GENERATING CHANNEL ESTIMATES IN A RADIO RECEIVER

Abstract

A method and system for generating channel estimates for processing signals received through first and second transmission channels in a wireless communications network, each channel corresponding to a separate transmit antenna, and each signal comprising a plurality of samples derived from symbols transmitted in the signal by: generating first variable z1 (k) and second variable z2 (k); and providing a set of filter coefficients (I) and generating first and second channel estimates using first and second variables and a set of filter coefficients.

Description

TECHNICAL FIELD

The present invention relates to a radio receiver in a wireless communications system, and to a method of generating channel estimates for processing radio signals.

BACKGROUND

The transmission of radio signals carrying data in modern wireless communications can be realized based on a number of different communications systems, often specified by a standard. Mobile radio receiver devices include analog radio frequency (RF)/intermediate frequency (IF) stages which are arranged to receive and transmit wireless signals via one or more antennas. The output of the RF/IF stage is typically converted to baseband, wherein an analog to digital converter (ADC) converts incoming analog signals to digital samples, which are then processed for signal detection and decoding of the data in the form of reliability values. The ADC may alternatively operate directly at IF, in which case the conversion to baseband is performed in the digital domain.

In a wideband CDMA (wideband code division multiple access) cellular system, different physical channels are multiplexed in the code domain using separate spreading sequences called orthogonal variable spreading factor (OVSF) codes. In a case where multiple transmit antennas are used, the same spreading and scrambling codes modulate symbols transmitted from both antennas, but the symbol sequence is different. FIG. 1 is a schematic diagram illustrating two transmit antennas A1, A2. For each antenna A1, A2, a symbol sequence is supplied to a multiplier M1, M2 which multiplies a symbol sequence with a spreading/scrambling code. The transmit antennas A1, A2 are intended to transmit on the same channelization code (in the case of FIG. 1 the CPICH (common pilot channel) downlink code). An adder ADD1, ADD2 allows other channels to be added into the transmission from each of the antennas.

FIG. 1 schematically illustrates dual antenna transmission in the case where the CPICH channel comprises a plurality of pilot symbols, with the data being transmitted using the open-loop transmit diversity (space-time transmit diversity (STTD)) scheme specified by the 3GPP WCDMA system (see, e.g., the 3GPP specification TS 25.211, “Technical Specification Group Radio Access Network; Physical Channels and Mapping of Transport Channels onto Physical Channels (FDD)”, December 2005), or the closed-loop transmit diversity (transmit adaptive array (TxAA)) scheme specified by the 3GGP WCDMA system (see, e.g., the 3GPP specification TS 25.214, “Technical Specification Group Radio Access Network; Physical Layer Procedures (FDD)”, June 2005). The use of multiple transmit antennas requires the estimation of the channel from each transmit antenna at the receiver in the user equipment (UE).

As shown in FIG. 1, for support of channel estimation at the user equipment (UE) receiver, a different symbol sequence is transmitted from each antenna. The modulated CPICH symbol sequence for each transmit antenna is shown in FIG. 2. The symbol S in FIG. 2 is constant, S=(1+j)/√{square root over (2)}. The CPICH spreading factor is 256 and the same spreading and scrambling codes modulate the symbols for both antennas. Antenna A1 transmits the symbol d₁(k) always equal to S,

d ₁(k)=S,

where antenna A2 transmits the symbol d₂(k) equal to +S or −S,

d ₂(k)=ξ(k)·S

where

ξ(k)=(−1)^{└(k+1)/2┘},

and k is the symbol index counted from the CPICH frame boundary.

As can be seen from FIG. 2, the sign verifies the following property,

ξ(2k)+ξ(2k+1)=(−1)^k+(−1)^k+1=0.

FIG. 3 shows the descrambling/despreading circuitry of the CPICH at the UE, for different values of delay applied to the received signal samples corresponding to the taps of the channel impulse response. The circuitry comprises a set of fingers indicated by the reference numeral 2, each for descrambling a delayed version of the received signal. The signal at the output of the CPICH descrambling/despreading corresponding to the l-th delay of the channel, l=1, . . . , L, is

y(l,k)=Sh₁(l,k)+ξ(k)Sh₂(l,k)+n(l,k),

where h₁(l,k) (respectively h₂(l,k)) is the channel gain from antenna A1 (respectively antenna A2) corresponding to the l-th channel delay, and k is the symbol index.

The channel estimation is performed in the same way for each value of delay. For application to the conventional rake receiver processing, the selection of the delays for which the channel estimation is performed is done in a way to capture most of the channel energy. The exact implementation of the delay selection is out of the scope of the present description and is known to a person skilled in the art.

For simplicity, in the following we will omit the delay index, so that the signal after descrambling/despreading is written as

y(k)=Sh₁(k)+ξ(k)Sh₂(k)+n(k).

The estimation of the l-th channel tap is performed separately for the channel from each antenna, h₁(k) and h₂(k).

The problem then is how to exploit best the received pilot symbols for the channel estimation.

We introduce the variables:

$z_1\left(k\right)=\frac{y\left(2k\right)+y\left(2k+1\right)}{2}·S^{*}$ $z_2\left(k\right)=\frac{\xi \left(2k\right)y\left(2k\right)+\xi \left(2k+1\right)y\left(2k+1\right)}{2}·S^{*},$

where asterisk denotes complex conjugate.

For a slowly varying channel h₁(2k+1)≈h₁(2k) and h₂(2k+1)≈h₂(2k), and due to the property ξ(2k)+ξ(2k+1)=0

$z_1\left(k\right)\approx h_1\left(2k\right)+\frac{n\left(2k\right)+n\left(2k+1\right)}{2}·S^{*}.z_2\left(k\right)\approx h_2\left(2k\right)+\frac{\xi \left(2k\right)n\left(2k\right)+\xi \left(2k+1\right)n\left(2k+1\right)}{2}·S^{*}.$

Therefore, for a slowly varying channel, z₁(k) (respectively z₂(k)) is a noisy estimate for the channel h₁(2k) (respectively h₂(2k)).

One approach (e.g., as cited in U.S. patent application Ser. No. 10/139,904, “Transmit Diversity Pilot Processing”, published November 2003) exploits the above property. The channel estimate is performed once every two CPICH symbols

h ₁(2k+1)≈h₁(2k)=f(z₁(k+L₁), . . . , z₁(k+1), z₁(k), z₁(k−1), . . . , z₁(k−L₂))

h ₂(2k+1)≈h₂(2k)=f(z₂(k+L₁), . . . , z₂(k+1), z₂(k), z₂(k−1), . . . , z₂(k−L₂)),

where (•) is a filtering function, with L₁ and L₂ the length of the anti-causal and causal parts of the filter response, respectively. For an infinite impulse response (IIR) filter, the length of the causal part is infinite, L₂=+∞.

For highly time-varying channels, the approximation h₁(2k+1)≈h₁(2k) and h₂(2k+1)≈h₂(2k) are no more valid, and the above approach causes a degradation of the quality of the estimated channel.

Another method based on the estimation of the sum and the difference of the channels of the two antennas A1 and A2, is proposed in U.S. patent application Ser. No. 10/139,904, “Transmit Diversity Pilot Processing”, published 6 Nov. 2003. However, the method proposed in this patent does not update the channel estimate for each antenna every CPICH symbol, and leads to an increased complexity if a finite impulse response (FIR) filter is used to improve the channel estimation. As the pattern of the received signal of the sum and the difference of the channels from the two antennas is not uniform, this requires the use of a different FIR filter depending on the position.

SUMMARY

In one aspect there is provided a method of generating channel estimates for processing signals received through first and second transmission channels in a wireless communications network, each channel corresponding to a separate transmit antenna, and each signal comprising a plurality of samples derived from symbols transmitted in the signal. The method comprises: generating a first variable z₁(k) and a second variable z₂(k), the first variable being a sequence of values, each k-th value being a function of the 2k-th and 2k+1-th samples and the complex conjugate of the symbol S, the second variable being a sequence of values, each k-th value being a function of the 2k-th and 2k+1-th samples, the sign of the 2k-th and 2k+1-th symbols transmitted on one of the antennas (e.g. a second one of first and second antennas) and the complex conjugate of the symbol S; and providing a set of filter coefficients {w(l)}_l=−L1^L2 and generating first and second channel estimates using the first and second variables and the set of filter coefficients.

According to another aspect of the invention there is provided circuitry for first and second transmission channels in a receiver of a wireless communications system. The circuitry comprises: a variable generation unit operable to generate first and second variables, the first variable being a sequence of values, each k-th value being a function of the 2k-th and 2k+1-th samples and the complex conjugate of a symbol S, the second variable being a sequence of values, each k-th value being a function of the 2k-th and 2k+1-th samples, the sign of 2k-th and 2k+1-th symbols transmitted on one of the antennas and the complex conjugate of the symbol S; storage means holding a set of filter coefficients; and means for generating first and second channel estimates for the first and second transmission channels using the set of filter coefficients and the first and second variables.

Other aspects provide a wireless receiver utilizing such circuitry, and a computer program product for implementing the method.

BRIEF DESCRIPTION OF THE DRAWINGS

For a better understanding of the present invention and to show how the same may be carried into effect, reference will now be made by way of example to the accompanying drawings in which:

FIG. 1 is a schematic diagram illustrating transmit diversity for pilot channels;

FIG. 2 is a table illustrating transmitted symbols for the pilot channels;

FIG. 3 is a schematic block diagram of receive circuitry;

FIG. 4A is a schematic block diagram of one embodiment of the invention;

FIG. 4B is a schematic block diagram of another embodiment of the present invention;

FIG. 5 illustrates derivation of new filter coefficients;

FIG. 6 is a graph showing a performance comparison; and

FIG. 7 illustrates a plot of one of the two transmit diversity channels versus the corresponding estimated channel values before filtering.

DETAILED DESCRIPTION

Those skilled in the art to which this application relates will appreciate that other and further additions, deletions, substitutions and modifications may be made to the described embodiments. FIG. 4A is a schematic block diagram of one embodiment of the present invention. Antenna A3 is the receive antenna at the user equipment as in FIG. 3, and reference numeral 2 denotes a single finger. In practice, there will be multiple fingers as indicated in FIG. 3. There will also be multiple samples y(k) for each delay index l, as described above with reference to the prior art, but for the sake of clarity these have been omitted from FIG. 4A and FIG. 4B and from the following description.

The samples are supplied into a variable generation unit 4a which operates to generate two variables z₁(k) and z₂(k). In FIGS. 4A and 4B, these are shown schematically coming from different respective blocks, but this is only a diagrammatic representation. In practice, the variable generation unit 4a could be implemented as a suitably programmed processor, e.g., a processor running individual selectable code sequences. The variable generation unit 4a receives from a store 6 the constant symbol value S and the sign values ξ(k). The symbol value S and the sign values ξ(k) are known because they constitute the known pilot symbols that are transmitted on the CPICH channel as discussed already. The variable generation unit 4a can calculate the complex conjugate S* of the symbol S, or can receive it directly from the store 6.

The variables z₁(k) and z₂(k) are generated according to

$\begin{array}{cc}{z}_1\left(k\right)=\frac{y\left(2k\right)+y\left(2k+1\right)}{2}·S^{*}& \mathrm{Equation}1\\ z_2\left(k\right)=\frac{\xi \left(2k\right)y\left(2k\right)+\xi \left(2k+1\right)y\left(2k+1\right)}{2}·S^{*}.& \mathrm{Equation}2\end{array}$

FIGS. 4A and 4B also illustrates a filter 8 having a plurality of filter coefficients w(l). The filter can be a finite impulse response (FIR) filter, where there is a filter coefficient for each filter tap extending from −L₁ to L₂, with L₁<+∞, L₂<+∞.

In the case of a single transmit antenna (single channel) the filter corresponds for example to a low-pass Wiener filter (i.e., based on Wiener filter theory) used to improve the quality of the single-channel estimation.

With reference to FIG. 4A, the filter coefficients w(l) can be used to generate the channel estimates, by introducing the variables y₁(k), y₂(k) in a variable generation unit 4b

y ₁(2k)=y₁(2k+1)=z₁(k)

y ₂(2k)=y₂(2k+1)=z₂(k).

The channel estimates ĥ₁(k) and ĥ₂(k) are obtained by filtering the variables y₁(k) and y₂(k) in a channel estimation unit 12,

${\hat{h}}_1\left(k\right)=\sum _{l=-L_1}^{L_2}w\left(l\right)y_1\left(k-l\right)$ ${\hat{h}}_2\left(k\right)=\sum _{l=-L_1}^{L_2}w\left(l\right)y_2\left(k-l\right).$

In the following, we also propose low-complexity implementations of this method as illustrated in FIG. 4B. A filter coefficient generation unit 10 receives the filter coefficients w(l) and generates new filter coefficients w₀(l), w₁(l) from the original filter coefficients as shown in FIG. 5,

$\begin{array}{cc}{w}_0\left(\frac{-L_1}{2}\right)=w\left(-L_1\right)+w\left(-L_1+1\right)& w_1\left(\frac{-L_1}{2}\right)=w\left(-L_1\right)\\ ⋮& w_1\left(\frac{-L_1}{2}+1\right)=w\left(-L_1+1\right)+w\left(L_1+2\right)\\ w_0\left(\frac{L_2}{2}-1\right)=w\left(L_2-2\right)+w\left(L_2-1\right)& ⋮\\ w_0\left(\frac{L_2}{2}\right)=w\left(L_2\right)& w_1\left(\frac{L_2}{2}\right)=w\left(L_2-1\right)+w\left(L_2\right)\end{array}$

That is, for the filter coefficients w₀(l), each adjacent pair of filter coefficients w(l) is combined until the final coefficient which is taken as a single value (w₀(L₂/2)=w(L₂)). For the filter coefficients w₁(l), the first value is taken as a single value and then subsequent adjacent pairs of filter coefficients w(l) are combined.

The variables z₁(k), z₂(k) and the new filter coefficients w₀(l), w₁(l) are supplied to a channel estimation unit 12′ which obtains respectively the channel estimates ĥ₁(k) and ĥ₂(k) by filtering z₁(k) and z₂(k) as follows:

$\begin{array}{cc}{\hat{h}}_1\left(2k\right)=\sum _{l=-L_1/2}^{L_2/2}w_0\left(l\right)z_1\left(k-l\right),{\hat{h}}_1\left(2k+1\right)=\sum _{I=-L_1/2}^{L_2/2}w_1\left(l\right)z_1\left(k-l\right)& \mathrm{Equation}3\\ {\hat{h}}_2\left(2k\right)=\sum _{l=-L_1/2}^{L_2/2}w_0\left(l\right)z_2\left(k-l\right),{\hat{h}}_2\left(2k+1\right)=\sum _{l=-L_1/2}^{L_2/2}w_1\left(l\right)z_2\left(k-l\right)& \mathrm{Equation}4\end{array}$

In FIG. 4B, this is shown diagrammatically by the blocks 12a, 12b, 12c, 12d, where 12e is a block which represents the combining of the even (2k) and odd (2k+1) values to generate the final channel estimates ĥ₁, ĥ₂.

FIG. 6 illustrates the effectiveness of the above described method. The figure shows a performance comparison in terms of channel estimation mean square error (MSE) for different CPICH chip energy (CPICH E_c/I_or). The channel is flat fading, the Doppler frequency is 250 Hz and the cell geometry I_or/I_oc=3 dB. The channel estimation is performed using an FIR filter

$w=\left[w\left(-L_1\right)\dots w\left(0\right)\dots w\left(L_2\right)\right]=\frac{1}{9}\left[\begin{array}{ccccccccc}1& 1& 1& 1& 1& 1& 1& 1& 1\end{array}\right],$

with L₁=L₂=4.

The performance of the method described herein is labeled “Proposed Method”. The curve labeled “Alternative Approach” refers to the performance of the method based on the slow varying channel approximations h₁(2k+1)≈h₁(2k) and h₂(2k+1)≈h₂(2k).

For comparison, we also include the case of no transmit diversity, labeled “No Transmit Diversity”, where we account for the difference in CPICH power per antenna compared to the case of transmit diversity by subtracting 3 dB. Also, in this case there is no interference between the pilot patterns transmitted from the two antennas, and therefore the resulting curve corresponds to an upper bound on the achievable performance.

The results shows that the proposed method achieves the optimal performance of the upper bound, and gives a MSE gain of 1 dB for CPICH E_c/I_or=−10 dB.

There follows a description explaining the basis for the proposed method.

Consider the new sequences y₁(k) and y₂(k) in which we replicate z₁(k) and z₂(k) for odd and even symbols indices,

y ₁(2k)=y₁(2k+1)=z₁(k)

y ₂(2k)=y₂(2k+1)=z₂(k).

FIG. 7 shows an example of the actual channel h₁(k) and its estimate y₁(k). From the plot it is clear that, although y₁(2k) and y₁(2k+1) are equal, they correspond to estimates of the channel at instants 2k and 2k+1 (h₁(2k) and h₁(2k+1)), and are affected by different noise terms:

$y_1\left(2k\right)=h_1\left(2k\right)+\underset{\underset{{\mathrm{noise}}_1\left(2k\right)}{}}{\left\{\begin{array}{c}\frac{h_1\left(2k+1\right)-h_1\left(2k\right)}{2}+\\ \frac{\mathrm{sign}\left(2k\right)h_2\left(2k\right)+\mathrm{sign}\left(2k+1\right)}{2}+\\ \frac{n\left(2k\right)+n\left(2k+1\right)}{2}·S^{*}\end{array}\right\}}$ $y_1\left(2k+1\right)=h_1\left(2k+1\right)+\underset{\underset{{\mathrm{noise}}_1\left(2k+1\right)}{}}{\left\{\begin{array}{c}\frac{h_1\left(2k+1\right)-h_1\left(2k\right)}{2}+\\ \frac{\mathrm{sign}\left(2k\right)h_2\left(2k\right)+\mathrm{sign}\left(2k+1\right)h_2\left(2k+1\right)}{2}+\\ \frac{n\left(2k\right)+n\left(2k+1\right)}{2}·S^{*}\end{array}\right\}}$

The same applies to y₂(2k) and y₂(2k+1):

y ₂(2k)=h₂(2k)+noise₂(2k)

y ₂(2k+1)=h₂(2k+1)+noise₂(2k+1).

From this perspective, the inventors noticed that the channel estimation performance can be improved by filtering the sequences y₁(2k) and y₁(2k+1) to generate the channel estimate, rather than applying the filtering to z₁(k) and z₂(k) and using the approximations h₁(2k+1)≈h₁(2k) and h₂(2k+1) h₂(2k) as done in the old approaches.

The inventors noted that the channel estimation using the CPICH with transmit diversity can be obtained, for each channel delay l, l=1, . . . , L, according to the following procedure.

1. Calculate z₁(k) and z₂(k) according to

$z_1\left(k\right)=\frac{y\left(2k\right)+y\left(2k+1\right)}{2}·S^{*}$ $z_2\left(k\right)=\frac{\xi \left(2k\right)y\left(2k\right)+\xi \left(2k+1\right)y\left(2k+1\right)}{2}·S^{*}$

2. Obtain y₁(k) and y₂(k) from z₁(k) and z₂(k) according to

y ₁(2k)=y₁(2k+1)=z₁(k)

y ₂(2k)=y₂(2k+1)=z₂(k)

3. Obtain the channel estimates ĥ₁(k) and ĥ₂(k) by filtering y₁(k) and y₂(k)

ĥ ₁(k)=f(y₁(k+L₁), . . . , y₁(k+1), y₁(k), y₁(k−1), . . . , y₁(k−L₂))

ĥ ₂(k)=f(y₂(k+L₁), . . . , y₂(k+1), y₂(k), y₂(k−1), . . . , y₂(k−L₂))

Depending on the specific filtering which is used, the inventors noticed that this procedure can be simplified to save complexity. In particular if an FIR filter is used, with coefficients w(l), −L₁≦l≦L₂, L₁<+∞, L₂<+∞, filtering is performed through a simple convolution

$f\left(y_1\left(k+L_1\right),\dots ,y_1\left(k+1\right),y_1\left(k\right),y_1\left(k-1\right),\dots ,y_1\left(k-L_2\right)\right)=\sum _{l=-L_1}^{L_2}w\left(l\right)y_1\left(k-1\right)$ $f\left(y_2\left(k+L_1\right),\dots ,y_2\left(k+1\right),y_2\left(k\right),y_2\left(k-1\right),\dots ,y_2\left(k-L_2\right)\right)=\sum _{l=-L_1}^{L_2}w\left(l\right)y_2\left(k-1\right).$

For simplicity, we assume that both L₁ and L₂ are even, and if this is not the case we can append zeros at both ends.

We define two new sets of filter coefficients {w₀(l)} and {w₁(l)}, derived from {w(l)}_l=−L1^L2 as

$\begin{array}{cc}{w}_0\left(\frac{-L_1}{2}\right)=w\left(-L_1\right)+w\left(-L_1+1\right)& w_1\left(\frac{-L_1}{2}\right)=w\left(-L_1\right)\\ ⋮& w_1\left(\frac{-L_1}{2}+1\right)=w\left(-L_1+1\right)+w\left(L_1+2\right)\\ w_0\left(\frac{L_2}{2}-1\right)=w\left(L_2-2\right)+w\left(L_2-1\right)& ⋮\\ w_0\left(\frac{L_2}{2}\right)=w\left(L_2\right)& w_1\left(\frac{L_2}{2}\right)=w\left(L_2-1\right)+w\left(L_2\right)\end{array}$

It is now sufficient to observe that

$f\left(y_1\left(2k+L_1\right),\dots ,y_1\left(2k+1\right),y_1\left(2k\right),y_1\left(2k-1\right),\dots ,y_1\left(2k-L_2\right)\right)=\sum _{l=-L_1/2}^{L_2}w\left(l\right)z_1\left(k-1\right)$ $f\left(y_1\left(2k+1+L_1\right),\dots ,y_1\left(2k+1\right),y_1\left(2k\right),y_1\left(2k-1\right),\dots ,y_1\left(2k+1-L_2\right)\right)=\sum _{l=-L_1/2}^{L_2}w\left(l\right)z_2\left(k-1\right).$

This new implementation of the FIR filtering requires half the number of multiplications and additions than previously.

The above-described embodiment of the present invention has the additional advantage that in the case of a single transmit antenna, the filter does not have to be replaced but can be used with its existing filter coefficients w(l). This has an advantage in that a common filter can be used in the case of single transmit antenna and multiple transmit antennas.

Claims

1.-14. (canceled)

15. A method of generating channel estimates for processing signals received through first and second transmission channels in a wireless communications network, each channel corresponding to a separate transmit antenna, and each signal comprising a plurality of samples derived from symbols transmitted in the signal, the method comprising: generating a first variable z1(k) and a second variable z2(k), the first variable being a sequence of values, each k-th value being a function of the 2k-th and 2k+1-th samples and the complex conjugate of a symbol S, the second variable being a sequence of values each k-th value being a function of the 2k-th and 2k+1-th samples, the sign of the 2k-th and 2k+1-th symbols transmitted on one of the antennas and the complex conjugate of the symbol S; andproviding a set of filter coefficients {w(l)}l=−L1L2 and generating first and second channel estimates using the first and second variables and the set of filter coefficients.

16. A method according to claim 15, comprising generating a third variable y1(k) and a fourth variable y2(k) by repeating in time the first variable z1(k) and the second variable z2(k), respectively; and generating the channel estimates by applying a single set of filter coefficients {w(l)}l=−L1L2 to the fourth and third variables.

17. A method according to claim 15, comprising providing a first set of filter coefficients {w0(l)} and a second set of filter coefficients {w1(l)} derived from a single set of filter coefficients {w(l)}l=−L1L2 by combining respectively adjacent pairs of said single set, the first set commencing from a first value from the single set and the second set commencing from the second value of the single set; and generating the channel estimates by applying the first and second sets of filter coefficients to the first and second variables.

18. A method according to claim 15, comprising the step of selectively using the single set of filter coefficients for generating a channel estimate in the case where a signal is received through one transmission channel from a single antenna.

19. A method according to claim 15, wherein the first and second variables are generated according to:

20. A method according to claim 17, wherein the first and second filter coefficients are provided according to:

21. A method according to claim 17, wherein the step of generating the channel estimates comprises generating even indexes for the channel estimates using the first set of filter coefficients and generating odd indexes for the channel coefficients using the second set of filter coefficients.

22. A method according to claim 20, wherein the channel estimates are generated according to:

23. Circuitry for generating channel estimates for first and second transmission channels in a receiver of a wireless communications system, the circuitry comprising: a variable generation unit operable to generate first and second variables, the first variable being a sequence of values, each k-th value being a function of the 2k-th and 2k+1-th samples and the complex conjugate of a symbol S, the second variable being a sequence of values, each k-th value being a function of the 2k-th and 2k+1-th samples, the sign of 2k-th and 2k+1-th symbols transmitted on one of the antennas and the complex conjugate of the symbol S;storage means holding a set of filter coefficients; andmeans for generating first and second channel estimates for the first and second transmission channels using the set of filter coefficients and the first and second variables.

24. Circuitry according to claim 23, comprising a filter coefficient generation unit for providing first and second sets of filter coefficients from a single set of filter coefficients by combining respectively adjacent pairs of said single set, the first set commencing from a first value from the single set and the second set commencing from the second value of the single set.

25. Circuitry according to claim 23, comprising a further variable generation unit for generating a third variable y1(k) and a fourth variable y2(k) by repeating in time the first variable z1(k) and the second variable z2(k) respectively.

26. A wireless receiver for a communications system, the receiver comprising circuitry according to claim 23.

27. A wireless communications system comprising first and second transmit antennas each arranged to transmit signals via first and second transmission channels; and a wireless receiver according to claim 26.

28. A computer program product for generating channel estimates for processing signals received through first and second transmission channels in a wireless communications network, each channel corresponding to a separate transmit antenna, and each signal comprising a plurality of samples derived from symbols transmitted in the signal, the computer program product comprising program code embodied on a computer readable medium and configured so as when executed on a processor to: generate a first variable z1(k) and a second variable z2(k), the first variable being a sequence of values, each k-th value being a function of the 2k-th and 2k+1-th samples and the complex conjugate of a symbol S, the second variable being a sequence of values each k-th value being a function of the 2k-th and 2k+1-th samples, the sign of the 2k-th and 2k+1-th symbols transmitted on one of the antennas and the complex conjugate of the symbol S; andprovide a set of filter coefficients {w(l)}l=−L1L2 and generating first and second channel estimates using the first and second variables and the set of filter coefficients.

29. A computer program product according to claim 28, wherein the code is configured to: generate a third variable y1(k) and a fourth variable y2(k) by repeating in time the first variable z1(k) and the second variable z2(k), respectively; andgenerate the channel estimates by applying a single set of filter coefficients {w(l)}l=−L1L2 to the fourth and third variables.

30. A computer program product according to claim 28, wherein the code is configured to: provide a first set of filter coefficients {w0(l)} and a second set of filter coefficients {w1(l)} derived from a single set of filter coefficients {w(l)}l=−L1L2 by combining respectively adjacent pairs of said single set, the first set commencing from a first value from the single set and the second set commencing from the second value of the single set; andgenerate the channel estimates by applying the first and second sets of filter coefficients to the first and second variables.

31. A computer program product according to claim 28, wherein the code is configured to selectively use the single set of filter coefficients for generating a channel estimate in the case where a signal is received through one transmission channel from a single antenna.

32. A computer program product according to claim 28, wherein the first and second variables are generated according to:

33. A computer program product according to claim 30, wherein the first and second filter coefficients are provided according to:

34. A computer program product according to claim 30, wherein the generation of the channel estimates comprises generating even indexes for the channel estimates using the first set of filter coefficients and generating odd indexes for the channel coefficients using the second set of filter coefficients.

35. A computer program product according to claim 33, wherein the channel estimates are generated according to: