WO2009074603A1

WO2009074603A1 - Method for ascertaining the clock rate of a data signal, which is received from a node of a communication system, active star coupler for executing the method, and communication system having an active star coupler of this type

Info

Publication number: WO2009074603A1
Application number: PCT/EP2008/067218
Authority: WO
Inventors: Juergen Schirmer; Thorsten Huck
Original assignee: Robert Bosch Gmbh
Priority date: 2007-12-11
Filing date: 2008-12-10
Publication date: 2009-06-18
Also published as: DE102007059554A1

Abstract

The invention relates to a method for ascertaining the clock rate of a data signal (40), which is transmitted in a communication system (200), which comprises a data bus (202) and multiple subscribers (100; 204) connected thereto, from a transmitting subscriber via the data bus (202) to a receiving subscriber. The invention additionally relates to a communication system (200) and an active star coupler (100) of the communication system (200) for implementing a method according to the invention. In order to allow a data transmission in the context of a predefined protocol specification even via complex network structures, in particular in regard to the occurring asymmetrical delays, and in order in particular to provide a possibility for ascertaining the actual clock rate of the transmitted data signal (40) at the receiver in a simple and efficient way, it is proposed that, to overlay the data signal (40) with additional information in regard to the clock rate, the data signal (40) is transformed at the transmitting subscriber, the transformed data signal is transmitted via the data bus (202) and transformed back at the receiving subscriber, wherein said additional information in regard to the clock rate is reclaimed in the context of the back transformation. The receiver is preferably part of the active star coupler (100) according to the invention.

Description

description

Title:

Method for determining the clock rate of a data signal received from a subscriber of a communication system, active star coupler for carrying out the method and communication system with such an active star coupler

State of the art

The present invention relates to a method for determining the clock rate of a data signal which is transmitted in a communication system comprising a data bus and a plurality of connected thereto subscribers from a transmitting subscriber via the data bus to a receiving subscriber. The invention also relates to a communication system for carrying out the method and to an active star coupler as a receiving subscriber of the communication system.

Star couplers are known, for example, for FlexRay communication systems from the FlexRay specification v2.1. The structure and operation belong to the specified physical layer (so-called physical layer) of the FlexRay communication system. Active star couplers are important in communication networks in which the communication link (for example, a data bus) splits up, that is to say has a star topology, and a data signal is to be split over several branches of the data bus. In addition, active star couplers are important when it comes to the transmission of data signals over complex network topologies and longer distances, since they can also amplify the signal in addition to or as an alternative to dividing the data signal over several branches.

Electrical and electronic components (ie the corresponding ICs) for an active star coupler (a so-called Active Star) for use in a FlexRay Communication systems are offered by Philips Semiconductors (now NXP). In a star coupler, FlexRay communication controllers of type "SJA 2510" according to specification v2.1 and an ARM9 microcontroller can be integrated. On a known active star coupler several ports are provided to which a plurality of branches of the communication link are connected. The ports can be configured either as input for incoming data signals and / or as output for outgoing data signals. The star coupler has a bus driver, for example of the "NXP TJA 1080" type, for amplifying an outgoing data signal at each connection A digital data signal received via one of the connections is forwarded digitally to a central processing logic of the star coupler, which is a computing device, eg In the form of a Field Programmable Gate Array (FPGA), a microcontroller or a digital signal processor (DSP), corresponding complete active star couplers are offered, for example, by TZM (Transfer Center Microelectronics, Göppingen) with semiconductors from NXP (formerly Philips Semiconductors) ,

The central processing logic allows further processing of the data signal (e.g., monitoring and / or router function) in the active star coupler. If a communication controller is provided in the star coupler, the latter can assume a monitoring and / or router function depending on the content of the incoming data signal decoded in the central processing logic. A likewise optionally provided bus guardian can monitor and / or control the access of the star coupler to the data bus. The received in the central processing logic received digital data signal is applied via bus driver (so-called bus driver), which convert the bit sequence of the digital signal back into one or more analog signals and amplify, to the communication link.

Prior art TZM active star couplers include "NXP TJA 1080" bus drivers similar to those of FlexRay transceiver units, so the known star coupler is nothing more than linking multiple transceivers to a hub Subscriber (or node) of a communication network via a branch of the communication connection incoming data to all other participants of the communication system and at the same time amplifies the signal to be forwarded. In the development of the physical layer for FlexRay, the "NXP TJA 1080" bus driver represents the state of the art. In addition, the use of integrated energy and data transmission is currently being sought (FlexRay-PLC, Power Line Communications) In accordance with Chapter 2.1 of the "Electrical Physical Layer Specification" FlexRay specification, Version 2.1, any physical layer can be used as a FlexRay communication channel as long as the specified basic requirements.

In the FlexRay PLC, the data is modulated onto a high-frequency carrier by means of DBPSK (differential binary phase shift keying) modulation and transmitted via the energy on-board network. The conditioned electrical system allows a simple design of the transmission system, since no expensive equalizers are required. In this respect, the digital signal processing is limited, which is why a fast data transfer of 10 Mbit / sec with a propagation delay (Propagation Delay) of a maximum of 2.5 microseconds (as required in the FlexRay specification) can be achieved.

FlexRay is a fast, deterministic and fault-tolerant bus system, especially for use in motor vehicles. The FlexRay protocol operates on the principle of Time Division Multiple Access (TDMA), whereby the subscribers or the messages to be transmitted are assigned fixed time slots in which they have exclusive access to the communication connection. The time slots are repeated in a fixed cycle, so that the time at which a message is transmitted over the bus, can be accurately predicted and the bus access is deterministic.

In order to get the most out of bandwidth for the transmission of messages on the bus system, FlexRay divides the cycle into a static and a dynamic part or into a static and a dynamic segment. The fixed time slots are located in the static part at the beginning of a bus cycle. In the dynamic part, the time slots are specified dynamically. In it is now the exclusive bus access only for a short time, for the

Duration of at least one so-called minislot enabled. Only if bus access actually occurs within a minislot does the time slot become the time required for access extended. Thus, bandwidth is only consumed when it is actually needed. There are two channels in FlexRay. FlexRay communicates via one or two physically separate lines per channel at a maximum data rate of 10 Mbit / sec. Of course, FlexRay can also be operated at lower data rates. The lines of the two channels correspond to the physical layer, in particular the so-called OSI (Open System Architecture) layer model. The two channels are mainly used for the redundant and thus fault-tolerant transmission of messages, with the same data being transmitted over both channels. Alternatively, however, different messages can be transmitted via the two channels, which could then double the data rate in the FlexRay communication system. It is also conceivable that the signal transmitted via the connecting lines results from the difference of signals transmitted via the two lines. Finally, only one channel can be used for data transmission, the other channel is then unused and free. The physical layer is designed such that it enables electrical or optical transmission of the signal or signals via the line (s) or a transmission in another way.

In order to realize synchronous functions and to optimize the bandwidth by small distances between two messages, the subscribers in the communication network need a common time base, the so-called global time. Synchronization messages are transmitted in the static part of the cycle for the synchronization of local clocks of the subscribers, whereby the local time of a subscriber is corrected with the aid of a special algorithm according to the FlexRay specification in such a way that all local clocks synchronously run to one global clock.

In the transmission of data or messages via such a bus system pulses are distorted because high-to-low and low-to-high edges are delayed in different ways on the transmission path. If the transmitted pulse in the receiver with the sampling rate present there (the so-called sampling rate) is scanned several times (for example n times per bit, in particular 8 times per bit), the position of the sampling point, ie the selection of precisely one of these n, decides Samples whether the date is sampled correctly or incorrectly. This is particularly difficult when the sampling instant refers to an edge of the signal and, relative to that, over many periods of the sample clock, several binary data values (bits) of the Transmitter evaluates. In addition to a pulse distortion (according to specification: asymmetrical delay), the clock frequency deviation between transmitter and receiver also has an effect here. In this case, the signal to be sampled can be pretreated, for example, to filter out short-term disturbances. Such a filter can be realized by evaluating a plurality of scanning signals in the chronological order with a majority decision (so-called voting). Especially in the specification of the FlexRay protocol, where n κ-1

Network node 2 V z can give different transmission paths (every conceivable path has

2 transmitter-receiver combinations), it has been found that the occurring asymmetric delays (ie rising and falling edges propagate with different delays in the network) on the different transmission paths can lead to timing problems. The delay between the rising and falling edge of a signal is also referred to as pulse distortion.

Asymmetric delays can have both systematic and stochastic causes. With the FlexRay protocol, systematic delays affect only the rising edges because they are synchronized to the falling edges. Stochastic delays affect both rising and falling edges and are caused by noise or EMC jitter. Each component of the network must be examined separately with respect to the asymmetrical delay caused by it, whereby in the prior art the delays occurring in the individual components must be added together in order to obtain the maximum asymmetric delay of the selected transmission path (so-called worst-case analysis). ,

In the realization of FlexRay data transmission systems, especially in complex systems comprising several active star couplers and passive networks, it has also been found that the theoretically determined worst-case asymmetric delay times are so great that they exceed a predetermined by the FlexRay protocol time budget. In accordance with the FlexRay protocol, a sampling counter is synchronized with the falling BSS (Byte Start Sequence) edge, ie reset to 1. At a count of 5 is scanned. With an 8-fold oversampling (so-called oversampling), as it is currently provided in FlexRay, between the sampling time (5th sample) and the 8th sample still remains 3 Sample clocks corresponding to a Kommunikationscontro Her clock of 80 MHz each 12.5 ns, in total so a time budget of 37.5 ns. This time budget actually serves to compensate for asymmetrical delays due to the difference between falling and rising edge steepness and the deviations of the local clocks between transmitter and receiver. If, however, as is the case with complex network topologies, the asymmetric one

Delay exceeds the intended time budget, this leads to a wrong value being determined for a sample at the 5th sample clock (counter reading of the sample counter at 5), because the bit that should have been sampled actually becomes one due to the asymmetrical delay earlier date and no longer due to the early flank change. A similar treatment applies to an asymmetric delay after late. Then a time budget of 4 sampling clocks corresponding to 50 ns is available. The consequence of exceeding the time budget to early or late are decoding errors, so wrong data is received.

Although these decoding errors can be detected by suitable error detection algorithms, so that a retransmission of the bit or the entire data frame can be caused. As an error detection algorithm, for example, a so-called. Parity bit or a so-called. Cyclic Redundancy Check (CRC) can be used. The disadvantage of frequent response of the error detection algorithm, however, lies in the associated poorer availability of the data transmission system, since the faulty data, for example, be retransmitted or discarded.

In summary, it can be said that the FlexRay protocol makes specifications that the physical layer can not maintain - at least for complex network topologies. It has been shown that the occurring asymmetric delays prohibit the construction of complex network structures, as these no longer meet the timing requirements of the FlexRay specification. For this reason, in addition to the clock control already implemented in FlexRay (sending a binary sequence for synchronization by correlation measurement), additional measures must be taken to improve the clock control and to reduce the asymmetrical delays. This may require that a receiving subscriber have information about the exact clock rate of a received data signal used to send the data signal, which is in FlexRay is currently not the case.

Disclosure of the invention

The present invention is therefore based on the object of enabling data transmission within the scope of a predetermined protocol specification, in particular with regard to the occurring asymmetrical delays, even over complex network structures. In particular, with the present invention, a possibility is to be created to determine the actual clock rate of the transmitted data signal at the receiver in a simple and efficient manner.

To solve this problem is proposed starting from the method of the type mentioned above, that for superimposing the data signal with additional information regarding the clock rate transformed the data signal at the sending party, the transformed data signal is transmitted via the data bus and transformed back at the receiving party, wherein in In the case of the inverse transformation, the additional information relating to the clock rate is recovered.

According to the invention, it is therefore proposed to superimpose the data signal in the transmitter with additional information regarding the clock rate of the signal transmission. The data signal is then transmitted along with the overlaid information over the data bus. At the receiver, the data signal is then transformed back to recover the clock rate information. This information is used to determine the clock rate of the transmitted data signal. The determined clock rate can be used to correct the sampling of the received signal in the receiver, in particular in the context of clock control in the receiver.

The information relating to the clock rate can be superimposed on the data signal, for example in the form of at least one setpoint edge change, at a defined position within the data bits of the data signal. Such a desired edge change can be, for example, by means of a

Generate baseband encoding at the transmitter. The receiver must then be provided with appropriate means for baseband decoding. As baseband coding is preferably a Manchester coding, a Biphase-Mark coding or a Miller coding used. Theoretically, coding by means of Barker sequences, by means of a bipolar code, by means of an AMI code or by means of an HDBn code would also be conceivable. Alternatively, it is conceivable that a reference edge change at a defined position within the data bits of the data signal is generated as information about the clock rate of the transmitted data signal by a scrambler at the sending participant. At the receiver then a corresponding descrambler must be provided. When using a scrambler or descrambler, however, the resulting delays must be taken into account.

The receiver is preferably part of an active star coupler of a star topology communications system in which the incoming data signal is decoupled from the transmitted data signal for asymmetric delay and the asymmetrical delay of the incoming data signal is reduced such that the data signal leaving the star coupler has less asymmetric delay has as the incoming data signal. To reduce the asymmetric delay, the information recovered in the receiver of the star coupler with respect to the clock rate of the signal transmission is used. For this purpose, an asynchronous FIFO memory element is used in the star coupler at the transition between the receiver and transmitter, which is fed by two (non-synchronous) clocks, namely on the one hand to the extracted using the extracted information with respect to the clock rate of the communication system, with the

Data signal from the receiver is received, and the independent local clock of the transmitter, with which the data signal from the star coupler is sent again. The synchronized clock signal is thus used in the receiver (or at the input) of the star coupler, while the local signal in the transmitter (or at the output) of the star coupler is used. The clock signal in the receiver is synchronized with the aid of the transmitted information relating to the clock rate of the transmission of the data signal present at the input, so that the asymmetrical delay is reduced and the correct data bit is always sampled at the sampling instant.

In the transmission of a data signal from a transmitting subscriber to a receiving subscriber via a complex network topology with at least one active star coupler, the active star coupler is thus extended to a significant reduction, if possible even elimination of the asymmetric delay which the data signal was impressed on the transmission line from the transmitter to the receiver of Sternkopp coupler, can be made. This only applies to the distance between the sending user of the communication system and the receiver of the active star point. Starting from a significantly reduced asymmetric delay at the output of the active Sternkopp coupler, the data signal on the last leg of the transmission path between the active star coupler and a receiving subscriber while a new share of the asymmetric delay can be impressed. However, when considered over the entire transmission path, the use of the described active star couplers can significantly reduce the asymmetrical delay of the data signal arriving at the receiver. In a worst case consideration, the classification of a

Communication system is required, the decoupling can be achieved by the present invention at each described in the described manner star coupler with respect to the calculation of the maximum asymmetric delay of the entire network.

This makes it possible to achieve a simple, very efficient and cost-effective reduction of the asymmetrical delay on the transmission path. For this purpose, it is already sufficient, if not all, but only some star couplers are formed in a communication network in the proposed manner. Preferably, however, all star couplers are formed in a communication network in the manner proposed, unless this conflicts with the maximum allowable propagation delay. Overall, this can increase the acceptance of FlexRay, since FlexRay communication systems were previously limited to relatively simple network topologies, which is now no longer the case. Very complex topologies can also be realized with the described star coupler without the asymmetrical delay on the transmission paths reaching values beyond the tolerance window permitted according to the FlexRay specification. The present invention forms for the

Requirement by the star coupler, the additional information regarding the clock rate of the signal transmission is provided, the local clock of the star coupler can be synchronized to this clock rate as quickly as possible.

The dependent claims have advantageous embodiments of the invention the subject. Your

Advantages and further embodiments of the invention can be found in detail in the following description of the figures. Brief description of the drawings

FIG. 1 shows an active star coupler according to the invention; FIG. 2 shows a central processing logic of the active star coupler on FIG. 1;

Figure 3 shows waveforms of a transmitted data signal and a received data signal for

Explanation of the asymmetric delay; Figure 4 is a circuit diagram of a latch for use in a central processing logic of Figure 2 of the active star coupler of Figure 1; Figure 5 shows waveforms of the latch of Figure 4;

FIG. 6 shows a communication system with a complex network topology; FIG. 7 shows an example for explaining a Manchester coding; and FIG. 8 shows an example for explaining a BMC coding.

Embodiment (s) of the invention

In FIG. 2, an active star coupler according to the invention for a FlexRay communication system is designated in its entirety by the reference numeral 100. Star coupler 100 is also referred to as Active Star in the FlexRay protocol specification. The star coupler 100 shown in Figure 2 may be connected to up to four branches of a communication link (e.g., a data bus). The communication link may comprise one or more electrical lines, one or more optical lines or transmission channels of a different type, for example radio links.

The star coupler 100 shown in FIG. 2 is connected to a line of a first branch via a first terminal 101 and a second terminal 102, respectively

Communication connection (not shown) connected. Via further connections 103, 104; 105, 106 and 107, 108, the active star coupler 100 can be connected in a corresponding manner to two lines of further branches of the communication link. The star coupler 100 could thus in a FlexRay communication system with a Communication connection with star topology and up to four branches (or arms) are used.

The active star coupler 100 has its own bus driver (so-called bus driver, BD) 109, 110, 111, 112 for each of the four branches. The bus drivers 109-112 correspond in function to a transceiver unit (so-called transceiver). In the bus drivers 109-112, the difference between the values transmitted via the two lines of the branches of the communication connection and at the terminals 101, 102; 103, 104; 105, 106 or 107, 108 applied signals and generated (for sending) for each branch a difference signal. Furthermore, the bus drivers 109-112 convert an analog data signal or differential signal received via one of the branches into a digital signal for further processing in the star coupler 100. The bit sequence of the received digital signal is forwarded by the bus drivers 109-112 to a central processing logic 113 comprising a computing device for processing the incoming digital data signal, for example as an FPGA (Field Programmable Gate Array), a microcontroller DSP (Digital Signal Processor), a CPLD (Complex Programmable Logic Device) or simply designed as a discrete logic circuit.

Optionally and therefore shown in dashed lines, the star coupler 100 comprises a communication controller (CC) 114 which further processes and / or evaluates the data received via one of the bus drivers 109 - 112 and preprocessed in the central processing logic 113 to realize a monitoring and / or router functionality. To realize the router function, it would be conceivable to analyze the content of the incoming and preprocessed data in order to determine for which of the branches connected to the star coupler 100 or for which of the subscribers the data received via one of the channels are intended. This information could be taken from a header segment (so-called header) of the incoming data packets. However, such a realization of a router functionality is not (yet) provided according to the current FlexRay specification. If a communication controller 114 is provided in the star coupler 100, a receive (RxD) port 115, a transmit port (Transmit, TxD) 116, and a transmit enable (TxEN) port 117 are provided at the star coupler 100 which are routed to the communication controller 114. Also shown as optional and therefore dashed, the Sternkopp ler 100 includes a monitoring unit in the form of a bus guardian 118, which monitors the access of the communication controller 114 to the communication link and / or controls. The bus guardian 118 can be controlled via a bus guardian trigger connection (Bus Guardian Enable BGE) 119 and a receive trigger connection (receive enable, RxEN) 120 from outside of the star coupler 100.

Finally, the active star coupler 100 also includes a supply voltage source 121 to which an operating voltage is applied via a terminal (Vcc) 125 and ground via a further terminal 123 (GND). In addition, a battery voltage interruption port (inhibit, INH) 124 and a terminal 30/15 port 122 (V _Ba t) may be provided, at which a vehicle electrical system voltage, eg depending on the design, a battery voltage of 6V, 12V or 24V , is present.

The star coupler 100 is part of a communication system, such as is designated in FIG. 6 in its entirety by the reference numeral 200, for example. The communication system 200 includes a physical data bus 202 and a plurality of subscribers 204 connected thereto. In addition, the communication system 200 has two star couplers 100. A star coupler 100 receives and transmits data signals similar to a subscriber 204. For this reason, the term "subscriber" in the claims also includes a star coupler In order to realize synchronous functions in the communication system and to optimize the bandwidth by small distances between two messages, the subscribers 204 in the communication network 200 need a common time base , the so-called global time. For the synchronization of local clocks the participant 204 becomes

Synchronization messages transmitted in the static part of the cycle, using a special algorithm according to the FlexRay specification, the local time of a subscriber 204 is corrected so that all local clocks run synchronously to a global clock. The same applies to star coupler 100, whose local clock signal from a local clock is also synchronized to the global clock of the communication system.

In the implementation of FlexRay data transmission systems, especially in complex Network topologies with multiple star couplers and passive networks have shown that the asymmetric delay times that occur in the system are so great that they exceed a time budget imposed by the FlexRay protocol. According to FlexRay protocol v.2.1, a data bit of the received signal is 8-times oversampled. Approximately in the middle of the data bit, the data bit is sampled at a sampling instant. This is realized by synchronizing a sample counter with falling BSS (Byte Start Sequence) edge, ie setting it back to 1 and starting it. At a count of 5 is scanned. With an 8-fold oversampling (so-called oversampling), as currently provided in FlexRay, between the sampling time (5th sample) and the 8th sample, there are still 3 sampling cycles which, at a communication controller clock of 80 MHz, remain at 12.5 ns, in total therefore correspond to a time budget of 37.5 ns. This time budget actually serves to compensate for asymmetrical delays due to the difference between falling and rising edge steepness and the deviations of the local clocks or clock signals / quartz tolerances between transmitter and receiver. However, if, as can be the case with complex network topologies, the asymmetric delay exceeds the intended time budget, this will result in an incorrect value being determined for a sample at the 5th sample clock (count of the sample counter at 5) because that bit, that should have been sampled, due to the asymmetric delay at an earlier time and applied by the early edge change is no longer present. An analogous treatment applies to an asymmetric delay after late. Then a time budget of 4 sampling clocks corresponding to 50 ns is available. The consequence of exceeding the time budget to early or late are decoding errors, so wrong data is received.

To explain the concept of asymmetric delay, reference is made to FIG. There, a signal (TxD) designated by a transmitting subscriber of the communication system with the reference numeral 10 and the corresponding signal (RxD) as received by a receiving subscriber are shown. The rising edge delay is designated by the reference numeral 16 and the falling edge delay by the reference numeral 18. Depending on the difference of the delays for rising edges 16 and falling edges 18 results in an asymmetric delay 20 of the transmitted data signal 10 on the considered transmission path. In the transmission of data or messages via a bus system with such delays 16, 18, pulses are distorted because high-to-low- or Low-to-high edges on the transmission are delayed differently. This ultimately leads to the asymmetric delay 20.

Each component on a particular transmission path through the communication system contributes to the total asymmetric delay. Each component must be considered separately and the asymmetrical delays occurring in the components (so-called Slewrates) must be added together, which then results in a worst-case scenario for the asymmetrical delay of the communication channel. The FlexRay system must take appropriate precautions to remain within preset limits for the asymmetric delay. The aim is to be able to meet the requirements for a secure and reliable data transmission by suitable reduction or compensation of the asymmetric delay in communication systems 200 with complex topologies. Important for the reduction or compensation of the asymmetrical delay is the most accurate possible determination of the clock rate of the received signals, for example by measuring the bit or symbol duration of the signals.

In order to be able to provide the clock rate of a signal transmission on the receiver side, it is proposed according to the invention that the data signal in the transmitting subscriber is superimposed with additional information relating to the clock rate of the data signal. This additional information may, for example, in the form of at least one of the data signal

Target edge change to be superimposed on a defined position within the data bits of the data signal. Such a desired edge change can be generated, for example, by means of a baseband coding at the transmitter. The data signal is then transmitted along with the overlaid information over the data bus. At the receiver, the data signal is then transformed back again (for example by means of a baseband decoding corresponding to the baseband coding) in order to recover the transmitted information regarding the clock rate. This information is used to determine the clock rate of the transmitted data signal. The determined clock rate can be used to correct the sampling of the received signal in the receiver, in particular in the context of clock control in the receiver. As baseband coding is preferably a Manchester

Coding (MC), a Biphase-Mark coding (BMC) or a Miller coding used. In Fig. 7, a Manchester encoded signal is designated by the reference numeral 11, and the corresponding decoded value of the data bits of the signal 11 is indicated below and designated by the reference numeral 12. In Manchester coding, the data bits are encoded into the edges of the signal. The information-bearing edge change always occurs at the middle of the bit. A falling edge means, for example, a logical 1, a rising edge means a logical 0 (or vice versa depending on the definition). Therefore, there is at least one edge per data bit. For the transmission of the Manchester coded signal twice the bit rate of the actual useful signal is required. The code is by definition not invertible. In addition, there is the differential Manchester code, in which, in contrast to the Manchester coding at the beginning of the bit only at zeros, an edge change takes place. As a result, the fixed assignment between the direction of the edge change and logical signal state is lost.

Miller coding is also known as delay coding. By means of this coding, each logical 1 in the serial data stream is coded with a clock edge in the middle of the bit, while a logical 0 either has no edge (if it is followed by a 1) or it is coded with an edge to the beginning of the bit. As a result, edge changes occur in the multiples 1.0, 1.5, and 2.0 of the bit times.

In FIG. 8, a clock signal for the coding with the reference symbol 13, the data signal with the reference symbol 14 to be coded and the signal coded with the BMC with the reference symbol 15 are designated. In the BM encoding, two zero crossings of the BMC signal 15 correspond to logic 1 per two clock cycles, and one zero crossing of the BMC signal per two clock cycles equals to logic 0. In other words, each bit begins with an edge, in the case of a logical 1 occurs in the middle of the bit another edge, while for 0 the edge is missing. The bit rate of the data signal 14 corresponds to half the frequency of the clock signal 13. The code is invertible since there is no exact assignment of the edges.

Otherwise, there are a number of other codes, such as the kBnT code (k binary, n ternary), each of which summarize several bits and ensure that enough edge changes occur within these packets and allow synchronization of the sampling clock at the receiver. In this way, the bandwidth requirement can be kept very low (ie an increase of about 25% is sufficient when choosing a suitable code, the to transmit additional information regarding the clock rate). Decisive for the quality of a code is always the effectiveness or the redundancy measure. Of course, a kBnT code is more recommendable, but it has an effect on the propagation delay (propagation delay). Of course, there are a number of other codes that are particularly well suited for synchronization, such as Barker episodes, but their use is associated with a much larger bandwidth needs. However, if enough bandwidth is available because, for example, the data signal only partially utilizes the bandwidth available in the communication system, this is not an insignificant limitation. The use of a Barker code for coarse synchronization of the receiver is particularly well suited Start of transmission. Since here the data due to the shortening of the data frame (so-called.

Truncation) can be discarded anyway, this period can be used to configure the recipient. Truncation is the part of the message that does not actually contain any information. The point is that there is a physical signal on the channel and the system can "tune in" to get "useful" information. This part of the frame, i. usually the first 14 bits, may also be lost or adulterated arrive as they contain no information for the receiver.

Other codes, such as the Bipolar or Alternate Mark Inversion (AMI) code, also allow recovery of the clock (better here, however, the HDBn (High Density Bipolar) code). However, they also include a cancellation of the signal, which in this case can be equated with an amplitude modulation.

Alternatively, scramblers (self-synchronizing scramblers) could be used which scrambled the data bits according to a particular algorithm. These provide a transformation of the data signal to force more frequent zero crossings independent of the source signal. In contrast to the above codes, this does not require a higher bandwidth and no additional redundancy.

In order to ensure safe recovery of the bit duration, the coding method used must have a reference edge change at a defined position within the data bit. For this purpose, on the one hand, the Manchester encoding and on the other hand, the BMC encoding is particularly well suited. To transfer the data in a FlexRay PLC system, a combination of the DBPSK method and a corresponding baseband coding with reference edge change is used. In the case of a pure FlexRay communication system (not power supply lines form the physical level, but a conventional one

Data bus), the insertion of a baseband coding is sufficient. In both cases, this results in a transmission method with a bandwidth requirement that is increased by up to twice, which permits a defined recovery of the clock information from the setpoint edge changes at a defined position within the data bit in certain data bits, preferably in each data bit. This is necessary in order to be able to ensure the demand of a defined asymmetrical delay of the communication system at any time of the communication.

The clock rate of the transmitted data signal determined in the receiver, in particular in star point 100, can be used to implement a clock control. The concept of a clock control is described in detail, for example, on page 233 in Kroschel, Kristian: Datenübertragung, ISBN 3-540- 53746-5, Springer Verlag, 1991 in detail. Usually, the clock control takes place in two stages: in an initialization phase after the modem is switched on, a known binary sequence, the so-called preamble, is sent, which is used for coarse synchronization of the receiver by means of correlation measurement. This period must be within the truncation period and could be, for example, a cash reward. If the actual data is transmitted in the second stage, the fine adjustment takes place in the manner described in detail below, taking advantage of the information transmitted in addition to the data signal with respect to the clock rate of the transmission.

By using the superimposed baseband coding with reference edge change, a long synchronization process can be bypassed since the set edges of two successive bits of the data signal immediately provide information about the current clock rate of the data signal. This makes the recovery of the clock very easy, but includes a limited catch range on which the frequency can "lock", so there will always be a small, practically negligible error. This error is as follows. The receiver (eg, the star coupler 100) oversamples the incoming data signal according to the FlexRay specification at 8 times the frequency of the bit duration or symbol rate. The sampling frequency in the receiver must therefore be at least 80 MHz. Thus, every 12.5ns results in a new sample. To compensate for the offset between transmitter and receiver and deviations arising through the channel, the receiver adapts the data period by varying the clock frequency. Thus, the receiver does not always use 8 clock cycles to recover the incoming coded and sampled signal, but optionally 7, 8 (ideal) or 9 clock cycles.

By limiting the maximum phase correction to 1/8 of the entire data period, the theoretical error in the detection of a desired edge also results in 1/8 of the bit duration. When calculating the asymmetric delay of a FlexRay communication system, this error must be taken into account, resulting in a maximum delay of 12.5ns at a sampling frequency of 80 MHz. Although this error is caused by the proposed concept at the input of the active star coupler 100 (as is also the case in every FlexRax node, by the way), it is not passed on to the transmission path by the concept presented.

The receiver in which the clock rate of the transmitted data signal is recovered is preferably embodied as an active star coupler 100 of the communication system in which the incoming data signal is decoupled from the transmitted data signal for asymmetric delay and the asymmetrical delay of the incoming data signal is reduced the data signal leaving the star coupler has a smaller asymmetrical delay than the incoming data signal. In order to reduce the asymmetric delays 20 in a communication network, a

Extension of the star coupler 100 to the effect that a reduction or even elimination of the asymmetric delay, which was impressed on the data signal until reaching the star coupler 100, is made. This only relates to the distance between the transmitting subscriber and the star coupler 100, while on the last leg of the transmission path between active star coupler 100 and the receiving subscriber, a new proportion of asymmetrical delay can be impressed on the data signal. In particular, an extension of the star coupler 100 is proposed, which leads to a decoupling of the asymmetric delay on the transmission link. Thus, for a worst-case consideration of the asymmetric delay, only the path between a node (or subscriber) and an active star coupler according to the invention, as well as between two active star couplers according to the invention has to be investigated. This means that, unless the sum of all asymmetric delays per component occurring on these two sub-transmission links does not violate the maximum asymmetric delay allowed, no further analysis of the later overall configuration of the communications network is required. Of course, the propagation delay (Propagation Delay) must be adhered to.

The extension of the star coupler 100 proposed according to the invention is implemented in the central processing unit 113. An example of an extended processing unit 113 is shown in FIG. Instead of a simple branching matrix for controlling the incoming and outgoing signals within the central processing logic 113, as well as the monitoring by the optional bus Guardian 118, a complete node function is implemented in the inventive active star coupler 100. The star coupler 100 is thus a real participant of the communication system with its own local clock (Oscillator 132) whose local clock signal 134 is synchronized in a synchronization unit 136 to the clock signal 138 of the transmitting subscriber. The recovered clock rate of the incoming data signal according to the inventive method is supplied to the synchronization unit 136, so that it can synchronize the local clock signal 134 to the global clock signal 138.

One of the bus drivers 109-112 of the inventive star coupler 100 is configured as a receiver and at least one other bus driver 109-112 as a transmitter. In the embodiment of FIG. 1, the incoming signal is, for example, received by the bus driver 110 and sent out of the star coupler 100 again via the bus driver 112. The central processing unit 113 of the star coupler has means for decoupling the asymmetrical delay on the transmission link. These means physically separate the input (bus driver 110) from the output (bus driver 112) of the Star coupler, so that a to be transmitted via the star coupler data signal 40 can not be transmitted directly from the receiving bus driver 110 of the star coupler to the transmitting bus driver 112. In addition, the star coupler 100 has means for reducing the asymmetrical delay of the incoming data signal 40 so that the outgoing data signal 41 has a lower asymmetric delay than the incoming data signal 40.

According to the embodiment of the invention illustrated in FIG. 1, both the means for reducing the asymmetrical delay and the means for decoupling the asymmetrical delay comprise a memory element 130, which is realized, for example, as an asynchronous FIFO memory. A FIFO is a special form of shift register. The common feature is that the data appears in the same order at the output of the FIFO memory as it was input. The first written data word (First In) is read out first (First Out). In the case of a FIFO memory, in contrast to a shift register, this process can be carried out completely asynchronously, that is to say the read-out clock is independent of the input clock.

To implement the method according to the invention, the incoming data signal 40 is first decoded with the globally synchronized clock signal 138 in a decoding unit 140. The decoding of the received signal 40 on the one hand comprises a baseband decoding with which the baseband coded signal 40 can be decoded again. The clock rate of the data transmission can be recovered in the decoder 140 from the additional edge change introduced by way of baseband coding into the data signal at defined times. This determined in the receiver of the star coupler 100 clock rate of data transmission is used to synchronize the clock signal in the receiver. Furthermore, the

Decode the incoming signal 40 in the decoder 140, the signal 40 bits for each bit repeatedly, eg. 8-fold, oversampled. By a majority decision on the sampled signal values of a data bit sampled at different sampling times, the value of the data bit is determined. For example, if 8 samples are "8" and only 2 samples are "0", assuming that samples "0" are erroneous (due, for example, to the asymmetric delay), and if the Data bit actually has the value "1". The data bit is stored in the memory 130 with the value "1".

As already mentioned, the receiving part of Sternkopp coupler 100 corresponds to that of a conventional node, that is, the clock is synchronized to the receive sequence and thus to the clock of the transmitting node. Due to the asymmetric delay occurring in the signal 40, shifts may occur during the synchronization. Therefore, before the data in this form is put back on the channel, the asymmetrical delay is removed from the signal 40. For this purpose, the data is buffered in the asynchronous FIFO memory 130 and then recoded in a coding unit 142 with the local system clock 134 of the active star coupler 100, that is not with the synchronized clock signal 138 of the communication system. The output of the active star coupler then in turn corresponds to a transmitting node, which reads the data of the communication controller (CC), in this case from the buffer memory 130 and optionally sends this amplified through the physical layer. The

Computation of the asymmetrical delay can now again start again, since by reading out the transmission data from the memory 130, only the asymmetric delay of the oscillator 132, that is the active star coupler 100 itself, has an influence on the data signal 41 at the output of the star.

The use of the FIFO asynchronous memory 130 is required because of the frequency offset between the synchronized clock 138 and the clock of the transmitting subscriber and the clock 134 of the active star 100 no direct connection of data lines is permitted. Otherwise, transmission errors could occur with longer data frames due to the two asynchronously running clock signals 134, 138. For this reason, the asynchronous FIFO memory 130 is used as an interface between the two clock levels. The asynchronous FIFO memory 130 can be used for data exchange between completely independent clock domains. The removal of the asymmetrical delay from the data signal 40 is obtained by buffering the data and thus by slightly increasing the propagation delay (the propagation delay). According to the

FlexRay specification V.2.1 must be taken to ensure that the maximum delay of a active star coupler does not exceed two bit times. The following limits apply to the timing of an active star coupler 100 according to the specification:

- Transmission delay (so-called Propagation Delay) via an active star coupler: max. 250 ns

- Asymmetric delay (so-called asymmetry propagation delay): max. 8 ns

Additional shortening of the data frame (so-called truncation) by inserting an active star coupler into a transmission path between node M and node N: max. 450 ns

For the worst case consideration, a 10 Mbit / s connection is considered, since at lower data rates larger asymmetric delays may occur, in order to still ensure error-free reception of the transmitted data. If one expresses the transmission delay in bit times for a 10 Mbit / sec connection, one comes to a maximum allowable delay of less than or equal to 2.5 bit times. A data bit may thus be stored in the FIFO memory 130 for a maximum of 2 bit times since 3 bit times would already be too long. At the beginning of the transmission or the data frame (so-called frames), 4.5 bits will be lost (so-called truncation).

Conventional FIFO memories are used to store 8-bit data words, although other variants are also available (see data sheets of the manufacturers of FIFO memories, for example Texas Instruments, Inc.). In general, the information is stored in a random access memory (RAM). For use in an active star coupler 100 according to the invention the storage of complete data words is not required. It is sufficient to store only one bit in a memory at a time. The "durability" of the stored information must also not be very long, that is, the memory can be made short because of the maximum allowable transmission delay. In the present case, a 4-bit FIFO memory is sufficient.

In the context of the present invention, a special storage concept is introduced. FIG. 4 shows a 1-bit asynchronous FIFO memory 130 in ring structure of length 4. The FIFO memory 130 was deliberately programmed as a ring structure, since in this type the Throughput times play a lesser role compared to a conventional shift register. Because with this type of FIFOs, not the data are moved, but only pointers that point to the stored data. Thus, however, a memory with separate read and write clock is needed. The name ring memory should be traced back to the overflow of the pointers, since this creates a quasi ring of the memory. The respective pointers are obtained by clock dividers 150 to 154. The incoming data bits of the decoded receive data 160 are sequentially stored in the four D flip-flops 162-168. The storage takes place synchronously with the synchronized clock 138 with the clock signals Tl to T4. The selection of the flip-flops 162-168 is done by dividing the synchronized clock 138 by a factor of four and an output shifted by one clock time.

The output of the data or the re-send and encode with the encoding unit 142, however, takes place with the local system clock 134 of Sternkopp coupler 100, as the figure 4 can be seen. The selection of the corresponding memory location for the read operation is similar to that of the write operation. Likewise, by dividing the system clock 134 by means of clock dividers 156 and 158, the respectively following memory location is selected by a switch (so-called switch) 170. The more memory cells the ring buffer would contain, the greater the allowable difference in clock frequencies between synchronized clock 138 and local system clock 134 could be. However, with the number of memory cells, the transmission delay also increases.

Figure 5 shows the time course of the signals of the 1-bit asynchronous FIFO memory 130 in ring structure of length four from Figure 4, the arrival of a data signal (SOF; Start of Frame), after initialization and the subsequent beginning of caching of the data bits.

The system is at the beginning of the waveforms of Figure 5 in a standby state. During this period, the flip-flops are initialized according to the values named INIT. Thereafter, the beginning of the data frame (so-called data frame), which was designated SOF (Start of Frame) in FIG. In the line "Data In" the incoming data can be recognized, while at this time a synchronized clock ("CIk syn") is present. In addition, according to the FlexRay specification V.2.1 there is an additional margin of approx. 4 Bits (see FIG. 3). By the clock divider from the corresponding trigger or delayed clock signals Tl to T4 are obtained. In the following line is indicated by arrows, which bit is stored in which positive flank-triggered register DO to D3.

In parallel with the detection of the beginning of the data frame (SOF), the system clock 134 ("Sys-Clk") is also switched to the output of the asynchronous FIFO memory 130. Due to the initialization, this starts to read the data bits by two register values offset from the ring buffer. Thus, the first two bits (D2, D3) applied to the output ("Output") must be discarded before the first payload data (DO, Dl,...) Are read from the FIFO memory 130. It can easily be seen from FIG. 5 that propagation delay (propagation delay) was increased by use of the present invention, but with a duration of 2 bit times exactly in the range of the permissible limit value according to the FlexRay specification.

If one further sets the offset by a maximum of one bit duration as the maximum permissible frequency offset of the two clock frequencies (with enough space for this assumption due to the ring structure of length four with start offset by two memory locations after initialization in positive and negative offset directions), this results in Quartz quality to:

Goodness =! L-- N _frame + ± _ _χ

N frame

with r _b as the data rate and N _Fram e as the number of bits per data frame (so-called frame). A FlexRay data frame contains a maximum of 2096 bits according to the specification. For a worst case consideration, this maximum frame length is important, resulting in a maximum frequency tolerance of

Quartz tolerance = \ f _c - f _c

results. Conventional quartz oscillators even have a tolerance range of only 100 ppm.

In summary, the present invention thus provides a particularly advantageous possibility of reducing the asymmetrical delay even in complex network structures to a very low value, which lies without problems within the tolerance window for the asymmetrical delay permissible according to the protocol specification. In order to reduce the asymmetrical delay at the receiver, the information transmitted with the data signal is used with respect to the clock rate of the signal transmission. This information is transmitted in the form of a predetermined edge change at a defined time within a data bit of the data signal. The reference edge change is introduced into the data signal in the transmitter and extracted again at the receiver from the received data signal in order to determine the clock rate of the signal transmission. The present invention can be easily combined with further measures to reduce the asymmetric delay or otherwise improve the data transmission behavior in the network structure, for example a variable adaptation of the sampling instant.

Claims

claims

1. A method for determining the clock rate of a data signal in a

Communication system comprising a data bus and a plurality of subscribers connected thereto is transmitted from a transmitting subscriber via the data bus to a receiving subscriber, characterized in that for superimposing the

Data signal with additional information regarding the clock rate transforms the data signal at the sending party, the transformed data signal is transmitted over the data bus and transformed back at the receiving party, wherein the additional information regarding the clock rate is recovered in the context of the inverse transformation.

2. The method according to claim 1, characterized in that by the transformation of

Data signal at a defined position within the data bits of the data signal at least one reference edge change is generated.

3. The method according to claim 2, characterized in that the data signal is encoded at the transmitting subscriber with a baseband coding and decoded at the receiving subscriber with a corresponding baseband decoding.

4. The method according to claim 3, characterized in that as baseband coding a Manchester coding, a Biphase-Mark coding, a Miller coding, a coding by means of Barker sequences, an encoding by means of a bipolar code, an AMI code or a HDBn Codes is used.

5. The method according to claim 2, characterized in that the data signal is transformed at the transmitting subscriber by means of a scrambler and transformed back at the receiving subscriber by means of a descrambler.

6. The method according to any one of claims 1 to 5, characterized in that the transmitted information of a received data signal is used to correct the clock rate at the receiving subscriber of the data signal.

7. The method according to claim 6, characterized in that the incoming data signal is oversampled at the receiving party and the location of the sampling point

Variation of a local clock frequency in the receiving subscriber in dependence on the transmitted information is varied.

8. The method according to claim 7, characterized in that the incoming coded data signal is oversampled 8 times at the receiving subscriber and the local clock frequency in the receiving subscriber is varied in dependence on the transmitted information such that the receiving subscriber 7 to 9, preferably 8 , Clock cycles used to decode the incoming coded data signal.

9. The method according to any one of claims 6 to 8, characterized in that the clock rate of a received in an active star coupler of the communication system with star topology data signal is detected, wherein the data transmitted via the star coupler data signal decoupled from the incoming data signal with respect to an asymmetric delay in the star coupler and reducing the asymmetric delay of the data signal transmitted via the star coupler in the star coupler, such that the data signal leaving the star coupler has a smaller asymmetric delay than the incoming data signal.

10. The method according to claim 9, characterized in that the information superimposed on the data signal of the incoming data signal with the synchronized clock rate of the communication system are cached in a memory element of the star coupler.

11. The method according to claim 10, characterized in that the cached information of the incoming data signal with a local clock rate of the active star coupler is read from the memory element, wherein the local clock of the active star coupler is independent of the synchronized clock of the communication system and the local clock of active star coupler is regulated.

12. A communication system comprising a data bus and a plurality of subscribers connected thereto for transmitting and / or receiving a data bus transmitted via the data bus, characterized in that a transmitting subscriber for superimposing the data signal to be transmitted with additional information regarding the clock rate means for transforming the data signal and a receiving subscriber for the purpose of recovering the additional information relating to the clock rate has means for inverse transformation of the incoming data signal.

13. Communication system according to claim 12, characterized in that the transformation of the data signal generates at least one reference edge change at a defined position within the data bits of the data signal.

14. Communication system according to claim 13, characterized in that the transmitting subscriber encodes the data signal to be transmitted with a baseband coding and the receiving subscriber decodes the received encoded data signal with a corresponding baseband decoding.

15. Communication system according to claim 14, characterized in that the

Baseband coding comprises a Manchester coding, a Biphase-Mark coding or a Miller coding.

16. A communication system according to claim 13, characterized in that the transformation means of the transmitting subscriber as a scrambler and the inverse transform means are formed at the receiving subscriber as a descrambler.

17. Communication system according to one of claims 12 to 16, characterized in that the receiving subscriber is part of an active star coupler with at least one input for the incoming data signal and at least one output for an outgoing data signal.

18. An active star coupler for use in a star topology communication system, the star coupler having at least one input for an incoming data signal and at least one output for an outgoing data signal, characterized in that the star coupler comprises means for receiving one of a transmitting one Subscriber of the communication system has transmitted data signal, wherein the transmitting party has previously transformed the data signal to be transmitted to superimpose the data signal to be transmitted with additional information regarding the clock rate of the data signal, and the star coupler means for inverse transformation of the received data signal for the purpose of recovering the additional information having.

19. Star coupler according to claim 18, characterized in that the star coupler detects at least one reference edge change at a defined position within the data bits of the data signal in the context of the inverse transformation of the received data signal and determines therefrom the clock rate of the received data signal.

20. The star coupler according to claim 19, wherein the star coupler decodes the received, previously coded data signal with a baseband decoding corresponding to the baseband coding used to encode the data signal to be transmitted in the transmitting subscriber of the communication system.

21. Star coupler according to claim 20, characterized in that the baseband coding comprises a Manchester coding, a Biphase-Mark coding or a Miller coding.

22. A star coupler according to claim 19, characterized in that the inverse transforming means are designed as a descrambler, which transforms back the received data signal previously transformed by the sending subscriber by means of a scrambler.