1 Human genes 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 230000014509 gene expression Effects 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 230000011664 signaling Effects 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L25/00—Baseband systems
- H04L25/02—Details ; arrangements for supplying electrical power along data transmission lines
- H04L25/0202—Channel estimation
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B7/00—Radio transmission systems, i.e. using radiation field
- H04B7/005—Control of transmission; Equalising
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L7/00—Arrangements for synchronising receiver with transmitter
- H04L7/04—Speed or phase control by synchronisation signals
- H04L7/041—Speed or phase control by synchronisation signals using special codes as synchronising signal
Abstract
ABSTRACT
Radio signals are transmitted over a fading radio channel and are filtered and sampled in a receiver to obtain a digital signal (y(n)). A channel equalizer (5) has an equalizing filter (CEST) in which a channel estimate (h(n/n-L)) is formed with the aid of a known synchronizing sequence. A Viterbi analyzer (VIT) decides symbols (d(n-L)) with the aid of the channel estimate. An es-timated time derivative of the channel estimate (h(n/n-L)) is formed and adapted with the aid of the decided symbols (d(n-L)).
The channel estimate (h(n/n-L)) is adapted to the radio channel by the derivative estimate and the decided symbols (d(n-L)). In the event of fading with rapidly varying signal strength (y(n)), the channel estimate (h(n/n-L)) is adapted slowly and is predicted mainly through the derivative estimate which varies relatively evenly. The digital signal (y(n)) is stored in a memory (M) and an initial value is formed for the derivative estimate as a dif-ference between two known channel estimates divided by a time difference therebetween. The known channel estimate may belong to two separate synchronizing sequences or may belong to one synchronizing sequence and a fading time point in which the channel estimate has zero value.
Publication Figure: 7
Radio signals are transmitted over a fading radio channel and are filtered and sampled in a receiver to obtain a digital signal (y(n)). A channel equalizer (5) has an equalizing filter (CEST) in which a channel estimate (h(n/n-L)) is formed with the aid of a known synchronizing sequence. A Viterbi analyzer (VIT) decides symbols (d(n-L)) with the aid of the channel estimate. An es-timated time derivative of the channel estimate (h(n/n-L)) is formed and adapted with the aid of the decided symbols (d(n-L)).
The channel estimate (h(n/n-L)) is adapted to the radio channel by the derivative estimate and the decided symbols (d(n-L)). In the event of fading with rapidly varying signal strength (y(n)), the channel estimate (h(n/n-L)) is adapted slowly and is predicted mainly through the derivative estimate which varies relatively evenly. The digital signal (y(n)) is stored in a memory (M) and an initial value is formed for the derivative estimate as a dif-ference between two known channel estimates divided by a time difference therebetween. The known channel estimate may belong to two separate synchronizing sequences or may belong to one synchronizing sequence and a fading time point in which the channel estimate has zero value.
Publication Figure: 7
Description
A METHOD OF FO~MING A CXANNEL E8TIMATE FOR A TIME-VARYING ~ADIO
C~ANN13L
TEC~NICAL FIELD
The present invention relates to a method of forming a channel estimate for a time-varying radio channel, in which radio signals are transmitted over the radio channel between a transmitter and a receiver and are subjected to disturbances such as multipath propagation, fading and noise, and wherein the radio signals are sampled to obtain received symbols which include information-carrying symbols and at least one symbol-containing synchronizing sequence, said method including the following method steps:
- forming a receiver channel estimate, i.e. an appreciation of parameters in a radio-channel transmission function with the aid of at least one of said synchronizing sequences; and - forming a derivative estimate, i.e. an appreciation of parame-ters in at least a first time derivative of the channel estimate.
BACKGROUND ART
In radio communication, echo signals can occur over a radio channel as a result of multipath propagation of a transmitted radio signal. The transmitted signal reaches a receiver both directly and also via one or more reflected or otherwise deflected signal paths. In the case of digital transmission systems, the echo signals give ri~e to int ~ in~terference. The problems resulting from this interference are well known and solutions to the problems are described in the literature, for instance in an article in IEEE TRANSACTIONS ON INFORMATION THEORY, Vol. IT-18, No. 3, May 1972, G.D. Forney: "Maximum-Likelihood Sequence Estimation of Digital Sequences in the Presence of Intersymbol Interference". The receiver has an adjustable filter which is set with the aid of a known synchronizing sequence. The filter is an image of a sampled impulse response for the channel, usually designated a channel estimate, whose parameters are used to establish the values of transmitted symbols. If the channel is changed with time, the channel estimate is adapted, possibly with 2 207~D61 the aid of the established symbols, for instance as described in an artiale in IEEE TRANSACTIONS ON INFORMATION THEORY, January 1973, pages 120-124, F.R. Magee and J.G. Proakis: "Adaptive Maximum-Likelihood Sequence Estimation for Digital Signaling in the Presence of Intersymbol Interference". In the case of channels which are changed quickly in relation to the transmitted bit frequency, further problems arise because the adaptation process must be effected so quickly that it generates noise itself. The channel estimation herewith becomes sensitive to errorneous decisions. Swedish Patent Application No. 8903526-5 describes an equalizer which overcomes these problems. The equalizer is provided with an analyzer which operates in accordance with a Viterbi algorithm having a requisite number of states. Each state is assigned a channel estimate, which is adapted in accordance with selected state transitions in the Viterbi algorithm. This adaptation is effected without time delay in the analyzer. The problem of erroneous decisions during adaptation of the channel estimation is ~articularly apparent in the case of fading, where the signal strength rapidly decreases and fades away and thereaf-ter rapidly increases. The fading phenomenon is caused byinterfering radio signals and is described in detail by William C.Y. Lee in Mobile Communications Engineering, Chapters 6 and 7, McGraw-Hill, Inc., 1982. In order to overcome the estimation problems caused by fading, there is reguired a better model of the channel than that obtained with the aforesaid methods. One proposal or such improved channel estimation is given in IEEE
TRANSA~TIONS ON coNMnNIcATIoNs~ Vol. 37, No. 9, September 1989, A.P. Clark and S. Hariharan: "Adaptive Channel Estimation for an HF Radio Lin~". This article sugqests generally the use of the channel estimate derivative in channel estimation processes. The article, however, gives no indication as to how the derivative shall be initiated, for instance, during the estimation process so that a reliable derivative will be obtained. When a plurality of symbols are lost, particularly with fading it is essential that such a reliable derivative is obtained.
The object of the present invention is to provide a channel estimate with the aid of a reliable estimate of the time deriva-tive of the channel estimate. Fundamental to the invention is that S complex transmitted symbols have a real and an imaginary partwhich vary relatively regularly with slowly varying derivatives, even when fading is pronounced. A derivative estimate of the channel impulse response is used to adapt and predict a channel estimate, particularly during and after fading occasions in which a number of symbols are lost. The derivative estimate is cal-culated with a starting point from channel estimates for synchro-nizing sequences. According to one alternative, there is formed an initial value for the derivative estimate, such as a difference between two known channel estimates divided by a time distance between said channel estimates. The channel estimate is adapted to received signals and is predicted with the aid of the derivatiYe estimate, which may also be subseguently adapted. A second derivative and derivatives of higher orders may also be used in the channel estimation.
BRIFF DF8CRIPTION OF TH~ DRAWING8 The invention will now be described in more detail with reference to the accompanying drawings, in which Figure 1 is a block schematic of a radio transmission system:
Figure 2 illustrates time slots and signal sequences for a time-; 25 shared radio transmi~sion system;
Figure 3 is a diagram illustrating a complex numeric plan with complex symbols;
Figure 4 is a view of buildings between which radio signals are reflected, and also shows a mobile radio receiver;
Figure S is a diagram illustrating signal strengths of fading radio signals:
Figure 6 is a diagram illustrating levels of the components of the complex symbols;
Figure 7 is a block schematic of a channel equalizer; and Figure 8 is a block schematic of a channel estimation filter.
, ~ ~
- , . . ~-: :
.
., 2~760~
BE8T MODE OF CAR~YINO OUT THg IN~ENTION
A known radio transmission system for time-shared radio com-munication is shown schematically in Figure 1. A transmitter has a unit 1 which generates digital complex-value symbols S(n). These symbols are converted digital/analogues and are transmitted as a signal Y from a unit 2 to a receiving unit 3 of a receiver. The re-ceiver modulates the signal down to an analog received signal y(T) in the baseband. The siqnal is sampled in an analog-digital converter 4 to a received digital signal y(n) and is delivered to a channel equalizer 5. This equalizer produces, with a given time delay, estimated symbols _(n-L) which perform an appreciation of the transmitted complex value symbols S(n). The designation (n) denotes a sampling time point having the number n and the desig-nation (n-L) denotes that the estimated symbols are delayed by a number L sampling intervals. The double signal-paths shown in ~he Figure indicate that the channel between the units 2 and 3 subject the transmitted signal Y to time dispersion. The signal A repre-sents a disturbing signal on the same frequency as that used between the units 2 and 3, a so-called co-channel disturber. Noise and signal-fading also disturb transmission, as explained in - further detail herebelow. The radio transmission system is time-shared with individual time slots l-N according to Figure 2, in which T designates time. A signal sequence comprising at least one synchronizing sequence and one data sequence containing the information desired to be transmitted can be transmitted in each time slot. A signal sequence SSF, which according to the il-lustrated example has two synchronizing sequences SFl and SF2 and two data sequences Dl and D2, is transmitted in a time slot F.
Correspondingly, a signal sequence SSG in an adjacent time slot G
has synchronizing sequences SGl and SG2. It should be noted that, for instance, SF2 need not be a synchronizing sequence in the actual meaning of the ter~, but may be another known sequence. One example in this regard is a so-called CDVCC-sequence, which is used to decide whether or not a received signal belongs to its own connection or derives from the co-channel disturber A. The signal sequences include binary signals, although the aforesaid symbols S(n) are complex and modulated in accordance with QPSK-modulation, ' 20760~1 for instance, as illustrated in ~`igure 3. In a complex numeric plan, with the axis designated I and Q, the four possible values of the symbols S(n) are mar~ed in each quadrant w~th the binary numbers 00, 01, 10 and 11. The time taken to transmit a thus modulated symbol is referred to as symbol time TS. With regard to the analog signal y(T) in the baseband, this signal can be expressed as y(T) = I(T)+~Q(T), where ; = ~
The aforementioned signal fading occurs in the following manner.
Shown in Figure 4 are two buildings 20 and 21 which reflect the transmitted signal Y. The reflected signals interfere with one another between the buildings and a complicated interference pattern is liable to occur with alternating maximum and minimum signal strengths. A mobile receiver 22 which carries the receiver 3, 4, s and which moves through the interference pattern will repeatedly pass the signals of minimum strength. The signal strength is there extremely low and falls beneath the noise level or the signal level from the co-channel disturber A.
Included in the diagram of Figure 5 is a curve 23 which illustrates how the strength P(T) of the signal received by the mobile 22 can vary over the time period of the signal sequence SSF. The noise level is shown by a chain line 24 and the Figure shows how the signal strength P(T) falls beneath a threshold value PO during a time interval TF, which is a centre for a fading occasion. The signal strength is calculated in accordance with a relationship P(T) = I (T) + Q (T) for the aforesaid complex signal y(T).
Whereas the signal strength P(T) varies quickly during fading, the components I(T) and Q(T) of the complex signal vary more evenly and relatively slowly. This is illustrated in Figure 6 with curves B and C, which show that the time derivatives of the complex components are constant, or almost constant, even during the fading occasion TF.
The present invention takes its starting point from the regular occurrence of the time derivatives of these complex signal -- --2076~61 components. The invention will be described in the following with the aid of examples and also with reference to the aforesaid time-shared radio communication system.
Figure 1 illustrates schematically the time-varying radio channel between the transmitter 2 and the receiver 3. Parameters for this radio channel are described in a known manner with its sampled impulse response ~(n) = [hO(n), hl(n),----,hm(n)] (1) where the underlining indicates a vector and index T denotes a vector transposition. The received sample signal y(n) can be described as a convolution between this impulse response for the actual radio channel and the signal d(n) according to y(n) = ~ (n) d(n) + w(n) (2) In this case, d(n) = [d(n), d(d-l),----,d(n-m)] (3) is the synchronizing sequence or the decided data that is considered to be known. The index H signifies Hermit transposition, that is to say transposition and complex con-jugation, and w(n) signifies a disturbance.
There is formulated for the time-varying channel a channel model, with whose help data is decided from the received signal y(n).
According to the invention, the model includes a description of channel parameters and also time derivatives of said parameters.
If only the first derivative is included, the following general relationship is obtained ~(n) = h(n-l) + TS h(n-l) + vl(n) h(n) = h(n-l) + v2(n) (4) The bold style signifies that the magnitudes belong to the channel model which is an estimate of the actual channel. A time deriva-tive is signified with a dot over the symbol concerned, in the usual manner, and vl(n) and v2(n) are magnitudes that are contin-gent on a selected adaptation algorithm. If an LMS-algorithm (Least Mean Square) is selected for vs(n) and v2(n), there i8 ob-tained ~(n) = h(n-l) + TS ~(n-l + ~1 d(n) e*(n) ~(n) = ~(n-l) + ~2 d(n) e*(n) (5) In this case, e(n) = y(n) - ~ (n-1) _(n) is an error signal in the adaptation. The symbol * signifies a complex conjugation and ~1 and ~2 are constants. According to one alternative, these constants may have different values for the different coefficients in h(n) and ~(n). It should be observed that the derivative estimate ~(n) is used when estimating the channel parameters h(n). The relationships (5) are given at a sampling time point n. Signal processing is delayed in the channel equalizer 5 by L number of sampling intervals and signals with this delay are shown in Figures 7 and 8.
The aforesaid algorithm accordir.g to the relationship (5) is used in the equalizer 5, which is shown schematically in Figure 7. The equalizer has a Viterbi analyzer VIT, an adaptive channel estimation filter CEST, a delay circuit DEL and a first memory circuit M. The signal sequences SSF are stored successively in this memory circuit prior to effecting other processing of signals in the equalizer 5. The Viterbi analyzer VIT receives the signal y(n) from the ~emory N and produces the symbols d(n-L), which are esti~ated with a delay of L number of sampling steps. The channel estimation filter CEST receives the estimated symbols ~n-L) and the signals y(n) and y(n-L). These latter signals are the received signals y(n) delayed by L number of sampling steps in the delay circuit DEL. The channel estimation filter CEST delivers the estimated impulse response h~n/n-L) to the Viterbi analyzer VIT.
It should be noted that in addition to the actual radio channel, the channel estimate also includes transmission filter and receiver filter. According to an alternative embodiment of the invention, preliminary decisions from the Viterbi analyzer VITare used instead of the estimated symbols d(n-L). This results in a delay which is shorter than the L number of sampling intervals.
, ' . . ~
.
. - :
' ':
Estimation of the impulse response h(n/n-L) will be described in more detail below with reference to Figure 8.
Figure 8 is a block schematic over the channel estimation filter CEST. The filter has delay elements 6, adjustable coefficient circuits 7, summators 8, a difference former 9, a level circuit 10, two adaptation circuits 11 and 12, a second memory circuit 13 and a prediction circuit 14 having an output 15. The number of coefficient circuits 7 provided will depend on the size of the time dispersion that the radio channel can have expressed in a number of sampling intervals, and according to the illustrated example, three circuits are provided. The delay elements 6 delay the symbols d(n-L) stepwise by one sampling interval. The symbols d(n-L), d(n-L-l) and d(n-L-2) are multiplied in the coefficient circuits 7 by coefficients hO(n-L), h1(n-L~ and h2(n-L) of the channel estimate. The values obtained are summated in the summators 8 to form an estimated signal ~(n-L). The error signal e(n-L) is calculated in the difference former 9 and is delivered to the adaptation circuits 11 and 12. The channel estimate h(n-L) and the derivative estimate h(n-L) are adapted in the circuits with the aid of the estimated symbols d(n-L~. The channel estimate is also influenced by the level P0, which is a threshold value for the signal strength of the received signal y(n). The threshold value, shown in Figure 5, is delivered from the level circuit 10.
The prediction circuit 14 functions to predict the obtained channel estimate with the aid of the derivative estimate. The output 15 of the prediction circuit is connected to the Viterbi analyzer VIT.
When signals are received without fading during the time interval TF, the channel estimation circuit CEST operates in the following manner. The channel estimate h(n-L) and its derivative ~(n-L) are calculated first, with the aid of the synchronizing sequence SFl.
This is effected with the aid of the synchronizing sequence d(n-L) stored in the memory circuit 13 and the received, observed syn-chronizing sequence y(n-L). In this case, there is first effected , 35 a channel correlation which gives an initial value h(0) for the channel estimate according to a relationship ^: :
. , , :`
, 9 2~76~
~(O) = 1/K ~ d (i) y~ l ) (6) 1 =1 The relationship is shown here for the second coefficient hl(n) in the channel estimate h(n) = [hO(n),h1(n),h2(n)] . K signifies the number of symbols in the synchronizing sequence SFl used for the correlation. A zero value can be used as the initial value of the derivative. The synchronizing seguence SFl is then used once more, by successively adapting the channel estimate and the derivative estimate with the aid of the adaption circuits 11 and 12 according to the algorithm in the relationship ~5) above. The error signal e(n-L) is minimized in this procedure so that the coefficients ho(n-L), hl(n-L) and h2(n-L) formthe channel estimate. Adaptation of the channel estimate and the derivative estimate is then continued successively during the transmission of the data sequence Dl, with the aid of decided data d(n-L) from the Viterbi algorithm VIT. The procedure is repeated for the synchronizing sequence SF2 and the data sequence D2. According to a simplified alternative, adaptation to the data sequence is performed directly after the channel correlation. The channel estimation obtained in the aforesaid manner applies to the time point (n-L). The channel estimation is predicted in the prediction circuit 14 from the time point (n-L) to the time point (n) with the aid of the derivative estimation. ~his results in a channel estimation intended for the Viterbianalyzer, this estimation being designated h(n/n-L~ in the Figure. Thust the derivative estimate h(n-L) is not only used to calculate the channel estimate according to the relationship (5), but is also used to predict the channel estimation up to time point (n). In this way, the values cf the estimated symbols d~n-L1 obtained from the Viterbi analyzer VIT will be more certain, which in turn enables the channel estimation in the circuits 11 and 12 to be improved. A simplified alternative excludes prediction in the circuit 14. The channel estimate is delivered to the Viterbi analyzer from the output 15.
o 2076061 The following procedural steps are carried out in the channel estimation circuit CEST in the event of fading. The power of the received signal y(n) is calculated in the level circuit 10 and is compared with the threshold value PO. This calculation is carried out with the aid of the signal y(n) received from the first memory circuit M prior to processing the entire signal sequence SSF in the channel equalizer. This provides information as to whether fading occurs before actual estimation of the symbols ~(n-L) takes place. If the signal strength is below the threshold value PO, that i8 to say fading is considered to prevail, a corresponding signal is sent to the adaptation circuit 11. In this case, the constant ~1 in the relationship (5) is given a small value or a zero value, so that the channel estimate h(n) is adapted substan-tially or completely with the aid of the derivative estimate h(n).
The signal strength P(T) is changed very quickly in the event of fading and is very low, see Figure 5, and adaptation with the error signal e(n) becomes very uncertain and generates noise. On the other hand, the derivatives of the signal components I ~T) and Q~T) vary much more slowly and more regularly than the signal strength P(T), as shown in Figure 6. This causes the derivative h(n) of the actual radio channel also to vary slowly and regularly. Conse-quently, adaptation with the aid of the derivative estimate h(n) becomes reliable, even during the fading occasion TF. If the constant ~1 is set to zero during the fading occasion, channel estimation becomes a simple prediction with the aid of the deriva-tive estimate h(n). It is also possible to hold the derivative estimate constant during fading, by setting the constant ~2 to zero. This is effected by sending a signal from the level circuit 10 to the adaptation circuit 12, as indicated by a broken-line connection in Figure 8. The constants ~1 and ~2 are set to their original values when the signal strength exceeds the threshold value P(O) at the end of the fading occasion. The threshold value PO can be set to a very low value or to a zero value in the case of a simplified method. In this case, the estimation according to the relationship (5) is also carried out during the fading occasion TF.
.
,.
, ' . : :
An initial value ~(o~ for the derivative estimate can be obtained in the following manner. The channel estimate h(0) and h(R) for the two known synchronizing sequences SFl and SF2 are calculated with the aid of the received signals y(n) stored in the first memory M. The initial value is obtained in accordance with the relationship h(0) = ~h(R) - h(0))/R (~) where R is the time distance between the synchronizing sequences, calculated as the number of symbols between said sequences. In the time-shared system according to Figure 2, a corresponding initial value can be calculated with a starting point, for instance, from SF2 and the synchronizing sequence SGl, which belongs to the bordering time slot G. The time slots and F and G lie on the same carrier freguency and the subscriber on the time slot F can be permitted to listen on the synchronizing sequence SG1 in order to be able to carry out the calculation according to the relationship (7) above. It should be observed that only the mobile can utilize this possibility with synchronizing sequences from separate time slots. It should also be observed that, for instance, SF2 need not be a true synchronizing sequence, but may be some other known sequence, as indicated above. It is also possible to utilize a fading occasion to calculate an initial value of the derivative e~timation. In the event of a fading occasion, the channel estima-te is set to ~(R~ = 0 and the initial value of the derivative estimate is calculated according to ~ (0) = -~(0)/R
where k(o) is the channel estimation for one of the synchronizing sequences and R is the number of symbols between the fading occasion and said synchronizing sequence. When calculating the start values for the derivative estimate as described above, the channel estimate can be calculated in a simplified fashion, without the aid of the derivative estimate. In this case, cal-culation of the channel estimate for the synchronizing sequences assumes that the synchronizing sequences have not been afflicted by fading.
In the aforedescribed exemplifying embodiment of the inventive method, only the first time derivative of the channel estimate has 12 20 76'Q 6 been included. It lies within the scope of the invention, however, to use derivatives of higher orders. For instance, the relation-ship (5~ can be expanded with an equation for calculating the second derivative of the channel estimation, and this is included in the expressions for the channel estimate and its first deriva-tive.
It is assumed that the modulation shown in Figure 3 is a signal modulation in which the symbol vectors are given in relation to a fixed reference axis. Alternatively, differential modulation can be applied, in which the phase position of a symbol is instead given in relation to the phase position of a preceding symbol. The invention can be used for signals modulated in this way. The phase position of the receiver is given a value, for instance a zero value, at the start of the first synchronizing word SFl. This enables a well-defined starting value h(0) to be calculated for the channel estimation. The start phase for the second synchroni-zing wora SF2 will depend on the phase positions of the symbols in the data sequence Dl. These phase positions are unknown and the starting phase for SF2 is therewith indefinite. If differential QPSK-modulation is used, four possible values for the starting phase in SF2 are found and therewith also four possible initial ~alues for the derivative estimate h(0), which is calculated according to the relationship (7) above. The most likely of these initial values is selected, by demodulating the data sequence Dl with a starting point from all four of the initial values and selecting the demodulation which gives the least error signal. In the exemplified case using a Viterbi equalizer, the error signal i8 equal to the metric value calculated in accordance with the Viterbi algorithm. In order to determine the least error signal and the initial value, it is sufficient to demodulate solely a part of the data sequence Dl or to only use the synchronizing sequence SFl as a training signal and train four times on this sequence.
Although the inventive method has been described in the aforegoing with reference to a Viterbi equalizer 5, it will be understood that the method can also be used for other types of equalizer, for ' 13 207~061 instance DFE-utilizer (Decision Feedback Equalizer). The method can be applied to advantage immediately the radio channel has a rapidly varying transmission function and not solely in the event of fading.
C~ANN13L
TEC~NICAL FIELD
The present invention relates to a method of forming a channel estimate for a time-varying radio channel, in which radio signals are transmitted over the radio channel between a transmitter and a receiver and are subjected to disturbances such as multipath propagation, fading and noise, and wherein the radio signals are sampled to obtain received symbols which include information-carrying symbols and at least one symbol-containing synchronizing sequence, said method including the following method steps:
- forming a receiver channel estimate, i.e. an appreciation of parameters in a radio-channel transmission function with the aid of at least one of said synchronizing sequences; and - forming a derivative estimate, i.e. an appreciation of parame-ters in at least a first time derivative of the channel estimate.
BACKGROUND ART
In radio communication, echo signals can occur over a radio channel as a result of multipath propagation of a transmitted radio signal. The transmitted signal reaches a receiver both directly and also via one or more reflected or otherwise deflected signal paths. In the case of digital transmission systems, the echo signals give ri~e to int ~ in~terference. The problems resulting from this interference are well known and solutions to the problems are described in the literature, for instance in an article in IEEE TRANSACTIONS ON INFORMATION THEORY, Vol. IT-18, No. 3, May 1972, G.D. Forney: "Maximum-Likelihood Sequence Estimation of Digital Sequences in the Presence of Intersymbol Interference". The receiver has an adjustable filter which is set with the aid of a known synchronizing sequence. The filter is an image of a sampled impulse response for the channel, usually designated a channel estimate, whose parameters are used to establish the values of transmitted symbols. If the channel is changed with time, the channel estimate is adapted, possibly with 2 207~D61 the aid of the established symbols, for instance as described in an artiale in IEEE TRANSACTIONS ON INFORMATION THEORY, January 1973, pages 120-124, F.R. Magee and J.G. Proakis: "Adaptive Maximum-Likelihood Sequence Estimation for Digital Signaling in the Presence of Intersymbol Interference". In the case of channels which are changed quickly in relation to the transmitted bit frequency, further problems arise because the adaptation process must be effected so quickly that it generates noise itself. The channel estimation herewith becomes sensitive to errorneous decisions. Swedish Patent Application No. 8903526-5 describes an equalizer which overcomes these problems. The equalizer is provided with an analyzer which operates in accordance with a Viterbi algorithm having a requisite number of states. Each state is assigned a channel estimate, which is adapted in accordance with selected state transitions in the Viterbi algorithm. This adaptation is effected without time delay in the analyzer. The problem of erroneous decisions during adaptation of the channel estimation is ~articularly apparent in the case of fading, where the signal strength rapidly decreases and fades away and thereaf-ter rapidly increases. The fading phenomenon is caused byinterfering radio signals and is described in detail by William C.Y. Lee in Mobile Communications Engineering, Chapters 6 and 7, McGraw-Hill, Inc., 1982. In order to overcome the estimation problems caused by fading, there is reguired a better model of the channel than that obtained with the aforesaid methods. One proposal or such improved channel estimation is given in IEEE
TRANSA~TIONS ON coNMnNIcATIoNs~ Vol. 37, No. 9, September 1989, A.P. Clark and S. Hariharan: "Adaptive Channel Estimation for an HF Radio Lin~". This article sugqests generally the use of the channel estimate derivative in channel estimation processes. The article, however, gives no indication as to how the derivative shall be initiated, for instance, during the estimation process so that a reliable derivative will be obtained. When a plurality of symbols are lost, particularly with fading it is essential that such a reliable derivative is obtained.
The object of the present invention is to provide a channel estimate with the aid of a reliable estimate of the time deriva-tive of the channel estimate. Fundamental to the invention is that S complex transmitted symbols have a real and an imaginary partwhich vary relatively regularly with slowly varying derivatives, even when fading is pronounced. A derivative estimate of the channel impulse response is used to adapt and predict a channel estimate, particularly during and after fading occasions in which a number of symbols are lost. The derivative estimate is cal-culated with a starting point from channel estimates for synchro-nizing sequences. According to one alternative, there is formed an initial value for the derivative estimate, such as a difference between two known channel estimates divided by a time distance between said channel estimates. The channel estimate is adapted to received signals and is predicted with the aid of the derivatiYe estimate, which may also be subseguently adapted. A second derivative and derivatives of higher orders may also be used in the channel estimation.
BRIFF DF8CRIPTION OF TH~ DRAWING8 The invention will now be described in more detail with reference to the accompanying drawings, in which Figure 1 is a block schematic of a radio transmission system:
Figure 2 illustrates time slots and signal sequences for a time-; 25 shared radio transmi~sion system;
Figure 3 is a diagram illustrating a complex numeric plan with complex symbols;
Figure 4 is a view of buildings between which radio signals are reflected, and also shows a mobile radio receiver;
Figure S is a diagram illustrating signal strengths of fading radio signals:
Figure 6 is a diagram illustrating levels of the components of the complex symbols;
Figure 7 is a block schematic of a channel equalizer; and Figure 8 is a block schematic of a channel estimation filter.
, ~ ~
- , . . ~-: :
.
., 2~760~
BE8T MODE OF CAR~YINO OUT THg IN~ENTION
A known radio transmission system for time-shared radio com-munication is shown schematically in Figure 1. A transmitter has a unit 1 which generates digital complex-value symbols S(n). These symbols are converted digital/analogues and are transmitted as a signal Y from a unit 2 to a receiving unit 3 of a receiver. The re-ceiver modulates the signal down to an analog received signal y(T) in the baseband. The siqnal is sampled in an analog-digital converter 4 to a received digital signal y(n) and is delivered to a channel equalizer 5. This equalizer produces, with a given time delay, estimated symbols _(n-L) which perform an appreciation of the transmitted complex value symbols S(n). The designation (n) denotes a sampling time point having the number n and the desig-nation (n-L) denotes that the estimated symbols are delayed by a number L sampling intervals. The double signal-paths shown in ~he Figure indicate that the channel between the units 2 and 3 subject the transmitted signal Y to time dispersion. The signal A repre-sents a disturbing signal on the same frequency as that used between the units 2 and 3, a so-called co-channel disturber. Noise and signal-fading also disturb transmission, as explained in - further detail herebelow. The radio transmission system is time-shared with individual time slots l-N according to Figure 2, in which T designates time. A signal sequence comprising at least one synchronizing sequence and one data sequence containing the information desired to be transmitted can be transmitted in each time slot. A signal sequence SSF, which according to the il-lustrated example has two synchronizing sequences SFl and SF2 and two data sequences Dl and D2, is transmitted in a time slot F.
Correspondingly, a signal sequence SSG in an adjacent time slot G
has synchronizing sequences SGl and SG2. It should be noted that, for instance, SF2 need not be a synchronizing sequence in the actual meaning of the ter~, but may be another known sequence. One example in this regard is a so-called CDVCC-sequence, which is used to decide whether or not a received signal belongs to its own connection or derives from the co-channel disturber A. The signal sequences include binary signals, although the aforesaid symbols S(n) are complex and modulated in accordance with QPSK-modulation, ' 20760~1 for instance, as illustrated in ~`igure 3. In a complex numeric plan, with the axis designated I and Q, the four possible values of the symbols S(n) are mar~ed in each quadrant w~th the binary numbers 00, 01, 10 and 11. The time taken to transmit a thus modulated symbol is referred to as symbol time TS. With regard to the analog signal y(T) in the baseband, this signal can be expressed as y(T) = I(T)+~Q(T), where ; = ~
The aforementioned signal fading occurs in the following manner.
Shown in Figure 4 are two buildings 20 and 21 which reflect the transmitted signal Y. The reflected signals interfere with one another between the buildings and a complicated interference pattern is liable to occur with alternating maximum and minimum signal strengths. A mobile receiver 22 which carries the receiver 3, 4, s and which moves through the interference pattern will repeatedly pass the signals of minimum strength. The signal strength is there extremely low and falls beneath the noise level or the signal level from the co-channel disturber A.
Included in the diagram of Figure 5 is a curve 23 which illustrates how the strength P(T) of the signal received by the mobile 22 can vary over the time period of the signal sequence SSF. The noise level is shown by a chain line 24 and the Figure shows how the signal strength P(T) falls beneath a threshold value PO during a time interval TF, which is a centre for a fading occasion. The signal strength is calculated in accordance with a relationship P(T) = I (T) + Q (T) for the aforesaid complex signal y(T).
Whereas the signal strength P(T) varies quickly during fading, the components I(T) and Q(T) of the complex signal vary more evenly and relatively slowly. This is illustrated in Figure 6 with curves B and C, which show that the time derivatives of the complex components are constant, or almost constant, even during the fading occasion TF.
The present invention takes its starting point from the regular occurrence of the time derivatives of these complex signal -- --2076~61 components. The invention will be described in the following with the aid of examples and also with reference to the aforesaid time-shared radio communication system.
Figure 1 illustrates schematically the time-varying radio channel between the transmitter 2 and the receiver 3. Parameters for this radio channel are described in a known manner with its sampled impulse response ~(n) = [hO(n), hl(n),----,hm(n)] (1) where the underlining indicates a vector and index T denotes a vector transposition. The received sample signal y(n) can be described as a convolution between this impulse response for the actual radio channel and the signal d(n) according to y(n) = ~ (n) d(n) + w(n) (2) In this case, d(n) = [d(n), d(d-l),----,d(n-m)] (3) is the synchronizing sequence or the decided data that is considered to be known. The index H signifies Hermit transposition, that is to say transposition and complex con-jugation, and w(n) signifies a disturbance.
There is formulated for the time-varying channel a channel model, with whose help data is decided from the received signal y(n).
According to the invention, the model includes a description of channel parameters and also time derivatives of said parameters.
If only the first derivative is included, the following general relationship is obtained ~(n) = h(n-l) + TS h(n-l) + vl(n) h(n) = h(n-l) + v2(n) (4) The bold style signifies that the magnitudes belong to the channel model which is an estimate of the actual channel. A time deriva-tive is signified with a dot over the symbol concerned, in the usual manner, and vl(n) and v2(n) are magnitudes that are contin-gent on a selected adaptation algorithm. If an LMS-algorithm (Least Mean Square) is selected for vs(n) and v2(n), there i8 ob-tained ~(n) = h(n-l) + TS ~(n-l + ~1 d(n) e*(n) ~(n) = ~(n-l) + ~2 d(n) e*(n) (5) In this case, e(n) = y(n) - ~ (n-1) _(n) is an error signal in the adaptation. The symbol * signifies a complex conjugation and ~1 and ~2 are constants. According to one alternative, these constants may have different values for the different coefficients in h(n) and ~(n). It should be observed that the derivative estimate ~(n) is used when estimating the channel parameters h(n). The relationships (5) are given at a sampling time point n. Signal processing is delayed in the channel equalizer 5 by L number of sampling intervals and signals with this delay are shown in Figures 7 and 8.
The aforesaid algorithm accordir.g to the relationship (5) is used in the equalizer 5, which is shown schematically in Figure 7. The equalizer has a Viterbi analyzer VIT, an adaptive channel estimation filter CEST, a delay circuit DEL and a first memory circuit M. The signal sequences SSF are stored successively in this memory circuit prior to effecting other processing of signals in the equalizer 5. The Viterbi analyzer VIT receives the signal y(n) from the ~emory N and produces the symbols d(n-L), which are esti~ated with a delay of L number of sampling steps. The channel estimation filter CEST receives the estimated symbols ~n-L) and the signals y(n) and y(n-L). These latter signals are the received signals y(n) delayed by L number of sampling steps in the delay circuit DEL. The channel estimation filter CEST delivers the estimated impulse response h~n/n-L) to the Viterbi analyzer VIT.
It should be noted that in addition to the actual radio channel, the channel estimate also includes transmission filter and receiver filter. According to an alternative embodiment of the invention, preliminary decisions from the Viterbi analyzer VITare used instead of the estimated symbols d(n-L). This results in a delay which is shorter than the L number of sampling intervals.
, ' . . ~
.
. - :
' ':
Estimation of the impulse response h(n/n-L) will be described in more detail below with reference to Figure 8.
Figure 8 is a block schematic over the channel estimation filter CEST. The filter has delay elements 6, adjustable coefficient circuits 7, summators 8, a difference former 9, a level circuit 10, two adaptation circuits 11 and 12, a second memory circuit 13 and a prediction circuit 14 having an output 15. The number of coefficient circuits 7 provided will depend on the size of the time dispersion that the radio channel can have expressed in a number of sampling intervals, and according to the illustrated example, three circuits are provided. The delay elements 6 delay the symbols d(n-L) stepwise by one sampling interval. The symbols d(n-L), d(n-L-l) and d(n-L-2) are multiplied in the coefficient circuits 7 by coefficients hO(n-L), h1(n-L~ and h2(n-L) of the channel estimate. The values obtained are summated in the summators 8 to form an estimated signal ~(n-L). The error signal e(n-L) is calculated in the difference former 9 and is delivered to the adaptation circuits 11 and 12. The channel estimate h(n-L) and the derivative estimate h(n-L) are adapted in the circuits with the aid of the estimated symbols d(n-L~. The channel estimate is also influenced by the level P0, which is a threshold value for the signal strength of the received signal y(n). The threshold value, shown in Figure 5, is delivered from the level circuit 10.
The prediction circuit 14 functions to predict the obtained channel estimate with the aid of the derivative estimate. The output 15 of the prediction circuit is connected to the Viterbi analyzer VIT.
When signals are received without fading during the time interval TF, the channel estimation circuit CEST operates in the following manner. The channel estimate h(n-L) and its derivative ~(n-L) are calculated first, with the aid of the synchronizing sequence SFl.
This is effected with the aid of the synchronizing sequence d(n-L) stored in the memory circuit 13 and the received, observed syn-chronizing sequence y(n-L). In this case, there is first effected , 35 a channel correlation which gives an initial value h(0) for the channel estimate according to a relationship ^: :
. , , :`
, 9 2~76~
~(O) = 1/K ~ d (i) y~ l ) (6) 1 =1 The relationship is shown here for the second coefficient hl(n) in the channel estimate h(n) = [hO(n),h1(n),h2(n)] . K signifies the number of symbols in the synchronizing sequence SFl used for the correlation. A zero value can be used as the initial value of the derivative. The synchronizing seguence SFl is then used once more, by successively adapting the channel estimate and the derivative estimate with the aid of the adaption circuits 11 and 12 according to the algorithm in the relationship ~5) above. The error signal e(n-L) is minimized in this procedure so that the coefficients ho(n-L), hl(n-L) and h2(n-L) formthe channel estimate. Adaptation of the channel estimate and the derivative estimate is then continued successively during the transmission of the data sequence Dl, with the aid of decided data d(n-L) from the Viterbi algorithm VIT. The procedure is repeated for the synchronizing sequence SF2 and the data sequence D2. According to a simplified alternative, adaptation to the data sequence is performed directly after the channel correlation. The channel estimation obtained in the aforesaid manner applies to the time point (n-L). The channel estimation is predicted in the prediction circuit 14 from the time point (n-L) to the time point (n) with the aid of the derivative estimation. ~his results in a channel estimation intended for the Viterbianalyzer, this estimation being designated h(n/n-L~ in the Figure. Thust the derivative estimate h(n-L) is not only used to calculate the channel estimate according to the relationship (5), but is also used to predict the channel estimation up to time point (n). In this way, the values cf the estimated symbols d~n-L1 obtained from the Viterbi analyzer VIT will be more certain, which in turn enables the channel estimation in the circuits 11 and 12 to be improved. A simplified alternative excludes prediction in the circuit 14. The channel estimate is delivered to the Viterbi analyzer from the output 15.
o 2076061 The following procedural steps are carried out in the channel estimation circuit CEST in the event of fading. The power of the received signal y(n) is calculated in the level circuit 10 and is compared with the threshold value PO. This calculation is carried out with the aid of the signal y(n) received from the first memory circuit M prior to processing the entire signal sequence SSF in the channel equalizer. This provides information as to whether fading occurs before actual estimation of the symbols ~(n-L) takes place. If the signal strength is below the threshold value PO, that i8 to say fading is considered to prevail, a corresponding signal is sent to the adaptation circuit 11. In this case, the constant ~1 in the relationship (5) is given a small value or a zero value, so that the channel estimate h(n) is adapted substan-tially or completely with the aid of the derivative estimate h(n).
The signal strength P(T) is changed very quickly in the event of fading and is very low, see Figure 5, and adaptation with the error signal e(n) becomes very uncertain and generates noise. On the other hand, the derivatives of the signal components I ~T) and Q~T) vary much more slowly and more regularly than the signal strength P(T), as shown in Figure 6. This causes the derivative h(n) of the actual radio channel also to vary slowly and regularly. Conse-quently, adaptation with the aid of the derivative estimate h(n) becomes reliable, even during the fading occasion TF. If the constant ~1 is set to zero during the fading occasion, channel estimation becomes a simple prediction with the aid of the deriva-tive estimate h(n). It is also possible to hold the derivative estimate constant during fading, by setting the constant ~2 to zero. This is effected by sending a signal from the level circuit 10 to the adaptation circuit 12, as indicated by a broken-line connection in Figure 8. The constants ~1 and ~2 are set to their original values when the signal strength exceeds the threshold value P(O) at the end of the fading occasion. The threshold value PO can be set to a very low value or to a zero value in the case of a simplified method. In this case, the estimation according to the relationship (5) is also carried out during the fading occasion TF.
.
,.
, ' . : :
An initial value ~(o~ for the derivative estimate can be obtained in the following manner. The channel estimate h(0) and h(R) for the two known synchronizing sequences SFl and SF2 are calculated with the aid of the received signals y(n) stored in the first memory M. The initial value is obtained in accordance with the relationship h(0) = ~h(R) - h(0))/R (~) where R is the time distance between the synchronizing sequences, calculated as the number of symbols between said sequences. In the time-shared system according to Figure 2, a corresponding initial value can be calculated with a starting point, for instance, from SF2 and the synchronizing sequence SGl, which belongs to the bordering time slot G. The time slots and F and G lie on the same carrier freguency and the subscriber on the time slot F can be permitted to listen on the synchronizing sequence SG1 in order to be able to carry out the calculation according to the relationship (7) above. It should be observed that only the mobile can utilize this possibility with synchronizing sequences from separate time slots. It should also be observed that, for instance, SF2 need not be a true synchronizing sequence, but may be some other known sequence, as indicated above. It is also possible to utilize a fading occasion to calculate an initial value of the derivative e~timation. In the event of a fading occasion, the channel estima-te is set to ~(R~ = 0 and the initial value of the derivative estimate is calculated according to ~ (0) = -~(0)/R
where k(o) is the channel estimation for one of the synchronizing sequences and R is the number of symbols between the fading occasion and said synchronizing sequence. When calculating the start values for the derivative estimate as described above, the channel estimate can be calculated in a simplified fashion, without the aid of the derivative estimate. In this case, cal-culation of the channel estimate for the synchronizing sequences assumes that the synchronizing sequences have not been afflicted by fading.
In the aforedescribed exemplifying embodiment of the inventive method, only the first time derivative of the channel estimate has 12 20 76'Q 6 been included. It lies within the scope of the invention, however, to use derivatives of higher orders. For instance, the relation-ship (5~ can be expanded with an equation for calculating the second derivative of the channel estimation, and this is included in the expressions for the channel estimate and its first deriva-tive.
It is assumed that the modulation shown in Figure 3 is a signal modulation in which the symbol vectors are given in relation to a fixed reference axis. Alternatively, differential modulation can be applied, in which the phase position of a symbol is instead given in relation to the phase position of a preceding symbol. The invention can be used for signals modulated in this way. The phase position of the receiver is given a value, for instance a zero value, at the start of the first synchronizing word SFl. This enables a well-defined starting value h(0) to be calculated for the channel estimation. The start phase for the second synchroni-zing wora SF2 will depend on the phase positions of the symbols in the data sequence Dl. These phase positions are unknown and the starting phase for SF2 is therewith indefinite. If differential QPSK-modulation is used, four possible values for the starting phase in SF2 are found and therewith also four possible initial ~alues for the derivative estimate h(0), which is calculated according to the relationship (7) above. The most likely of these initial values is selected, by demodulating the data sequence Dl with a starting point from all four of the initial values and selecting the demodulation which gives the least error signal. In the exemplified case using a Viterbi equalizer, the error signal i8 equal to the metric value calculated in accordance with the Viterbi algorithm. In order to determine the least error signal and the initial value, it is sufficient to demodulate solely a part of the data sequence Dl or to only use the synchronizing sequence SFl as a training signal and train four times on this sequence.
Although the inventive method has been described in the aforegoing with reference to a Viterbi equalizer 5, it will be understood that the method can also be used for other types of equalizer, for ' 13 207~061 instance DFE-utilizer (Decision Feedback Equalizer). The method can be applied to advantage immediately the radio channel has a rapidly varying transmission function and not solely in the event of fading.
Claims (11)
1. A method of forming a channel estimate for a time-varying radio channel in which radio signals are transmitted over the radio channel between a transmitter and a receiver and are subjected to disturbances such as multipath propagation, fading and noise, and in which the radio signals are sampled to obtain received symbols which include information-carrying symbols and at least one known symbol sequence such as a synchronizing se-quence, said method comprising the steps of:
- forming a receiver channel estimate, i.e. an appreciation of parameters in a transmission function of the radio channel, with the aid of at least one of the synchronizing sequences; and - forming a derivative estimate, i.e. an appreciation of parame-ters in at least a first time derivative of the channel estima-tion, c h a r a c t e r i z e d in that the method further comprises the steps of:
- registering the signal strength of the received radio signals (y(n));
- assigning an initial value to the parameters of the derivative estimate (h(n-L));
- adapting the derivative estimate (h(n-L)) in dependence on the subsequently received symbols (y(n)); and - adapting the channel estimate (h(n-L)) in the case of a low registered signal strength that lies beneath a threshold value (PO), essentially with the aid of the derivative estimate (h(n-L)).
- forming a receiver channel estimate, i.e. an appreciation of parameters in a transmission function of the radio channel, with the aid of at least one of the synchronizing sequences; and - forming a derivative estimate, i.e. an appreciation of parame-ters in at least a first time derivative of the channel estima-tion, c h a r a c t e r i z e d in that the method further comprises the steps of:
- registering the signal strength of the received radio signals (y(n));
- assigning an initial value to the parameters of the derivative estimate (h(n-L));
- adapting the derivative estimate (h(n-L)) in dependence on the subsequently received symbols (y(n)); and - adapting the channel estimate (h(n-L)) in the case of a low registered signal strength that lies beneath a threshold value (PO), essentially with the aid of the derivative estimate (h(n-L)).
2. A method according to Claim 1, in which the known symbol sequences reoccurs at desired time intervals and received symbols are stored in a memory provided in the receiver, c h a r a c t e-r i z e d in that the initial value (h(O)) of the derivative estimate parameters is formed in accordance with the following method:
- forming two channel estimates (h(R), h(O)) belonging to two of the known symbol sequences, of which at least one is one of the synchronizing sequences (SF1, SF2);
- calculating a time distance (R) between the two channel estimations (h(R), h(O) );
- forming a difference (h(R)-h(O)) between the two channel estimations; and - forming a quotient between the aforesaid difference and the calculated time distance (R), said quotient constituting the initial value (h(O)) of the derivative estimate (h(n-L)).
- forming two channel estimates (h(R), h(O)) belonging to two of the known symbol sequences, of which at least one is one of the synchronizing sequences (SF1, SF2);
- calculating a time distance (R) between the two channel estimations (h(R), h(O) );
- forming a difference (h(R)-h(O)) between the two channel estimations; and - forming a quotient between the aforesaid difference and the calculated time distance (R), said quotient constituting the initial value (h(O)) of the derivative estimate (h(n-L)).
3. A method according to Claim 1, in which received symbols are stored in a memory in the receiver, c h a r a c t e r i z e d in that the initial value of the parame-ters of the derivative estimate (h(n-L)) is formed by the fol-lowing method:
- establishing a fading time point (TF) with disappearing signal strength:
- assigning zero values to the parameters of the channel estimate (h(R)) in the established fading time point (TF);
- forming a channel estimate (h(O)) for the synchronizing sequence (SF1):
- calculating a time distance (R) between the synchronizing sequence (SF1) and the established fading time point (TF); and - forming a quotient between said channel estimate (h(O)) and the calculated time distance (R), said quotient constituting the initial value (h(O)) of the derivative estimate.
- establishing a fading time point (TF) with disappearing signal strength:
- assigning zero values to the parameters of the channel estimate (h(R)) in the established fading time point (TF);
- forming a channel estimate (h(O)) for the synchronizing sequence (SF1):
- calculating a time distance (R) between the synchronizing sequence (SF1) and the established fading time point (TF); and - forming a quotient between said channel estimate (h(O)) and the calculated time distance (R), said quotient constituting the initial value (h(O)) of the derivative estimate.
4. A method for forming a channel estimate for a time-varying radio channel in which radio signals are transmitted over the radio channel between a transmitter and a receiver and are subjected to disturbances, such as multipath propagation, fading and noise, and in which the radio signals are sampled to obtain received symbols which include information-carrying symbols and at least one known symbol sequence such as a synchronizing se-quence, said method comprising the following method steps:
- storing received symbols in a receiver memory;
- forming a receiver channel estimate, i.e. an appreciation of the parameters in a transmission function of the radio channel, with the aid of at least one of the synchronizing sequences; and - forming a derivative estimate, i.e. an appreciation of parame-ters in at least one first time derivative of the channel es-timate, c h a r a c t e r i z e d in that the method further comprises the following method steps:
- forming two time-separated channel estimates (h(O), h(R)), wherein at least one of said channel estimates belongs to one of the synchronizing sequences (SF1, SF2);
- calculating a time distance (R) between the two channel estimates;
- forming a difference between both channel estimates (h(O), h(R));
- forming a quotient between the said difference and the cal-culated time distance (R), said quotient constituting an initial value (h(O)) of the derivative estimate (h(n-L)): and - adapting the channel estimate (h(n-L)) in dependence on subsequently received symbols (y(n)) with the aid of the deriva-tive estimate (h(n-L)).
- storing received symbols in a receiver memory;
- forming a receiver channel estimate, i.e. an appreciation of the parameters in a transmission function of the radio channel, with the aid of at least one of the synchronizing sequences; and - forming a derivative estimate, i.e. an appreciation of parame-ters in at least one first time derivative of the channel es-timate, c h a r a c t e r i z e d in that the method further comprises the following method steps:
- forming two time-separated channel estimates (h(O), h(R)), wherein at least one of said channel estimates belongs to one of the synchronizing sequences (SF1, SF2);
- calculating a time distance (R) between the two channel estimates;
- forming a difference between both channel estimates (h(O), h(R));
- forming a quotient between the said difference and the cal-culated time distance (R), said quotient constituting an initial value (h(O)) of the derivative estimate (h(n-L)): and - adapting the channel estimate (h(n-L)) in dependence on subsequently received symbols (y(n)) with the aid of the deriva-tive estimate (h(n-L)).
5. A method according to Claim 4, c h a r a c -t e r i z e d by registering the signal strength of the received radio signals (y(T)) and - predicting the channel estimate (h(n-L)) with the aid of the derivative estimate (h(n-l)) when registering a low signal strength beneath a threshold value (PO).
6. A method according to Claim 4 or 5, c h a r a c t e r i z e d in that the second of the two time-separated channel estimates (h(R), h(O)) belongs to one of the known symbol sequences (SF1, SF2).
7. A method according to Claim 5, c h a r a c t e r i z e d in that the method includes the following method steps:
- establishing a time point (TF) of a fading occasion with disappearing signal strength; and - assigning a zero value to the parameters of the channel estimate (h(n-L)) in the established fading time point (TF) and making the second of the two time-separated channel estimates (h(R)).
- establishing a time point (TF) of a fading occasion with disappearing signal strength; and - assigning a zero value to the parameters of the channel estimate (h(n-L)) in the established fading time point (TF) and making the second of the two time-separated channel estimates (h(R)).
8. A method according to Claim 2 or 6, in which the radio transmission is time-shared with at least two separate time slots for separate receivers on a common carrier frequency, c h a r a c-t e r i z e d in that of the two known symbol sequences, one is the synchronizing sequence (SF1) for one time slot (F) and the other is a synchronizing sequence (SG1) for the other time slot (G).
9. A method according to any one of Claims 1-8, c h a r a c t e r i z e d in that when the signal strength lies above the aforesaid threshold value (PO), the derivative estimate (h(n-L)) is adapted in dependence on the subsequently received symbols (y(n)) and the channel estimate (h(n-L)) is adapted in accordance with both the derivative estimate (h(n-L)) and the subsequently received symbols (y(n)).
10. A method according to any one of Claims 1-9, in which the subsequently received digital signals are processed in an analyzer to estimated, time-delayed symbols, c h a r a c t e r i z e d in that said channel estimate (h(n-L)), which is applicable at a delayed time point (n-L) is predicted forwards in time a desired number of sampling time points (L) with the aid of the derivative estimate (h(n-L)) to a channel estimate (h(n/n-L)) intended for the analyzer (VIT).
11. A method according to any one of Claims 2, 4 or 5-10, in which the symbols are modulated differentially according to a rule which allows at least two separate phase positions which are given in relation to the phase positions of preceding symbols, c h a r a c t e r i z e d by - assigning a symbol (y(n)) in one of the known symbol sequences (SF1) a desired one of the different phase positions (00, 01, 10, 11);
- subsequently assigning a symbol in the other of the known symbol sequences (SF2) to each of the separate phase positions (00, 01, 10, 11);
- demodulating at least a part of the transmitted signal sequence (SSF) with a starting point from the assigned phase positions (00, 01, 10, 11) in the two known symbol sequences (SF1, SF2);
- selecting that phase position of the separate phase position (00, 01, 10, 11) in the said other of the known symbol sequences (SF2) which when effecting said demodulation produces a smallest error signal; and - demodulating the transmitted signal sequence (SSF) with the aid of the selected phase position.
- subsequently assigning a symbol in the other of the known symbol sequences (SF2) to each of the separate phase positions (00, 01, 10, 11);
- demodulating at least a part of the transmitted signal sequence (SSF) with a starting point from the assigned phase positions (00, 01, 10, 11) in the two known symbol sequences (SF1, SF2);
- selecting that phase position of the separate phase position (00, 01, 10, 11) in the said other of the known symbol sequences (SF2) which when effecting said demodulation produces a smallest error signal; and - demodulating the transmitted signal sequence (SSF) with the aid of the selected phase position.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| SE9102612A SE469052B (en) | 1991-09-10 | 1991-09-10 | PROCEDURE TO PICTURE A CHANNEL ESTIMATE FOR A TIME-ALIVE RADIO CHANNEL |
| SE9102612-0 | 1991-09-10 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CA2076061A1 true CA2076061A1 (en) | 1993-03-11 |
Family
ID=20383679
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CA002076061A Abandoned CA2076061A1 (en) | 1991-09-10 | 1992-08-13 | Method of forming a channel estimate for a time-varying radio channel |
Country Status (4)
| Country | Link |
|---|---|
| US (1) | US5465276A (en) |
| JP (1) | JPH05218987A (en) |
| CA (1) | CA2076061A1 (en) |
| SE (1) | SE469052B (en) |
Families Citing this family (33)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| DE4311604C1 (en) * | 1993-04-08 | 1994-07-28 | Ant Nachrichtentech | Reliability-controlled data detection in receivers for TDMA mobile radio systems |
| JP3893151B2 (en) * | 1994-02-25 | 2007-03-14 | コーニンクレッカ フィリップス エレクトロニクス エヌ ヴィ | Multiple access digital transmission apparatus and radio base station and receiver for use in such apparatus |
| FI105514B (en) * | 1994-09-12 | 2000-08-31 | Nokia Mobile Phones Ltd | Reception procedure and recipients |
| FR2724513A1 (en) * | 1994-09-13 | 1996-03-15 | Philips Electronique Lab | SYNCHRONIZABLE DIGITAL TRANSMISSION SYSTEM ON ITS INITIALIZATION SEQUENCES |
| SE503522C2 (en) * | 1994-10-31 | 1996-07-01 | Ericsson Telefon Ab L M | Channel estimation method and apparatus |
| US5592514A (en) * | 1995-03-08 | 1997-01-07 | Lucent Technologies Inc. | Method of performing signal reconstruction at the receiving end of a communications system, such as for GSM |
| FR2742619B1 (en) * | 1995-12-15 | 1998-02-06 | Thomson Csf | MULTI-SENSOR EQUALIZATION METHOD FOR MULTI-SENSOR RECEPTION IN THE PRESENCE OF INTERFERENCE AND MULTI-TRAVEL PATHWAYS, AND RECEIVER FOR IMPLEMENTING SAME |
| US5790598A (en) * | 1996-03-01 | 1998-08-04 | Her Majesty The Queen In Right Of Canada | Block decision feedback equalizer |
| US5881073A (en) * | 1996-09-20 | 1999-03-09 | Ericsson Inc. | Convolutional decoding with the ending state decided by CRC bits placed inside multiple coding bursts |
| US6026130A (en) * | 1997-03-04 | 2000-02-15 | Advanced Micro Devices, Inc. | System and method for estimating a set of parameters for a transmission channel in a communication system |
| JPH1141115A (en) * | 1997-07-23 | 1999-02-12 | Mitsubishi Electric Corp | Digital data demodulating device |
| IT1295863B1 (en) * | 1997-10-22 | 1999-05-28 | Telital Spa | METHOD AND APPARATUS FOR TRANSMISSION AND RECEPTION OF DIGITAL SIGNALS AND ESTIMATION OF COMMUNICATION CHANNELS |
| US6226336B1 (en) | 1998-02-20 | 2001-05-01 | Telefonaktiebolaget Lm Ericsson (Publ) | Method and apparatus for detecting a frequency synchronization signal |
| FR2782212B1 (en) * | 1998-08-04 | 2004-07-23 | Nortel Matra Cellular | DYNAMIC BORING OF A TRANSMISSION CHANNEL |
| US6658050B1 (en) | 1998-09-11 | 2003-12-02 | Ericsson Inc. | Channel estimates in a CDMA system using power control bits |
| US6535554B1 (en) | 1998-11-17 | 2003-03-18 | Harris Corporation | PCS signal separation in a one dimensional channel |
| US6542560B1 (en) * | 1999-04-23 | 2003-04-01 | Lucent Technologies Inc. | Method of channel estimation and compensation based thereon |
| DE19935480A1 (en) * | 1999-07-28 | 2001-02-22 | Infineon Technologies Ag | Method for estimating the channel impulse responses of a mobile radio channel |
| US6775521B1 (en) * | 1999-08-09 | 2004-08-10 | Broadcom Corporation | Bad frame indicator for radio telephone receivers |
| US6430528B1 (en) * | 1999-08-20 | 2002-08-06 | Siemens Corporate Research, Inc. | Method and apparatus for demixing of degenerate mixtures |
| DE60129381D1 (en) * | 2000-02-22 | 2007-08-30 | Koninkl Philips Electronics Nv | MULTI-RENDER RECEIVER WITH CHANNEL ESTIMATE |
| DE60023280D1 (en) * | 2000-06-23 | 2005-11-24 | St Microelectronics Nv | Method and device for estimating the impulse response of a transmission channel, in particular for a cellular mobile telephone |
| US6970520B1 (en) | 2000-11-13 | 2005-11-29 | Telefonaktiebolaget Lm Ericsson (Publ) | Methods and systems for accumulating metrics generated by a sequence estimation algorithm |
| US6947475B2 (en) * | 2001-01-26 | 2005-09-20 | Qualcomm Inc. | Method and apparatus for estimating channel characteristics using pilot and non-pilot data |
| JP4164364B2 (en) * | 2001-02-22 | 2008-10-15 | コーニンクレッカ フィリップス エレクトロニクス エヌ ヴィ | Multi-carrier transmission system with channel response estimation with reduced complexity |
| US6940914B1 (en) | 2001-06-11 | 2005-09-06 | Cingular Wireless Ii, Llc | Turbo channel estimation for OFDM systems |
| KR100393192B1 (en) * | 2001-07-11 | 2003-07-31 | 삼성전자주식회사 | Signal receiving apparatus and method adapted for time-variant channel |
| US20040005010A1 (en) * | 2002-07-05 | 2004-01-08 | National University Of Singapore | Channel estimator and equalizer for OFDM systems |
| US6999539B2 (en) * | 2003-05-06 | 2006-02-14 | Nokia Corporation | Linear derivative equalizer, and associated method, for a radio communication system |
| KR100620914B1 (en) * | 2004-04-07 | 2006-09-13 | 삼성전자주식회사 | Apparatus and method for switching between amc mode and diversity mode in broadband wireless communication system |
| JP2009520429A (en) * | 2005-12-20 | 2009-05-21 | コーニンクレッカ フィリップス エレクトロニクス エヌ ヴィ | Method and apparatus for receiving signals in an OFDM system |
| US8750089B2 (en) * | 2010-01-05 | 2014-06-10 | Broadcom Corporation | Method and system for iterative discrete fourier transform (DFT) based channel estimation using minimum mean square error (MMSE) techniques |
| TWI583229B (en) * | 2015-03-26 | 2017-05-11 | 晨星半導體股份有限公司 | Time-varying channel discriminating device and method thereof |
Family Cites Families (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US4516241A (en) * | 1983-07-11 | 1985-05-07 | At&T Bell Laboratories | Bit compression coding with embedded signaling |
| CA1238368A (en) * | 1983-10-14 | 1988-06-21 | Takayuki Ozaki | Digital radio receiving apparatus |
| US4564952A (en) * | 1983-12-08 | 1986-01-14 | At&T Bell Laboratories | Compensation of filter symbol interference by adaptive estimation of received symbol sequences |
| US4631734A (en) * | 1985-03-21 | 1986-12-23 | At&T Bell Laboratories | Cross-polarization canceler/equalizer |
| SE464902B (en) * | 1989-10-24 | 1991-06-24 | Ericsson Telefon Ab L M | PROCEDURE TO ADAPT A VITERBIAL ALGORITHM TO A CHANNEL WITH VARIOUS TRANSFER PROPERTIES AND A DEVICE IMPLEMENTING PROCEDURE |
| SE465245B (en) * | 1989-12-22 | 1991-08-12 | Ericsson Telefon Ab L M | PROCEDURE TO PERFORM CHANNEL ESTIMATES FOR A FAITHFUL CHANNEL FOR TRANSFER OF SYMBOL SEQUENCES |
| US5297169A (en) * | 1991-06-28 | 1994-03-22 | Telefonaktiebolaget L M Ericsson | Equalizer training in a radiotelephone system |
-
1991
- 1991-09-10 SE SE9102612A patent/SE469052B/en unknown
-
1992
- 1992-08-13 CA CA002076061A patent/CA2076061A1/en not_active Abandoned
- 1992-09-09 JP JP4240278A patent/JPH05218987A/en active Pending
- 1992-09-09 US US07/942,270 patent/US5465276A/en not_active Expired - Lifetime
Also Published As
| Publication number | Publication date |
|---|---|
| SE9102612L (en) | 1993-03-11 |
| SE469052B (en) | 1993-05-03 |
| JPH05218987A (en) | 1993-08-27 |
| US5465276A (en) | 1995-11-07 |
| SE9102612D0 (en) | 1991-09-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US5465276A (en) | Method of forming a channel estimate for a time-varying radio channel | |
| EP0443997B1 (en) | A method of reducing the influence of fading of a Viterbi-receiver having at least two antennas | |
| JP2715662B2 (en) | Method and apparatus for diversity reception of time division signals | |
| US6246732B1 (en) | Demodulator including adaptive equalizer and demodulating method in digital communications | |
| CA2204135C (en) | Method and apparatus for channel estimation | |
| US5204878A (en) | Method of effecting channel estimation for a fading channel when transmitting symbol sequences | |
| US5533067A (en) | Method and device for estimating transmitted signals in a receiver in digital signal transmission operations | |
| EP0998805B1 (en) | Determination of the length of a channel impulse response | |
| CN100403745C (en) | Method and apparatus for bidirectional demodulation of digitally modulated signals | |
| US5353307A (en) | Automatic simulcast alignment | |
| US5930296A (en) | Low-complexity bidirectional equalizer | |
| EP0500326A2 (en) | Digital mobile communication terminal equipment and receiving method therefor | |
| EP0449327A2 (en) | Noise-immune space diversity receiver | |
| US6411649B1 (en) | Adaptive channel tracking using pilot sequences | |
| WO2001054366A2 (en) | Parallel decision feedback equalizer with adaptive thresholding based on noise estimates | |
| EP1072133B1 (en) | Self-optimizing channel equalization and detection | |
| US6853681B1 (en) | Using different channel models for different training sequences | |
| EP1259040A2 (en) | Equalisation for QAM signals | |
| KR101059878B1 (en) | Timing offset compensation method of interference cancellation receiver | |
| JPH0865184A (en) | Detecting method for interference wave | |
| JP3537203B2 (en) | Antenna diversity receiver | |
| JPH02189034A (en) | Equalizer | |
| Wales | Co-channel interference suppression for TDMA mobile radio systems |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| EEER | Examination request | ||
| FZDE | Discontinued |