GB2362077A

GB2362077A - Precoder for data transmission with stability monitor

Info

Publication number: GB2362077A
Application number: GB0028479A
Authority: GB
Inventors: Malcolm Paul Sellars
Original assignee: Adaptive Broadband Ltd
Current assignee: Adaptive Broadband Ltd
Priority date: 1999-11-23
Filing date: 2000-11-22
Publication date: 2001-11-07
Anticipated expiration: 2020-11-22
Also published as: GB9927673D0; GB0028479D0; GB2362077B

Abstract

A precoding scheme for digital data transmissions across multipath channels is described, which does not suffer from instability. Before transmission, this precoding scheme uses a stability prediction algorithm to predict whether or not the precoder will be stable. If the precoder is predicted to be unstable, then the precoder is deactivated, and the data is transmitted without any precoding which means that the received signal requires equalization. If the precoder is predicted to be stable, the precoder is activated, and the data is precoded before transmission. If the precoder is activated, the received signal does not require equalization.

Description

2362077 PRECODER FOR DATA TRANSMISSION

TECHNICAL FIELD

The present invention relates to a precoder for use in data transmissions, and more particularly to an adaptive technique for precoders of the type used in data communications systems. In particular, the present invention is concerned with activating the precoder only if the precoder is stable.

PRIOR ART

The present invention has been developed in the context of a wireless Asynchronous Transfer Mode (ATM) networking infrastructure which is capable of supporting multimedia data traffic at high bit rates 111 local and wide areas, but the invention could also be used with other networking infrastructures, and in particular is not limited to the use of ATM.

In radio communications systems, the transmitted signal travels from transmitter to receiver over a channel consisting of a number of different paths, known as multipaths. The transmitted signal component travelling along a particular path experiences absorption, reflection or scattering by any objects located along that path. This causes the signal components arriving at the receiver to have different amplitudes, phases and delays so that they interfere with one another. This interference is known as multipath fading, and results in Intersymbol Interference (ISD at baseband. ISI describes the spreading out of the data symbols so that components of past and future symbols are superimposed on the current symbol, thus making symbol detection difficult, and causing symbol errors.

An equalizer is often used at the receiver to remove ISI caused by the channel. In order to operate correctly, the equalizer requires a known preamble (or training sequence) to be transmitted together with each burst (or packet) of data symbols. A decision feedback equalizer (DFE) is a type of equalizer commonly used in radio communication systems since it is able to deal with channels having 2 spectral nulls. Equalization introduces an overhead in data rate, since the preamble uses up symbols which could otherwise have been used to carry data. Equalization also introduces an overhead in packet delay, since the equalizer requires a certain period of time to train itself to the channel with the help of the preamble. A separate equalizer is required at both terminals (base station and mobile), which means that the equalization overhead is present on both the downlink (base station to mobile transmission) and on the uplink (mobile to base station transmission).

In a wireless ATM system with many mobile ternunals accessMg a single base station, the equalizer at the base station will be required to re-train for each received signal from a different mobile. This means that delays for re- training will be frequent, and will reduce the overall data rate.

Wireless packet-switched systems (such as ATM) may use time division duplex (TDD) operation so as to make the most efficient use of the asymmetric data rate requirements of the mobiles and base station. The TDD mode of operation allows the use of precoding, where the equalization load is shared asymnietrically between transmitter and receiver. The underlying assumption is reciprocity of the channel impulse response, since TDID performs 2-way transmission on the same carrier frequency. Thus the downlink channel and uplink channel are assumed to have the same impulse response, provided that the interval between downlink and uplink transmissions is short.

Mobile radio systems aim to make simpler and cheaper mobile terminals by moving the equalization load to the base station. A system with the equalizer and precoder located in the base station terminal is shown in Figure 1.

The term precoding is used to describe equalization performed at the transmitter. The precoder requires knowledge of the channel impulse response H(z). This may be estimated from the converged taps of the equalizer on the forward link. Ideally, the transfer function of the precoder should be the iriverse of the channel transfer fimetion. When the fading channel is not minimum- phase, a non-linear operation is needed to ensure stability of the precoder. - 3 A prior art method of precoding is TomIMson-Harashiina precoding (TH- precoding), which uses a modulo operator block at both transnuitter and receiver. The system is shown in Figure 2. The output of the modulo M block always has a magnitude III the range -W2 to +NV2, and this ensures stability of the precoder.

However, there are a number of practical difficulties with the modulo block used in TH precoding:

I. Signal amplitude at the receiver is distorted, and this causes the automatic gain control (AGQ level to vary due to signal level fluctuations.

2. Dynamic range of the received signal is increased, and therefore the number of useful quantizer levels at the receiver is reduced. The anti alias filters at the receiver will also require increased dynamic range.

3. Need for synchron-lzation between modulo blocks at the transmitter and receiver. The receiver has difficulty in distinguishing between 'spikes' (due to transmit modulo operations) and noise, so causing modulo errors.

The error rate increases due to reduction in size of the signal decision regions.

On channels for which the precoder is unstable, use of the precoder results III increased error rates or may require the use of stabilization methods for the precoder, as described in our earlier co-pending application no. GB2347054.

For these channels, it is argued that it may be better not to use precoding at all, rather than using stabilization techniques which inevitably cannot produce an ideal pre-coder response. On channels for which the precoder is stable, use of the precoder provides a simple and effective method for removing ISI from the signal at the transmitter rather than at the receiver. This allows shorter preambles to be transmitted with each data packet, and packet decoding delays to be reduced.

SUMMARY OF THE INVENTION

We propose a pragmatic precoder, which allows us to adaptively activate or deactivate the precoder, depending on whet-her or not the precoder is 4 predicted to be stable. This avoids the need for stabilization techniques such as modulo-operations, which are used in conventional precoders to ensure stability.

In order to meet the above, from a first aspect the present invention provides a precoder for pre-equalising data prior to transmission onto a communications channel, said precoder comprising:

equalisation means for substantially pre-equalising data to be transmitted onto the communications channel; and stability prediction means arranged to monitor the equalisation means to predict if said equalisation means is stable; wherein if said stability prediction means predicts that said equalisation means is stable then said data is pre-equalised by said equalisation means, else if said stability prediction means predicts that said equalisation means is not stable, then said data is not pre-equalised.

By "pre-equalization", we mean the same as "pre-coding", that is the equalization of the data to be transmitted with the inverse of the communications channel response, prior to transmission of the data onto the channel. The terms "pre-equalization" and "pre-coding" are therefore used herem interchangeably.

In addition to the above, in one embodiment the precoder of the present invention fixther comprises means for indicating to a receiver whether or not said data has been pre-equalised by the precoder prior to transmission.

The above described embodiment can provide particularly advantageous operation, as if the precoder is predicted to be unstable, then the precoder is deactivated, and the receiver is mdicated that no pre-coding has occurred. In this case the receiver can switch on the receive equalizer, since the received signal is to be transmitted without precoding. A longer preamble is attached to the front of the data packet to allow the receive equalizer to tram correctly. Alternatively, if the precoder is predicted to remain stable, then the precoder is activated, and the data is precoded prior to transmission, and the receiver is indicated that pre-coding is to occur. In this case the receiver switches off the receive equalizer, since the data is to be precoded before transmission. This allows a shorter preamble to be used in the data packet, since equalizer training C is not required. This also means that the packet decoding delay will be greatly reduced.

Furthermore, in the above described embodiment the indication means for indicating to a receiver whether or not the data has been pre- coded may ftuther comprise means for packetising the data prior to transmission, wherein the data is packetised into a data packet of a first type if it has been pre- equalised, and packetised into a data packet of a second type if it has not been pre- equalised.

By using different packet types to indicate whether or not pre-coding of the data has occurred, a receiver can differentiate whether or not precoding has occurred (and hence whether or not equalization is required) by distinguishing between the packet types without having to perform symbol detection of the data in the packet. More particularly, by "packet type" we do not mean that the packets are of different forrnats such as ATM or X.25, but rather merely that they are distinguishable on the basis of the raw received data, including any pre- cursors or post-cursors which lead to ISI, such that it is possible to tell whether or not equalization has been performed before any symbol detection has been performed.

In one embodiment of the above, the data packets of the first type and of the second type are of different lengths to each other, such that a receiver may differentiate between the two types of data packet (and hence tell whether or not pre-coding has been performed) based on length of the received packets.

In another embodiment of the above, the data packets of the first and second types each have differing correlation sequences, and detection is performed at a receiver by detecting the differing correlation sequences in a correlator. In one preferred embodiment the data packets of each of the first or second types have respective different correlation sequences which may be of the same length, and a receiver performs two correlation operations to determine which correlation sequence has been appended to a particular received packet. This has the 6 advantage that channel efficiency is retained at a maximum, but processing efficiency is reduced by having to perform two different correlation operations on different sequences. In an alternative embodiment, therefore, the same correlation sequence is appended two or more times to the packet, and the correlator at the output detects the two or more correlation peaks to determine the packet type.

In a further alternative embodime.nt, the precoder indicates to the receiver whether or not pre-coding has occurred by using a separate lowbit rate communications side-channel to signal to the receiver whether or not precoding of the data m any particular data packet has been performed.

In addition to the above, the present invention also provides a receiver for receiving data transmissions from a transmitter comprising a precoder according to any of the embodiments as substantially hereinbefore described, said receiver comprising:

means for receiving data packets from the transmitter over a communications channel; precoding detection means for detecting whether data received in a data packet has been pre-equalised based on an indication received from the precoder; and equalisation means for substantially equalising data received from the communications channel; wherein if said precoding detection means detects that a received data packet has been pre-equalised then said data received in the data' packet is not equalised by said equalisation means, and wherein if said packet detection means detects that a received data packet has not been pre-equalised then said data received m the data packet is equalised by said equalisation means.

Preferably, the precoding detection means detects whether or not the received data has been pre-equalised without performing any symbol detection, that is, by looking at the raw received data packet, which may include the pre- and post- cursors which cause ISI at baseband, to determine the packet type.

7 In a preferred embodiment, the pre-coding detection means comprises a correlator arranged to detect a correlation sequence pre- fixed to the front of the received data packet. In one alternative of the preferred embodiment the data packets of the different types have different correlation sequences such that the correlator performs two different correlations, the different sequences being respectively indicative of a particular packet type. In another alternative of the preferred embodiment a data packet of one of the two types has a greater number of the same correlation sequence pre-fixed thereto than the other type, such that the correlator outputs a greater number of correlation peaks from the same correlation sequence for one packet type than the other packet type.

From another aspect of the present invention, the present invention also provides a data communications system comprising a transmitter terminal and a receiver terminal, the transmitter terminal further comprising a precoddr for pre C equalising data prior to transmission onto a communications channel, the precoder comprising:

equalisation means for substantially pre-equalising data to be transmitted onto the communications channel; and stability prediction means arranged to monitor the equalisation means to predict if said equalisation means is stable; wherein if said stability prediction means predicts that said equalisation means is stable then said data is pre-equalised by said equalisation means, else if said stability prediction means is not stable, then said data is not pre-equalised; the receiver terminal further comprising an equalisation means for substantially equalising data received from the transmitter terrainal, wherein the equalisation means in the receiver terminal equalises the received data notwithstanding that the data may have been pre-equalised by the precoder.

In the above system, the precoder of the present invention does not 8 send any signals to the receiver. In this case, the equalizer at the receiver is always active. This allows error rates to be reduced, since any small amount of ISI which is not removed by the precoder will be removed by the equalizer at the receiver.

This also means that if the channel changes before re-transmission on the uplink (which means that the precoder coefficients will be inaccurate and ISI will be present), then the receiver equalizer will remove the remaining ISI.

DESCRIPTION OF THE FIGURES

Further features and advantages of the present invention will become apparent from the following descriptions of a number of preferred embodiments of the itivention below, presented by way of example only, and with reference to the accompanying figures, in which.

Figure I shows a block diagram of a communications system with the equalizer and precoder located at the Base Station (PRIOR ART);

Figure 2 shows a block diagram of a typical Tomlinson-Harasbima precoder (PRIOR ART);

Figure 3 shows a block diagram of a communications system on which the present invention is based, and where both the equalizer and precoder are located at Terminal B; Figure 4 shows a block diagram of an embodiment of the pragmatic precoder of the present invention, and of the receiver of the present invention; and Figure 5 shows data packets with different preambles, depending on whether the precoder is predicted to be stable or unstable, and used in an embodiment of the present invention.

DESCRIPTION OF THE PREFERRED EMBODIMENTS

The precoder and receiver of the present invention are particularly for use in a communications system with the equalizer and precoder located in a single terminal (e.g. Terminal B), as shown in Figure 3 (Terminal A and Terminal 1 9 B may be either mobile or base station terminals). Therefore, M order to allow for a fall understanding of the operation of the present invention, description is first undertaken of the operation of such a system, with reference to Figure 33.

Here, on the downlink Terminal A (1) transmits data to Terminal B (2). Data symbols (3) are input to the transmitter of Terminal A, where they pass into a modulator (4). The modulator converts the data into a radio signal for transmission (5). This radio signal is transmitted over the downlink channel (6), and arnves as a distorted signal (7) at Terminal B (2). At Terminal B. the received signal (7) is demodulated by a demodulator (8), and then passed into a DFE (9).

The DFE (9) consists of a feedforward filter and a feedback filter. The DFE removes ISI from the signal, and outputs a set of data symbols (10). At the end of a received data burst, the channel characteristics (11) measured by the DFE, are passed into a channel estimator block (12). The channel estimator generates a set of tap coefficients (13) which are transferred into the taps of the precoder (14).

Data symbols (15) which are to be transmitted from Tenninal. B to Terminal A, are passed through the precoder (14), to produce a stream of data samples (16) which have been precoded (or pre-equalized) with the inverse of the channel impulse response H(z) (6). The stream of precoded data samples (16) is passed through a modulator (17) to produce a radio signal (18) for transmission over the uplink channel ( 19). It is assumed that the uplink channel (19) has an identical channel impulse response H(z) to the downlink channel (6). After transmission over the uplink channel (19), the radio signal (20) which arrives at Terminal A is completely free from multipath interference, since it has already been precoded (pre-equalized) before transmission. This means that no further equalization is required at Tem-iinal A (1). At Terminal A, the received signal (2Q) is passed into a demodulator (2 1), which outputs a stream of data samples (22) which are free from hitersymbol Interference. (ISD due to the precoding. Since there is no ISI, the data samples (22) may be detected by a simple detector (23), without any equalization, to obtain the decoded data symbols (24). In a network where many mobiles cornmunicate with a single base station, this eliminates the problem of equalization and training delays at the base station.

Having described the background of the operation of the present invention, description can now be undertaken of the precoder and receiver of the present invention.

A first embodiment of the pragmatic precoder of the present invention and located at Terminal B in the data communication system as described earlier will now be undertaken with reference to Figure 4. Here, the pragmatic precoder of the first embodiment comprises a data input terminal Xk (52) arranged to feed input data to a switch Sp (62). The switch Sp is arranged to switch between a first position a (64) and a second position b (63). In the first position a the switch feeds data input at the input terminal to a precoder (14), such as a DFE loaded with inverse channel response filter coefficients, as described previously, and in the second position b over a feedforward line to an output terminal 69, thereby bypassing the precoder. Data from the output terminal is output from the precoder as output data Wk (55), which may or may not have been precoded dependent on the position of switch Sp.

To control the position of switch Sp, a stability predictor 60 is provided arranged to receive the same filter coefficients as the precoder and to predict the stability of the precoder therefrom. A control line SLp (6 1) is provided output from the stability predictor to the switch Sp, which acts to control the position of the switch Sp between the positions a and b.

The operation of the precoder of the present invention is as follows.

The converged filter taps from the DFE (9) (see Figure 3) are loaded directly into the precoder (14) and also into the Stability Predictor (60). The overall precoder transfer function is A(z) = C(z)/H,,(z) = C(z)/(I+B(z)) = 1/H(z), where H(z) is the transfer function of the channel (19). Precoder stabilization may be required if there are critical zeros in the transfer function HB(Z). The Stability Predictor (60) uses the converged taps of the DFE feedback filter to calculate reflection coefficient values. The magnitude of the largest coefficient value Irl, determines whether or not the precoder will be stable. If the value Irl is less than one, then all the zeros of HB(z) he within the unit circle, and so the precoder will remain stable.

If the value of Irl is greater than one, then some of the zeros of H,(z) lie outside the unit circle in the z-plane, which means that the precoder is unstable.

The Stability Predictor (60) uses the coefficients of 'the DFE feedback filter to calculate reflection coefficients as follows. For example, if the FB section of the precoder has a transfer function I/H,(z) where:

FB filter:

H,(z) = B(z) + I The coefficients of B(z) = [bz-', bz', bBZ-B] are transformed into the corresponding lattice filter coefficients G(z) = [g,z-', g,z', 9BZ-B1 using the Levinson recursion, or other iterative method. The magnitudes of the reflection coefficients Igl, Jg2j, 19BI are examined, and the largest value is denoted Irl. If Irl is greater than one, then the Stability Predictor (60) predicts that the. precoder (14) will be unstable. If Irl is smaller than one, then the Stability Predictor (60) predicts that the precoder (14) will be stable.

Following the above, the pragmatic precoder (50) functions as follows: if the Stability Predictor (60) predicts precoder instability, then a signal is sent along the precoder control line SLp (61) to move the switch Sp (62) to position b (63), such that the data Nk is not switched through the precoder and the data is not pre-equalised prior to transiMssion. In this case the data is fed forward along the line 63 to output terminal 69, from where it is then output for transmission.

In contrast if the Stability Predictor (60) predicts that the'precoder will be stable, then it is not necessary to send any message along the precoder control line SLp (61). In this case, switch Sp (62) will remain in position a (64) this means that the precoder (14) is activated. Data symbols (52) to be transnutted from Terminal B (2) to Terminal A (1) are input to the Pragmatic Precoder (50).

12 These data symbols pass through switch Sp (62), where they are directed irito, the Precoder (14), where they are pre-equalised with the inverse of the communication channel 19 impulse response. The pre-equalised data is then fed to the output terminal 69 for output for transmission as pre-equalised data.

With the above, the present invention provides a pragmatic precoder which allows for pre-coding (pre-equalization) of data to be performed prior to transnUssion of the data dependent on whether or not the pre-coder is predicted to be stable or not. This removes the need for stabilization techniques or the use of modulo operations at the transmitter side.

In use within a communications system, the pragmatic precoder of the first embodiment is preferably used m a transmitter which transmits to a receiver which is also provided with a conventional equaliser, such as DFE. As the pragmatic precoder in the transmitter may be switched off for some data packets (i.e. the data therein has not been pre-coded), the equaliser in the receiver should operate continuously, to compensate for those times when the precoder m the transmitter is turned off or bypassed. In such an operation, a minimum level of precoding or equalization is performed which should allow for adequate symbol detection, but where both pre-coding m the transmitter and equalization in the receiver are performed then error rates can be substantially reduced as almost all ISI in the received signal should be removed.

A second embodiment of the precoder of the present invention will now be described, again with reference to Figure 4, and with reference to Figure 5.

The second embodiment incorporates all of the features and flirictions of the previously described first embodiment, but is further arranged to provide an indication to a receiver as to whether or not the data contained within a particular data packet has or has not been precoded, such that the receiver can then perform equalization in a DFE or the like if no pre-equaliZation has occurred.

More particularly, the pragmatic precoder (50) of the second 13 embodiment of the invention functions as follows: if the Stability Predictor (60) predicts precoder instability, then a signal is sent along the precoder control line SLp (61) to move the switch Sp (62) to position b (63) such that the data bypasses the precoder and is not precoded, as described earlier in relation to the first embodiment. In the second embodiment, however, the resulting signal wk output from output terminal (69) and which contains data which has not been precoded is encapsulated in a longer than usual data packet (84), which has, a longer correlation sequence attached to the front of it, and which is used to 'indicate to the receiver that the receiver equalizer (70) must be switched on, as described in relation to the receiver later.

In contrast to the above, if the Stability Predictor (60) predicts that the precoder will be stable, then it is not necessary to send any message along the precoder control line SLp (61), and a shorter data packet (80) with a shorter preamble (81) is transmitted, to indicate to the receiver that the data has been pre coded. In this case, switch Sp (62) will remain in position a (64): this means that the precoder (14) is activated, and the receive equalizer (70) should be deactivated.

Data symbols (52) to be transmitted from Terminal B (2) to Terminal A (1) are input to the Pragmatic Precoder (50). These data symbols pass through switch Sp (62), where they are directed either into the Precoder (14). The data symbols arrive at the point (69), and are output as a precoded signal wk (55). The precoded signal (5 5) is transmitted over the uplink channel ( 19).

In Figure 5, the two types of transmitted data packet used in the second embodiment are shown. More particularly, if the stability predictor (60) predicts that the precoder will be stable, then the short data packet (80) will be transmitted, which has a short preamble (8 1) comprising correlation sequence 1, followed by the data (83). In contrast if the stability predictor (60) predicts that the precoder will be unstable, then the longer data packet (84) with the longer preamble comprising correlation sequences 1 (8 1) and 2 (82), Will be transmitted.

It will be appreciated that the correlation sequences I and 2 may be identical or 14 different. Furthermore, it will also be appreciated that in a finther alternative embodiment, the data packet of the second type carrying the non-precoded data could simply have a different correlation sequence (e.g. correlation sequence 2) to the correlation sequence of the data packet of the first type carrymg the precoded data.

A preferred embodiment of the receiver of the present invention will now be described with reference to Figures 4 and 5.

In Figure 4, a receiver accordmig to the present invention comprises a buffer (73) arranged to receive received data packets and buffer them therein, and a correlator (72) also arranged to receive received data packets, and to detect the one or more correlation sequences pre-fixed to each packet. A control line (7 1) is provided between the correlator (72) and the buffer (73) over which a control signal (7 1) is sent from the correlator to the buffer to allow the correlator to control the buffer. The first pole of a switch Sr (66) is arranged to receive data output from the buffer, and to switch the data to a first contact a (6), or a second contact b (67) in response to a signal SLr received from the correlator(72) to the switch Sr over a control 1Me (65), In position a, the switch Sr switches the received data to bypass an equaliser (70), such as a DFE or the like, such that the received data is fed directly to a symbol detector (23) for symbol detection without undergoing equalization. In position b the switch Sr switches the received data into the equaliser (70), such that the received data is fed through the equaliser for equalization and is then output from the equaliser to the symbol detector (23).

The operation of the receiver of the present invention is as follows, given the previously described operation of the precoder of the second embodiment.

A received signal Rk (57) at the Terminal A receiver (5 1) passes into the buffer (73) and also into the correlator (72). The correlator detects correlation peaks in the received signal, and uses this information to achieve timing synchronization. When synchronization is found, the correlator sends a tin-ling signal Tr (7 1) to the buffer (73), instructing the buffer to begin passing the received signal into switch Sr (66). In addition, the correlator also counts the number of correlation peaks resulting from the correlation operation, and uses this information to control whether or not the received data requires equalization or not. In particular, the correlator (72) sends the necessary signal along receive control lMe SLr (65) to set the position of switch Sr (66) dependent upon the number of correlation peaks in the received signal.

More particularly, if the receiver correlator (72) detects a long preamble in the received data packet (i.e. the data packet is of the second type which contains data which has not been pre-equalised) M that two or more correlation peaks have been detected, it will send a signal on the receive control line SLr (65) to move the switch Sr (66) to position b (67), which activates the receive equalizer (70). The long preamble (8 1) and (82) of the long data packet (84) is also used to train the receive equalizer (70). Alternatively, if the Stability Predictor (60) m the transmitter precoder has predicted that the precoder will be stable, then the data in the present data packet will have been precoded and a short data packet (80) with a short preamble (8 1) will have been received (i.e. a data packet of the first type containing precoded data), meaning that less correlation peaks (e.g. one) will be detected than in the case of the longer preamble. In this case, switch Sr (66) Will remain m position a (68): this means that receiver equalizer (70) is deactivated.

When the switch Sr is in position a (68), then the received signal (57) is assumed to be free from ISI, and so requires only a simple detector (23) instead of a full equalizer. When the Stability Predictor (60) predicted precoder instability, then switch Sr will be in position b (67). In this case the received signal (57) passes through the receive equalizer (70) where the ISI is removed, before passing into the detector (23), to obtain the decoded data symbols (24).

The above described receiver therefore complements the precoder of the second embodiment, in that it is arranged to receive an mdication of 16 whether the received data has or has not been pre-coded, and can then activate a receiver equaliser as approphate. This has the advantage that it becomes unnecessary to perform equaliser training at the receiver for every received burst, as training need only be performed for those bursts which have been indicated by the precoder in the transmitter as not having been pre-coded. This can have significant advantages in a point-multipoint TDMA data communications system, as it means that the base station does not have to retrain for evdr3 6rst received from a different subscriber terminal, but only for those bursts for which pre-coding was not possible in the transmitter.

Whilst the above described embodiments.have concentrated on using the correlation sequence to distinguish between packet types as to whether or not the data m a particular packet has been pre-coded or not, other means and methods for allowing the precoder in the transmitter to indicate to the receiver as to whether equalization is required are also apparent. In particular, in an alternative embodiment, the transmitter terminal B transmits a low bit rate side channel M addition to the main signal to signal to the receiver over the side channel as to whether a particular data packet is of the first type or the second type i.e. contams pre-equalised data or not. If the indication for a particular data packet is made at the same time as the data packet is transmitted over the primary channel, the side channel need only have the same capacity as the number of data packets to be transmitted per second, as signalling could be performed on a side channel symbol/per packet basis.

In another embodiment, the data packets of the first and second types could have the same correlation sequences but be of different lengths merely by one type having stuffing bits or a stuffing field m order to make the packet longer m order to allow for differentiation between the packet types. Here, merely by counting the length of the received packet as it is clocked into the buffer a differentiation may be performed. The drawback is encountered however that the entire packet would have to be received before the differentiation between whether 17 or not equalization was required could be made, thus introducing a one- packet delay into the system, which may be unacceptable in applications with low- latency requirements.

Further signalling methods and means for indicating to the receiver whether or not equalization is required will be apparent to the man skilled in the art, and which may be used with and/or fall within the scope of the present invention.

18

Claims

1. A precoder for pre-equalising data prior to transmission onto a communications channel, said precoder comprising:

equalisation means for substantially pre-equalising data to be transmitted onto the communications channel; and stability prediction means arranged to monitor the equalisation t means to predict if said equalisation means is stable; wherein if said stability prediction means predicts that said equalisation means is stable then said data is pre-equalised by said equalisation means, else if said stability prediction means predicts that said equalisation means is not stable, then said data is not pre-equalised.

2. A precoder according to claim 1, wherein said stability prediction means is further arranged to output a control signal indicative of the stability of the equalisation means; and wherein said precoder further comprises a signal switching means arranged to receive the control signal and to switch the received data so as not to pass through the equalisation means if the control signal midicates that the equalisation means is not stable.

3. A precoder according to any of the preceding claims, wherein said stability prediction means is arranged to calculate the magnitude of one or more reflection coefficients from filter coefficients used in the equalisation means, wherein if the magnitude of at least one of said reflection coefficients is greater than 1, then said stability prediction means predicts that said equalisation means is not stable.

4. A precoder according to any of the preceding claims, agd further 19 comprising pre-equalisation indication means for indicating to a receiver whether or not said data has been pre-equalised by the precoder prior to transirussion.

5. A precoder according to claim 4, wherein said pre-equalisation indication means further comprises means for packetising said data prior to transmission, wherein said data is packetised into a data packet of a first type if it has been pre-equalised, and is packetised into a data packet of a second type if it has not been pre-equalised.

6. A precoder according to claim 5, wherein said data packet of the first type is of a different length to said data packet of the second type.

7. A precoder according to claim 6, wherein said data packet of the first type is shorter than said data packet of the second type.

8. A precoder according to any of claims 5, 6, or 7 wherein said data packets of the first and second types have differing correlation sequences.

9. A precoder according to claim 8, wherein the correlation sequence of one type of said data packets of the first and second types is longer than the corTelation sequence of the other type of said types.

10. A precoder according to claim 8 or 9, wherein the correlation sequence of one type of said data packets of the first and second types comprises the correlation sequence of the other of said types repeated one or more times.

11. A receiver for receiving data transmissions from a transmitter comprising a precoder according to any of claims 4 to 10, said receiver comprising:- means for receiving data packets from the transmitter over a communications channel; precodffig detection means for detecting whether data received in a data packet has been pre-equalised, based on an indication received from the precoder;and equalisation means for substantially equalising data received from the communications channel; wherein if said precoding detection means detects that data in a received data packet has been pre-equalised then said data received in the data packet is not equalised by said equalisation means, and wherein if said precoding detection means detects that data in a received data packet has not been pre equalised then said data received in the data packet is equalised by said equalisation means.

12. A receiver according to claim 11, wherein said detection means is further arranged to output a control signal indicative of whether the data in the present data packet has been pre-equalised or not, and wherein said receiver further comprises a signal switching means arranged to receive the control signal and to switch the received data so as to pass through the equalisation means if the data in the present data packet has not been pre-equalised.

13. A receiver according to claims 11 or 12, wherein said precoding detection means comprises packet type detection means arranged to detect whether a received data packet is of a first type or a second type, wherein a packet of the first type contains data that has been pre-equalised, and a packet of the second type contains data which has not been pre-equalised.

14. A receiver according to claim 13, wherein said packet type detection means is flirther arranged to detect the length of the received data packets to 21 differentiate between data packets of the first type and data packets of the second type, said data packets of each respective type being of different lengths.

15. A receiver according to claim 14. wherein said data packets of the first type are shorter than said data packets of the second type.

16. A receiver according to claims 13, 14 or 15, wherein said packet type detection means is a correlator, said data packets of the first and second types having differing correlation sequences, said correlator bemig, further arranged to detect each respective correlation sequence to determine the packet type.

17. A receiver according to claim 16, wherein the correlation sequence of one type of said data packets of the first and second types is longer than the correlation sequence of the other ty-pe of said data packet types.

18. A receiver according to claims 16 or 17, wherem the correlation sequence of one type of said data packets of the first and second types comprises the correlation sequence of the other of said data types repeated one or more times.

19. A method of precodiiig data prior to transmission onto a communications channel, comprising the steps of monitoring an equalisation means to predict if said equalisation means is stable; and pre-equalismg said data m the equalisation means if it is predicted that said equalisation means is stable; wherein if it is predicted that said equalisation means is not stable, said pre-equalisation step is not performed.

20. A method according to claim 19, and further comprising the steps 22 of switching the data so as not to pass through the equalisation means if it is determined that the equalisation means is not stable.

21. A method according to claims 19 or 20, wherein said monitoring step further comprises the steps of - calculating the magnitude of one or more reflection coefficients from filter coefficients used in the equalisation means; and comparing the calculated magnitudes with a threshold value of 1 wherein if the magnitude of at least one of said reflection coefficients is greater than the threshold value, then said equalisation means is predicted to be unstable.

22. A method according to any of claims 19 to 21, and further comprising a step of indicating to a receiver whether or not said data has been pre equalised by the precoder prior to transmission.

23. A method according to claim 22, wherein said indicating step further comprises the step of packefising the data prior to transmission, wherein said data is packetised into a data packet of a first type if it has been pre- equalised, and is packetised into a data packet of a second type if it has not been pre- equalised.

24. A method according to claim 23, wherein said data packet of the first type is of a different length to said data packet of the second type.

25. A method according to claim 24, wherein said data packet of the first type is shorter than said data packet of the second type.

26. A method according to claims 23, 24 or 25, wherein said data packets of the first and second types have differing correlation sequences.

23

27. A method according to claim 26, wherein the correlation sequence of one type of said data packets of the first and second types is longer than the correlation sequence of the other type of said data packet types.

28. A method according to claim 26 or 27, wherein the correlation sequence of one type of said data packets of the first and second types comprises the correlation sequence of the other of said types repeated one or more times.

29. A method of data equalisation comprising the steps of.

receiving data packets containing data transmitted over a communications channel; detecting whether data received in each of said received data packets has been pre-equalised based on an indication received from a precoder in a transmitter; and equalising the received data if it is detected that the data was not pre equalised; wherein if it is detected that the received data m a data packet was pre-equalised then the received data is not equalised.

30. A method according to claim 29, further comprising the steps of switching the data so as not to pass through the equalisation means if it is detected that the received data was pre-equalised.

31. A method according to claim 29 or 30, wherein said detecting step flu-ther comprises the step of detecting whether a received data packet is of a first type or a second type, wherein a data packet of the first type contains data that has been pre-equalised, and a data packet of the second type contains data which has not been pre-equalised.

24

32. A method according to claim 3 1, wherein said detection step is further arranged to detect the length of the received data packets to differentiate between data packets of the first type and data packets of the second type, said data packets of each respective type being of different lengths.

3 53. A method according to claim 32, wherein said data packets of the first type are shorter than said data packets of the second type.

34. A method according to claims 3 1, 3 2 or 3 3, wherein said detection step further comprises a correlation step, and wherein said data packets of the first and second types have differing correlation sequences, said corTelation step detects the differing correlation sequences to determine the packet type.

35. A method according to claim 34, wherein the correlation sequence of one type of said data packets of the first and second types is longer than the correlation sequence of the other type of said data packet types.

36. A method according to claims 34 or 35, wherein the correlation sequence of one type of said data packets of the first and second types comprises the correlation sequence of the other of said data packet types repeated one or more times.

3

37. A data communications system comprising a transmitter terminal and a receiver terminal, the transmitter terminal ftirther comprising a precoder for pre-equalising data prior to transmission onto a commumcations channel, the precoder comprising:

equalisation means for substantially pre-equalising data to be transmitted onto the communications channel; and stability prediction means arranged to monitor the equalisation means to predict if said equalisation means is stable; wherein if said stability prediction means predicts that said equalisation means is stable then said data is pre-equalised by said equalisation means, else if said stability prediction means is not stable, then said data is not pre-equalised; the receiver terminal further comprising an equalisation means for substantially equalising data received from the trarisrnlitter terminal, - wherem the equalisation means m the receiver ternimal equalises the received data notwithstanding that the data may have been pre-equalised by the precoder.

38. A precoder as substantially hereinbefore described and/or with reference to the accompanying Figures 4 or 5.

39. A receiver as substantially herembefore described and/or With reference to the accompanying Figures 4 or 5.