GB2409384A - Maximum Likelihood Sequence Estimation Equaliser for Space Time Coded Data in Multiple Antenna Receivers - Google Patents

Abstract

A digital communications receiver having two or more antennas is described. The two or more antennas are arranged to receive first data in a first symbol period and second data in a second symbol period. The receiver also comprises an equaliser, which combines the first data and second data to produce a combined dataset, and estimates a transmitted symbol combination giving rise to the combined dataset. A trial dataset may be estimated for each possible transmitted symbol combination, and the transmitted symbol combinations most likely to produce the combined dataset identified. This allows the use of a single equaliser. The trial datasets may be stored in a lookup table. The data may be space time block coded. (STBC). The equaliser is particularly applicable to Bluetooth(RTM) systems.

Description

Maximum Likelihood Sequence Estimation Equaliser This invention relates to

receiving an information signal in a wireless cellular network, and in particular, although not exclusively, to applying Maximum Likelihood Sequence Estimation (MLSE) equalization to a multiple antenna, low complexity, short range wireless system, such as Bluetooth High Rate.

In recent years, the use of cellular networks for wireless communications has grown tremendously. In a cellular network, multiple wireless users within a designated area, or cell, communicate with a single base-station. In a Time Division Multiple Access (TDMA) cellular network, each user communicates with the base-station in a time multiplexed fashion. In other words, each user is allocated a slice of time (i.e., a TDMA time slot) during which it exchanges a burst (or packet) of data with the base-station. A burst is a sequence of digital symbols representing the data. The user must then wait until the other users have exchanged their bursts of data with the base-station before exchanging its next burst of data.

The quality of communication in a cellular network, often expressed as bit-error-rate (BER), can be degraded by a variety of factors. Three important factors that degrade the quality of communication and increase BER are multipath fading, noise (e.g., thermal noise), and interference.

There are essentially two types of multipath fading. Flat fading results when the primary ray of the transmitted signal arrives at the receiver at approximately the same time as one or more reflections of the transmitted signal. If the primary ray and the reflections have different amplitudes and phases, they combine at the receiver in a manner that produces variations in the received signal strength. These variations can include drops in signal strength over several orders of magnitude. Time dispersion is a second type of multipath fading that occurs when the reflections arrive at the receiver delayed in time relative to one another (i.e., their propagation paths have substantially different lengths).

If the relative time delays are a significant portion of a symbol period, then intersymbol interference (ISI) is produced, wherein the received signal simultaneously contains information from several superimposed symbols. Thus, both types of multipath fading can corrupt the received signal at the receiver.

In addition to multipath fading, noise, such as thermal noise in the analogue front end of a receiver, can also corrupt the received signal at the receiver. Noise typically has a white frequency distribution (e.g., constant energy at all frequencies) and a gaussian temporal distribution, leading to the term additive, white, Gaussian noise (AWGN).

The third factor that can corrupt the received signal at the receiver is co-channel interference (CCI). CCI is the result of receiving the desired signal along with other signals which were transmitted from other radios but occupy the same frequency band as the desired signal. There are many possible sources of CCI. For example, an indirect source of CCI is adjacent channel interference (ACI). ACI is the result of side-band signal energy from radios operating at neighbouring frequency bands that leaks into the desired signal's frequency band. A more direct source of CCI is signal energy from other radios operating at the same frequency band as the desired signal. For example, a cellular radio in a distant cell operating at the same frequency can contribute CCI to the received signal in the cell of interest.

The problem of flat or Rayleigh fading can be addressed by implementing a receiver with two or more physically separated antennas and employing some form of spatial diversity combining. Spatial diversity takes advantage of the fact that the fading on the different antennas is not the same. Spatial diversity can also address interference by coherently combining the desired signal (i.e., desired symbols) from each antenna while cancelling the interfering signal (i.e., interfering symbols) from each antenna.

Other effective techniques are time and frequency diversity. Using time interleaving together with coding can provide diversity improvement. The same holds for frequency hopping and spread spectrum. However, time interleaving results in unnecessarily large delays when the channel is slowly varying. Equivalently, frequency diversity techniques are ineffective when the coherence bandwidth of the channel is large (small delay spread).

Time diversity can be achieved using a simple block coding arrangement with symbols transmitted over a plurality of transmit channels, in conjunction with coding using only simple arithmetic operations such as negation and conjugation. This is generally known as Space Time Block Coding (STBC), and is described in US patent no. 6185258.

The optimal theoretical solution to the CCI and ISI problems is a receiver that employs diversity combining and a multi-channel maximumlikelihood-sequence-estimator (MLSE) equalizer wherein the individual channel vectors (i.e., the discrete-time channel impulse responses) are known for all signals (i.e. the desired signal and all its reflections and all the interferers). The MLSE receiver demodulates the desired signals.

MLSE equalization is a well-known technique described, for example, in J. G. Proakis, "Digital Communications", McGraw Hill, 3/e 1995. The technique is described in detail below but, broadly speaking, an estimate is made of a sequence of symbols representing binary bits by hypothesizing received symbol sequences, applying to the hypothesized sequences a channel estimate for a channel over which data has been transmitted, and comparing the result with the received data to see which estimated hypothesized sequence is the closest match. Typically the best match is found by determining the minimum mean-square error (MMSE), although other metrics may also be employed, and generally the procedure is implemented using a variant of the Viterbi algorithm.

The channel estimate comprises a set of numbers which models the transmission channel, for example comprising a complex number representing a magnitude and phase of the channel response at a particular delay. The channel response may be determined at delays of integer multiples of the symbol period, in effect defining a set of multipath components. Generally the channel response decreases at longer delays and it becomes zero after the longest multipath delay. In a digital system there may be more than one sample per symbol and, in this case, the channel response may be defined at a finer time resolution. All this is well known to the skilled person.

As the symbol period becomes significant compared to the time dispersion of the wireless channel, ISI becomes a particularly significant factor in the degrading of performance. The Maximum Likelihood metric provides optimal reception of data with ISI, but is relatively complex to implement. However where the time dispersion is relatively small, so that there are relatively few multipath components to consider at symbolspaced delays, the quantities of data to be processed are reduced. This is the case, for example, with short range radio links such as high rate Bluetooth links.

Aspects of the invention will therefore be described with reference to the High Rate Bluetooth link specification, although it should be understood that applications of the invention are not limited to this type of link.

When Space Time Block Codes (STBC) are applied in a multi-antenna environment exposed to ISI, it is common to change the transmission sequence with the inclusion of Zeros, the inversion of the transmitted sequence or the transmission of orthogonal data vectors. Modification of the transmitted sequence is performed in order to cope with the interference produced by the characteristics of the channel.

The generation of longer vectors, as in the case of the time reversal technique, requires the insertion of known bits between transmissions that prevents a non-intrusive implementation of this method to an existent packet format. Furthermore, the equalisation of data at the receiver becomes more complex as the correlation of the received data vectors increases the number of the equaliser taps to a total of 2(q-1) + 1 where q is the number of channel taps.

A general problem with the above systems is thus the complexity of the equalisation of data at the receiver. It is therefore desirable to combine the estimated symbols at the receiver in such a way that only one equaliser needs to be used when data is transmitted and received by one or more antennas simultaneously.

In accordance with a first aspect of the present invention there is provided a digital communications receiver, comprising: two or more antennas for receiving first data in a first symbol period and second data in a second symbol period; and an equaliser for combining the first data and second data received from the first and second antennas to produce a combined dataset and estimating a transmitted symbol combination giving rise to the combined dataset.

The equaliser preferably comprises estimation means for estimating a trial dataset for each possible transmitted symbol combination, and determining the error between each trial dataset and the combined dataset.

The receiver preferably further comprises a lookup table, coupled to the equaliser, for storing data estimates for all possible transmitted symbol combinations. Different coding or transmission schemes can also be decoded after loading the desired scheme structure into the lookup table.

The most likely transmitted symbol combinations are preferably identified using Maximum Likelihood Sequence Estimation (MLSE). Furthermore, an error between each trial dataset and the combined dataset may be determined, preferably by Minimum Mean-Squared Error (MMSE) estimation.

The system may be upgradeable in software (e.g. the population of the look-up table and combinational procedures) to allow the employment of different coding sequences or number of antennas without the necessity of changing the whole terminal.

The estimation of the received signal enables the use of only one MLSE equaliser, and this assists the exploitation of the diversity generated by ISI in a multi-antenna system.

The combinational method does not imply changes in the Space Time Block Code transmission strategy. This makes it compatible with non-ISI transmissions. No information about the number of channel taps is required at the transmitter.

Parallel Minimum Mean-square Error (MMSE) calculations can be realised instead of having several MLSE equalisers as the number of antennas increases.

In accordance with a second aspect of the present invention there is provided a method of equalising data received by two or more antennas in a digital communications system, the method comprising: receiving first data in a first symbol period at the two or more antennas; receiving second data in a second symbol period at the two or more antennas; combining the first data and second data received from the first and second antennas to produce a combined dataset; estimating a trial dataset comprising estimated first data and estimated second data received at the two or more antennas in the first and second symbol periods for each possible transmitted symbol combination; and identifying the transmitted symbol combinations most likely to produce the combined dataset.

Some preferred embodiments of the invention will now be described by way of example only and with reference to the accompanying drawings, in which: Figure 1 illustrates the packet structure used for the High Rate Bluetooth system; Figure 2 illustrates the structure of a segment within the payload of a Bluetooth( data packet; Figure 3 is a trellis diagram illustrating the effect of Maximum Likelihood Sequence Estimation (MLSE); Figure 4 is a block diagram showing the components of a transmitter and receiver for use with the present invention; Figure 5 is a trellis diagram showing the possible paths available with the receiver of Figure 4; and Figure 6 shows the improvement in error probability obtained by a method in accordance with the invention.

Although the present invention may be used with a variety of different cellular wireless systems, it is primarily devised for the High Rate Bluetooth packet structure, and a brief description of this system is presented. The general format of the high-rate packet is shown in Figure 1. The packet has five fields: a 9-byte preamble field 2, a 2-byte synchronization word 3, an 11-byte header 4, a payload 5 comprising up to 4096 bytes of user data, and a trailer field 6 of 2, 4 or 6 bits. The header 4 contains all the address information for the packet and some additional control information. The payload 5 contains user information, and may be further subdivided into segments 7, 8, 9, 10 if a sufficiently large quantity of data is contained therein.

All the segments and the header have Cyclic Redundancy Codes (CRCs) for error detection. The preamble 2, sync word 3 and the header 4 are modulated using Differential Binary Phase Shift Keying (DBPSK). The modulation format of the payload 5 is indicated in the header 4 and can be DBPSK, DQPSK (Differential Quadrature Phase Shift Keying), (which apply, respectively, rotations r/2 and r/4 when a new symbol is transmitted) or 8-DPSK. The modulation format of the payload 5 is indicated in the header 4 and can be DBPSK, DQPSK, or 8-DPSK. The trailer 6 is modulated in the same format as the payload 5. In the absence of a payload, the trailer 6 is modulated in the same format as the header 4 (DBPSK). Some of the most important fields relevant to the proposed invention are described below.

The preamble 2, which is intended to be used for supporting Antenna Diversity and AGC training, comprises a 72-bit sequence obtained by repeating the following 8-bit sequence nine times: 000011 1 1 The sync word 3 comprises a 16-bit sequence which has a high auto-correlation coefficient. It is used for frame synchronization. The sequence is: 00000010011 10101 The header 4 contains address and control information and has its own 24 bit CRC. The total length of the header, including the CRC, is 88 bits.

The following table lists the header fields, their size in bits, and their meaning. A clarification of each field relevant to this proposal follows the table.

Field Size Meaning

HR_ID High Rate channel ID DP_ADDR 8 Destination Point Address SP_ADDR 8 Source Point Address MOD_TYPE 2 Modulation Type XTD_HDR 2 Extended Header P_L 12 Payload Length in Bytes FEC_ON 1 Forward Error Correction Reserved 1 For future use Flow 1 Flow Control PKT_SAR 2 Packet Segmentation and Re-assembly

_

PLD_MN 1 Payload Message Number ARQMN 1 Arq Message Number ACK_REQ 1 Request Acknowledgement RN 8 Request Number SN_BM 8 Segment Error Bitmap H_CRC 24 Header CRC The HR_ID field (8 bits) is a high rate channel identification field used to distinguish between transmissions of different high-rate networks occupying the same high-rate RF channel. Since a device can only belong to one high-rate network it will only accept packets that have the same fixed HR_ID. A Bluetooth device may comprise several transceivers each associated with a different HR link.

The DP_ADDR field (8 bits) defines a destination point address. Each device participating in a Bluetooth high-rate link may have a number of"logical points". One Bluetooth(E) high-rate device may send information to a specific "logical point" on another device via the basic physical medium. The DP_ADDR field in the packet header indicates which logical point on the receiving device the packet is destined for.

In general, a single high-rate unit will be assigned multiple logical point addresses. The high-rate units are therefore able to accept and receive packets intended for multiple destination point addresses.

The H_CRC filed (24 bits) is the header CRC, a cyclic redundancy check for detecting errors in the header 4.

As mentioned above, the amount of user payload data that can be transmitted in a single packet is between O and 4095 bytes. If the amount of data to be transmitted with in the packet is above 128 bytes then the data is split into one or more segments 7, 8, 9, 10 which are transmitted sequentially within the payload section of the packet. Figure 2 shows the format of a segment 8. Each segment is numbered with a 1 byte (8-bit) sequence number SN 1 1 and has a 3 byte (24-bit) CRC 13 to detect errors. The sequence number 11 and CRC 13 sandwich a user information field 12 containing 128 bytes (1024 bits) per segment. The last segment 10 ofthe payload 5 can be partially filled with user information varying from 1 to 127 bytes.

The operation of Maximum Likelihood Sequence Estimation (MLSE) will now be described.

Consider a transmitter comprising a state machine which produces a sequence of outputs d = {d/, d2,..,dk.,i, where each di represents a transmitted symbol. The spacing between consecutive outputs is known as a symbol period. At any given time k, (which represents one out of a series of discrete times separated by the duration of a symbol period) it is possible to represent all the possible values of d by a vertical vector, within which every point is referred to as a state.

At any given time, a signal sent by a transmitter antenna experiences interference effects introduced by the channel traversed, which includes the transmit chain, the air-link, and the receive chain. The transmitted data is corrupted with noise, so that the signal at the receiver may be considered to be given by r(t) = d(t)h(t)+n(t) where n(t) is an Additive White Gaussian Noise (AWGN) component and d(t)h(t) is the convolution of the generated symbols d(t) with the channel component h(t).

If the transmission bandwidth is much greater than the coherence bandwidth (the bandwidth over which the channel frequency response stays relatively flat) then the channel varies greatly over the bandwidth of the transmission. This "frequency- selective fading channel" increases the probability of error, due to previous symbols interfering with the symbol that is currently being estimated. This effect is called Inter- Symbol Interference (ISI) and produces a received signal r(t) given by the sum of the generated symbols d convolved with the channel components h according to the span of the channel q at time t as shown in equation (1): r(t) = d(, q+,)h(q) + n(t) (I) q Here the "span" q of the channel is the length of the overall channel impulse response measured in symbol periods or, in terms of time, (q - 1)T where T is a symbol period.

For mitigating the ISI effect, an equaliser can be used to correct the multipath distortion.

Maximum Likelihood Sequence Estimation (MLSE) equalisers attempt to make a decision on an entire sequence of symbols. For this to be successful, perfect, or at least reasonably accurate, knowledge of the channel impulse response must be available at the receiver. All possible combinations of the transmit sequence can then be assessed and the process of detection of symbols performed.

In more detail, all possible received sequences are calculated at the receiver and convolved with the estimated channel impulse response. The output of this convolution process is then compared with the actual observation, and the error between the received signal and each of the possible sequences is determined.

The combination of expected data presenting the smallest error with respect to the received sampled signal under analysis is stored, becoming part of the only "survivor path" for this node or state. This process will continue until the end of a predefined length, or alternatively until a whole packet of information has been received. At the end of the process the sequence presenting the smallest error is termed the survivor path, and the sequence of states defined by this path is taken as the best estimate of the transmitted data, and thus becomes the decoded received data. The predetermined length or "truncation depth" after which a decision can be made is taken to be when the survivor paths for all possible states converge. This is generally taken to be a multiple of the channel span, such as between five and ten times the channel span.

The MLSE equalization technique can be graphically represented by a trellis diagram.

Figure 3 shows a trellis diagram for a system in which each symbol is transmitted as one of two possible states, but ISI occurs in the channel so that information from two consecutive symbols is received simultaneously. In the example of Figure 3, all possible variations of received symbols or states (4 in total) 21-24 are mapped in a vertical vector or column 31 -39. In this example, the four states 21 -24 correspond to the received symbol combinations [00], [01], [10] and [11], respectively. Each column 31-39 represents a subsequent sampling point in time, where the left hand column 31 corresponds to the oldest sampled symbol and the right hand column 39 is the most recently sampled symbol. The example of Figure 3 shows an exemplary system with BPSK modulation with 1 symbol ISI.

At each sample time 31-39, the predicted data received is calculated using equation (1) for all possible combinations of symbols transmitted. The received data r(t) depends on more than one consecutively transmitted symbol, and therefore the individual vectors 31-39 will not be entirely independent. A flag is set to mark which sequence of symbols has the lowest enror with respect to the received data. This sequence of symbols is known as the survivor path.

The solid arrows 25 represent the survivor path for one initial state 21 after reception of nine consecutive symbols. For simplicity, the survivor paths for the other three initial states 22-24 are not shown. The dashed lines 26 show the symbols that have been corrected once more information has been received at the receiver, and the dotted lines 27 represent all the possible transitions between states or nodes.

Thus, for example, at the fourth sampling point 34, the path leading to state 21 had the smallest error, although its enror value was similar to that of the path leading to state 22.

On receipt of further information, i.e. on receipt of the symbol for the next sampling point 35, the path was corrected to lead to state 22 at sampling point 34. This can be understood because, in the presence of inter-symbol interference, a received symbol carries information relating to previously received symbols.

In wireless communications, exploitation of the diverse propagation paths between transmitter and receiver helps to mitigate the random characteristics of the channels, since they do not fade equally. To facilitate the presence of more transmission paths, different techniques can be employed by the use of multiple antennas at the transmitter, or at the receiver, or both. One of the most efficient receive diversity techniques in terms of optimum utilization of received signals is Maximal Ratio Receive(r) Combine(r) (MRRC), which permits the coherent addition of the different incoming signals at the receiver after applying the appropriate weighting factor to each antenna branch. The main drawback of this technique is the fact that most of the antennas need to be placed in a receiver. The receiver is usually a mobile terminal where small size, cost and weight are important considerations, and the provision of many receivers is therefore not practical.

For the MRRC strategy, the theoretical probability of errorp(e)MRRC for PSK constellations is given by P(e)M"C = log (M) mat ax! j (2) where mr is the number of receiver antennas, M represents the number of symbols of the transmitter PSK constellation, Eb is the energy per symbol, No is the amplitude of the noise component introduced by the channel and lab is given by fib =log2(M)sin (M) (3) As mentioned earlier, the probability of error for this technique depends on the number of antennas at the receiver only.

When Space Time Block Codes (STBC) are used, the two terminals involved in a communication's process can have the same diversity order, as long as the number of antennas remains equal at both ends. Other schemes are possible, but the diversity order does not remain constant to both terminals. For this system, as for MRRC, the combination of signals is coherent while maintaining the system's throughput (i.e. no coding gain or overhead is added). The system has the considerable advantage that the majority of antennas can be placed at the transmitter or distributed between transmitter and receiver when peer to peer communication takes place.

The principle of STBC is the simultaneous transmission of orthogonal signals from different antennas. At the transmitter a minimum of two antennas is required, while at the receiver any number is possible. The simplest STBC configuration thus involves the use of two transmit antennas. The coding technique for two antennas can be extended to a higher number of transmitter antennas but the transmission matrix will increase proportionally.

In the first time period, two different symbols are transmitted: one from each antenna.

In the immediate following period, both symbols are complex conjugated and one is negated, and both are transmitted from the opposite antenna as in the following matrix representation, where the columns represent information transmitted from the same antenna and the rows represent continuous periods of time.

[Anfl Ant2] aiO aQ t+T - d; d (4) For STBC using PSK modulation, the theoretical probability of errorp(e)STBC for PSK constellations is given by I 2k)( 1-,Ud P(e)STBC = ( ) k=0 k): 4) (5) where m, is the number of antennas at the transmitter and is given by Ed = (6) Simple inspection of equations (3) and (5) reveals that the latter takes into account the number of antennas at both ends. A reduction in size of the antenna array at transmitter and receiver is therefore permitted.

A system in accordance with the present invention involves the combination and parallel calculation of data that has been transmitted and received simultaneously from multiple antennas to one or several antennas using STBC. An example will now be described using an array of two transmit and two receive antennas, although it will be appreciated that different configurations are possible. The process of equalization is described in more detail in British Patent Application No. 0219172.4.

Figure 4 shows a transmitter 41 having two antennas 42, 43 for emitting signals.

Signals pass via one of four channels 44-47 to two receive antennas 48, 49 in a receiver 50. The array of two transmit and two receive antennas is suitable for use with the Bluetooth High Rate base-band model, although it will be appreciated that different cone gurations are also possible.

In the transmitter, data 51 is passed to a Space-Time (S.T.) Mapper 52 which determines the information to go to each transmit antenna 42, 43. A constellation Mapper 53 generates signals based on this information and passes these to the antennas 42, 43, which send STBC M-DPSK symbols over different channels 44, 45, 46, 47. In the example of Figure 4, each channel is assumed to be composed of 3 taps, meaning that the effects of the channel lead to three overlapping symbols within one symbol period at the receiver. This is a consequence of the fact that the system is primarily intended to be used for indoor, short-range transmissions where ISI is present but not usually severe. Excluding the edge effect (beginning and end of transmission), every received sample at one of the receive antennas 48, 49 at a given time will be composed by 3 transmitted symbols convolved with the channel taps.

The data transmitted by the transmit antennas 42, 43 will be in the form described by equation (4) for the first two symbol periods. Subsequently, the antennas will transmit another set of orthogonal signals [Anti Ant2] t+2T1 d2 d31 ( ) Lt+3T] L d3* d2*] At the receiver SO, the received data after two consecutive symbol periods is a 4 component matrix R given bythe convolution of the channel matrix H (four channel vectors, each containing 3 taps) with the transmitted sequence matrix D (four sequence vectors, each containing three symbol combinations) from antenna 1 (A and 2 (2). The sequence vectors are determined from equations (4) and (7), and the sequence matrix-D is equal to the channel matrix in size. Additive White Gaussian Noise (AWGN) 57 is also introduced as the data passes from transmitter to receiver, so noise components N are included in the received signal.

R = HD+N (8) The matrix representation of equation (8) is given by Irk Oh, heard', dt''Tl En, nil Lr3 r4] Lh3 h4]Ld2' d(+)T] Ln3 n4] As mentioned before, after receiving data for two consecutive periods of time, four different values are available at the receiver 50 (two per antenna 48, 49) . With the inclusion of noise components nr, these signals are given by r, = h, ,d2 + h2,d3 - h, 2d, + h2 2do + h, 3do + h2 3d, + nil (10) r2 =-h,, d3 +h2,d2 +h,2d2 +h22d3 -h,3d, +h23do +n2 (11) r3 = h3,d2 +h4,d3 -h32d + h42do +h33do +h43d, +n3 (12) r4 =-h3,d3 +h4,d2 +h32d2 +h42d3 -h33d, +h4, 3do +n4 (13) where in the case of the channel components ELI, L represents the channel number (from Figure 4) and j the corresponding channel tap, where an increment in value represents regression in time. In the case of the data dk, k is the originated data's order before STBC mapping and * is the complex conjugate.

From equations (10) to (13), signals r' and r2 are those received at the first receiver antenna 48 at time t and t+Trespectively, and r3 and r4 are those received at the second receiver antenna 49 at time t and t+T.

Now, considering that this combinational strategy is known at the receiver, the received signals can be estimated 54, 55 using the following formulas: [ Ad (2) ] (14) r2 = [h, h 2: t2,T] (15) r^3 = [h 3 h 4 id t2' (16) r4 = [h 3 h 4 Ad A' ] (17) It will be appreciated that equations (14) to (17) correspond to the matrix representation of the previously shown set of signals in equations (10) to (13), where the h, vectors represent the estimated taps for channel L, and dime are the vectors containing the different combination of symbols (with length equivalent to the channel span) that have possibly been transmitted from the indicated antenna (m) and received at timej or equivalently t and t+T, where t is time and Ta symbol period. A lookup table 56 is filled with the estimated values.

Once the four signals have been received and the lookup table 56 filled with the estimated values, an error e is calculated by the Minimum MeanSquared Error (MMSE) modules as follows: 4 2 e = Mark -rk|| (18) k=l where rk is the received signal and rk is its estimation. This error calculation is performed for all possible received sequences, and the one with the smallest error value is assumed to be the transmitted data sequence.

It will be appreciated that two consecutive received signals will always contain the same symbols on an orthogonal basis. As can be seen from equations (10) to (13), where there are three channel taps, each received signal contains four different symbols (do, d', d2, d3). If the channels had four taps, each received signal would still have four symbols. In general, the number of symbols to be estimated will increase only when a new symbol set needs to be included, which is when the number of channel taps is odd.

A table indicating the number of different symbols per received signal according to the number of channel taps can thus be generated.

Number of Channel taps Number of different symbols 3 4 4 4 6 6 6 Table 2: Correspondence of Channel Taps and Symbol Stages Figure 5 is a conventional representation of the receiver trellis diagram for the system described above where each channel has three taps. The figure shows all of the possible MLSE transitions based on the values which have been stored in the Look Up Table.

The two received signals 62, 63 on the right hand side of the figure both include the symbols because of the STBC transmission characteristics. The previous received signal 61 is part of the previous orthogonal group (i.e. do and do). In other words, the last two received symbols 62, 63 form part of the same orthogonal matrix as equation 7 whereas the previous symbol 61 includes the orthogonal data from equation 4 (which can be confirmed only by data received at time t or and t+T). The receiver must therefore wait for at least two consecutive time periods before equalising.

The employment of this technique allows the use of only one MLSE equaliser at the receiver, with parallel MMSE modules when hardware allows. Instead of using an MLSE for every received stream of data, parallel error calculations can be realised with the help of a lookup table and control module in charge of calculating and loading the data into the parallel MMSE blocks as described in British Patent Application No. 0219172.4.

The response rate of the parallel distributed system will depend directly on the number of MMSE modules available. For instance, if the number of available MMSE blocks is not equal to the number of trellis states, some or all of them would need to be reused.

Figure 6 shows a graph of the probability of error, or Bit Error Rate (BER), against the signal to noise ratio (Eb/No) for a variety of practical and theoretical arrangements of antennas, and demonstrates that full advantage is taken of the energy present in the channel.

In Figure 6, the four dashed lines 71, 72, 73, 74 represent the response of an STBC system, not affected by ISI, having respectively one transmit and one receive antenna, two transmit and two receive antennas, three transmit and three receive antennas, and four transmit and four receive antennas. Thus the line 75 with triangular markers represents the response of a system, not affected by ISI, having two transmit and two receive antennas. This response is equivalent to that of a receiver employing the same diversity order MRRC but with a degradation in performance equivalent to 3 dB, as the same transmission power is assumed. Because two symbols are being transmitted at the same time, it is assumed that the power at the transmitter has been distributed equally between them; and the total transmitted power is equivalent to the power employed to transmit only I symbol where MRRC is employed. So, a reduction in power (of half per symbol), creates a degradation in performance of 3dB.

If the conventional receiver structure (i.e. without MLSE decoder or ISI connection and having two transmit and two receive antennas) is employed for decoding data when ISI fading takes place, then the response has an irreducible error floor having a constant value of Bit Error Rate (BER) = 1 0-t, regardless of any improvement in the signal to noise ratio (Eb/No).

The line 76 having starred markers shows the response obtained using a method in accordance with the invention, even under ISI conditions, for the same receiver structure (two transmit, two receive antennas). It can be seen that this response is equivalent to the response 74 of a system using four transmit and four receive antennas where ISI does not take place. As a result, it is possible to conclude, based on simulations and comparison with other works, that the proposed method gives;a diversity order equivalent to 2(ChL+I) where ChL is the channel length.

Claims (16)

1. CLAIMS: 1. A digital communications receiver, comprising: two or more

   antennas for receiving first data in a first symbol period and second data in a second symbol period; an equaliser for combining the first data and second data received from the first and second antennas to produce a combined dataset and estimating a transmitted symbol combination giving rise to the combined dataset.
2. 2. A receiver as claimed in claim 1, wherein the equaliser comprises estimation means for estimating a trial dataset for each possible transmitted symbol combination, and determining the error between each trial dataset and the combined dataset.
3. 3. A receiver as claimed in claim 1 or 2, further comprising a lookup table, coupled to the equaliser, for storing data estimates for all possible transmitted symbol combinations.
4. 4. A receiver as claimed in claim 1, 2 or 3, wherein the equaliser is arranged to identify the most likely transmitted symbol combinations using Maximum Likelihood Sequence Estimation (MLSE).
5. 5. A receiver as claimed in any preceding claim, wherein the equaliser is arranged to carry out error determination by Minimum Mean-Squared Error (MMSE) estimation.
6. 6. A receiver as claimed in any preceding claim, comprising only one equaliser.
7. 7. A receiver as claimed in any preceding claim, wherein the equaliser is upgradeable in software to enable the estimation of data encoded by different coding sequences.
8. 8. A receiver as claimed in any preceding claim, wherein the equaliser is upgradeable in software to enable the estimation of data received at more than two antennas.
9. 9. A receiver as claimed in claim 7 or 8, wherein the equaliser is upgradeable by changing the population of a look-up table.
10. 10. A method of equalising data received by two or more antennas in a digital communications system, the method comprising: receiving first data in a first symbol period at the two or more antennas; receiving second data in a second symbol period at the two or more antennas; combining the first data and second data received from the first and second antennas to produce a combined dataset; estimating a trial dataset comprising estimated first data and estimated second data received at the two or more antennas in the first and second symbol periods for each possible transmitted symbol combination; and identifying the transmitted symbol combinations most likely to produce the combined dataset.
11. 11. A method as claimed in claim 10, further comprising populating a lookup table with data estimates for all possible transmitted symbol combinations before identifying the transmitted symbol combinations most likely to produce the combined dataset.
12. 12. A method as claimed in claim 10 or 11, wherein the most likely transmitted symbol combinations are identified using Maximum Likelihood Sequence Estimation (MLSE).
13. 13. A method as claimed in claim 10, 11 or 12, further comprising, for each trail dataset, determining an error between that trial dataset and the combined dataset.
14. 14. A method as claimed in claim 13, wherein the error determination is performed by Minimum Mean-Squared Error (MMSE) estimation.
15. 15. A method as claimed in any of claims 10 to 14, comprising receiving data in more than two symbol periods from the two or more antennas before combining the data to form a combined dataset.
16. 16. A method of equalising data substantially as described herein with reference to the accompanying drawings. ( I,

    16. Processor control code to, when running, implement the method of any of claims to 15.

    17. A carrier carrying the processor control code of claim 16.

    18. A method of equalising data substantially as described herein with reference to the accompanying drawings.

    Amendments to the claims have been filed as follows CLAIMS: 1. A digital communications receiver, comprising: two or more antennas for receiving first data in a first symbol period and second data in a second symbol period; an equaliser for combining the first data and second data received from the first and second antennas to produce a combined dataset and estimating a transmitted symbol combination giving rise to the combined dataset, and a lookup table, coupled to the equaliser, for storing data estimates for all possible transmitted symbol combinations.

    2. A receiver as claimed in claim 1, wherein the equaliser comprises estimation means for estimating a trial dataset for each possible transmitted symbol combination, and determining the error between each trial dataset and the combined dataset.

    3. A receiver as claimed in claim 1 or 2, wherein the equaliser is arranged to identify the most likely transmitted symbol combinations using Maximum Likelihood Sequence Estimation (MLSE).

    4. A receiver as claimed in any preceding claim, wherein the equaliser is arranged to carry out error determination by Minimum Mean-Squared Error (MMSE) estimation.

    i 5 A receiver as claimed in any preceding claim, comprising only one equaliser.

    6. A receiver as claimed in any preceding claim, wherein the equaliser is upgradeable in software to enable the estimation of data encoded by different coding sequences.

    7. A receiver as claimed in any preceding claim, wherein the equaliser is upgradeable in software to enable the estimation of data received at more than two antennas.

    8. A receiver as claimed in claim 6 or 7, wherein the equaliser is upgradeable by changing the population of a look-up table.

    9. A method of equalising data received by two or more antennas in a digital communications system, the method comprising: receiving first data in a first symbol period at the two or more antennas; receiving second data in a second symbol period at the two or more antennas combining the first data and second data received from the first and second antennas to produce a combined dataset; estimating a trial dataset comprising estimated first data and estimated second data received at the two or more antennas in the first and second symbol periods for each possible transmitted symbol combination; populating a lookup table with data estimates for all possible transmitted symbol combinations and; identifying the transmitted symbol combinations most likely to produce the combined dataset.

    10. A method as claimed in claim 9, wherein the most likely transmitted symbol combinations are identified using Maximum Likelihood Sequence Estimation (MLSE).

    11. A method as claimed in claim 9 or 10, further comprising, for each trail dataset, determining an error between that trial dataset and the combined dataset.

    12. A method as claimed in claim 11, wherein the error determination is performed by Minimum Mean-Squared Error (MMSE) estimation.

    13. A method as claimed in any of claims 9 to l 2, comprising receiving data in more than two symbol periods from the two or more antennas before combining the data to form a combined dataset.

    14. Processor control code to, when running, implement the method of any of claims 9to 13.

    15. A carrier carrying the processor control code of claim 14.