GB2479549A - Selecting receiver antennas of an interference rejection receiver to provide highest SINR - Google Patents

Abstract

A low complexity antenna selection method 300 for interference rejection receivers that takes interference and noise power contributions received through receiver antennas that have already been selected into account when selecting the next receiver antenna. For each unselected receiver antenna under consideration, a signal to interference plus noise power ratio (SINR) of the receiver is determined when the unselected receiver antenna under consideration is used in combination with all previously selected receiver antennas. The unselected receiver antenna with the highest SINR is selected as the next receiver antenna. Additionally, a loss in SINR may be determined for each of the selected receiver antennas if the receiver were operated using all the selected receiver antennas but excluding the receiver antenna under consideration. The receiver antenna bringing about the smallest loss in SINR may be excluded.

Description

Low complexity Antenna Selection Methods for Interference Rejection Receivers

FIELD OF THE INVENTION

The present invention relates to a method of antenna selection. The present invention in particular relates to a method of selecting antenna for an interference rejection receiver.

BACKGROUND

CONVENTIONAL RECEIVER TECHNOLOGIES

Receiver diversity is a well-known method to improve the signal reception quality in fading channels. In this receiver, signals received over multiple antennae are combined to maximize the signal to noise ratio, depending on the available channel information. When all the channel information of the channels such as phase and amplitude is available, the best receiver is a maximal ratio combining (MRC) receiver (Win, M.Z.; Winters, J.H., "Analysis of hybrid selection/maximal-ratio combining in Rayleigh fading," IEEE Transactions on Communications, vol.47, no.12, pp.1773-1776, Dec 1999).

In cellular systems, the received signal consists of not only the desired signal but also the interference from co-channel users. Figure 1 illustrates an example, where there are two base stations (BS), each serving one user, as indicated by the solid lines in Figure 1. If two users share the same uplink spectrum, the received signal at a base station will include both desired and interference signals. The interference signals are illustrated by dashed lines in Figure 1. The interference signal can be significantly suppressed if multiple receive antennae are used (see, for example, Winters, J., "Optimum Combining in Digital Mobile Radio with Cochannel Interference," IEEE Journal on Selected Areas in Communications, vol.2, no.4, pp. 528-539, Jul 1984). A smart receiver suppressing interference signals in this way is called interference rejection combining (1RC) receiver.

IRC receivers have been successfully implemented in several mobile systems, using different multiplexing technologies, such as GSM-EDGE (Enhanced Data rates for Global Evolution; see Bladsjo, 0.; Furuskar, A.; Javerbring, S.; Larsson, E., "Interference cancellation using antenna diversity for EDGE-enhanced data rates in GSM and TDMA/136," IEEE VTS 50th Vehicular Technology Conference, 1999. VTC 1999 -Fall, vol.4, pp.1956-1960, 1999) and WCDMA (see Astely, D.; Artamo, A., "Uplink spatio-temporal interference rejection combining for WCDMA," IEEE Third Workshop on Signal Processing Advances in Wireless Communications, 2001.

(SPAWC 01), pp.326-329, 2001.).

In the IRC receiver, the received signals from multiple receive antennae are weighted and combined. The weights can be computed according to different criteria, such as to maximize signal-to-interference power ratio (SIR, see EP 1 667 341) or to maximize the signal-to-interference plus noise power ratio (SINR) (see Winters, J., "Optimum Combining in Digital Mobile Radio with Cochannel Interference," IEEE Journal on Selected Areas in Communications, vol.2, no.4, pp. 528-539, Jul 1984. and US200810212666). In terms of performance, the maximization of SIR in EP 1 667 341 is only suboptimal, compared to the maximization of SINR in Winters, J., "Optimum Combining in Digital Mobile Radio with Cochannel Interference," IEEE Journal on Selected Areas in Communications, vol.2, no.4, pp. 528-539, Jul 1984..

The performance of IRC receiver can be improved as the number of receive antennae grows. However, the number of radio frequency (RE) chains at the receivers may be constrained by the cost, power consumption and physical size. If the number of receive antenna N is larger than the number of RF chains K, the receiver will need to select K antennae out of N antennae. The question is how to select antennae in order to optimize the receiver's performance.

Femto-cell base stations find use in increasing the data rate of indoor users.

Being closer to the femto-cell base station, the transmit power of femto-cell users can be much lower compared to that of outdoor users, which are connected to the macro cell base station. Problems arise when a macro-cell user approaches the femto-cell base station. The signal of macro cell user has a higher transmit power and can distort the uplink signal of femto cell user. This is a major problem for femto cell base stations operating in a closed subscriber group operation mode. With multiple receive antennae, the femto BS may employ an interference rejection receiver to suppress the interference from macro users. Nevertheless, with constraints in terms of cost, size and power consumption, the number of available radio frequency modules in a femto BS may be less than the number of receive antennae. In this case, suitable receive antennae ought again be selected.

A MATHEMATICAL MODEL DESCRIBING IRC RECEIVERS

In the following a mathematical model describing ICR receivers is introduced. In Figure 2, a baseband diagram of a communication system is illustrated. The communication system comprises one multi-antenna receiver using an IRC receiver, one desired transmitter/user and M interference transmitters/users. Let h and g(m = 1,2,.. .,M) denote the channel column vectors respectively connecting a desired user and all interference transmitters to the receive antennae available at the receiver.

h[h1 h2... hN]T (1) g,, =[gi,m g2,,,, ... g,]T (2) wherein, as mentioned above, N is the number of available receiver antennae of the receiver. It is assumed that the receiver can estimate all the channel vectors h and g(m=1,2,...,M). Let s and m be the desired and interference data symbols, respectively. The average powers of s and im are normalized to unit.

The received signal vector at the receiver is: yhs+gi+z (3) where z is the thermal noise vector with covariance matrix = E[zz"] = diag(o-o.... where superscript H stands for Hermitian transpose, and diag(x) denotes a diagonal matrix with elements of vector x on the main diagonal. When all the receive antennae have the same noise power, then: (4) wherein K is a K x K identity matrix, and 02 is the noise power.

Assume that the channel vectors are unchanged during the transmission of a data packet. The covariance matrix of noise and interference, R, can be defined: R+g,g (5) where superscripts * and T represent matrix conjugate and transpose operations.

The SINR of each antenna i of the N receiver antennae is: h.12 SINR. (6) o+M1g, The optimal signal combining vector c to be applied to the received signal vector y is provided by Winters in "Optimum Combining in Digital Mobile Radio with Cochannel Interference," IEEE Journal on Selected Areas in Communications, vol. 2, no.4, pp. 528-539, Jul 1984 as being: c=Rh (7) With the combining vector c above, the SINR of the receiver is maximized and can be computed as: p = hTRh* (8) If there are N receive antennae in total and only K antennae are selected for data reception, the optimal selection criterion is to maximize the SINR in equation (8).

To do this, an exhaustive search over = combinations must be K) K!(N-K)! performed. This exhaustive search will be called the optimal selection method in the following.

In the optimal antenna selection, the number of antenna combinations that need to be investigated to find the optimal combination may be too large to be investigated tin practice. For example, when N 8,K, there are 70 combinations of receive antennae to be searched. In each search, one matrix inversion is required. Therefore, the complexity of exhaustive search may be too high for practical hardware.

CONVENTIONAL ANTENNA SELECTION METHODS

One alternative low complexity antenna selection method that is computationally less complex than the above referred to optimal selection method is to choose the antennae with highest SINR. This method is used, for example, in Win, M,Z.; Winters, J.H., "Analysis of hybrid selection/maximal-ratio combining in Rayleigh fading," IEEE Transactions on Communications, vol.47, no.12, pp.1773-1776, Dec 1999. This method will be referred to in the following as conventional method.

In the conventional antenna selection method, the complexity of the conventional antenna selection techniques is very low. The conventional antenna selection method is in fact optimal for MRC receiver (see, Win, M.Z.; Winters, J.H., "Analysis of hybrid selection/maximal-ratio combining in Rayleigh fading," IEEE Transactions on Communications, vol.47, no.12, pp.1773-1776, Dec 1999) but it is no longer optimal for IRC receiver.

Known antenna selection methods thus either suffer from high complexity or poor performance.

SUMMARY OF THE INVENTION

According to an aspect of the present invention there is provided a method of selecting a receiver antenna of an interference rejection receiver comprising a number of receiver antennae that is larger than a number of receiver chains of the receiver.

The method comprises determining for each of more than one unselected receiver antennae under consideration a signal to interference plus noise power ratio of the receiver, hereinafter referred to as SINR. The SINR is individually determined for each of the unselected receiver antennae that is under consideration under the assumption that the unselected antenna is operated in combination with all previously selected receiver antennae. The method also comprises selecting from among the more than one unselected receiver antennae as a next, k-th, receiver antenna the receiver antenna for which the determined SINR is the highest. The present invention thus takes cumulative interference and noise generated by the currently considered antenna in to account when this antenna is operated in combination with already selected other receiver antennae. This method can be determined iteratively, so that a number of receiver antenna is selected in an iterative fashion, selecting one receiver antenna one by one, each time taking the previously selected receiver antennae into consideration.

The receiver may be an interference rejection receiver for receipt of OFOMA signals in which data for a desired user is transmitted using more than one resource blocks. In this case the method can further comprise individually determining, for each of the more than one unselected receiver antennae, said SINR for more than one of the, or for each of the, more than one resource blocks. The SINR of a resource block can, for example be determined by determining the SINR of a single subcarrier within this resource block, if the single subcarrier is considered as being representative of the entire resource block. This is likely the case as discussed in more detail below in the

detailed description.

Knowledge of the SINR of each resource block and for each antenna can be useful in selecting receiver antennae. In one case this knowledge can form the basis for identifying the receiver antenna that, when operated in combination with all of the other previously selected receiver antennae, provides the maximum overall SINR for the receiver. For this purpose the SINR values determined for the various resource blocks can be summed for each antenna and the antenna with the highest sum can be selected as the next antenna.

Knowledge of the SINR of each resource block and for each antenna can moreover facilitate selecting receiver antennae by considering the lowest SINR value each receiver antennae has on a/its worst resource block. The next receiver antenna selected can be the receiver antenna that has the best/highest of these minimum SINR values for the worst resource block. In this case the method further comprises associating with each of the more than one unselected receiver antennae the lowest SINR for a resource block of the receiver antenna. The unselected receiver antenna that has the highest SINR associated with it can then be selected to ensure that the next selected antennae is the one that is most likely to reliably receive data for all resource blocks.

It may not be necessary to consider cumulative interference and noise when the first receiver antenna is selected because no such interferences are known or indeed present when the first receiverantenna is selected. The first receiver antenna can thus be selected based on a different selection criterion, for example so that the SINR generated by the first antenna, when considered on its own, is maximised. Alternative selection criteria for the first receiver antenna are discussed below with reference to OFDMA receivers.

The SINR can be determined for each of the more than one unselected receiver antennae (and in fact for each individual resource block or a representative subcarrier thereof for receivers that are to be used in OFDMA systems having the above discussed properties) by calculating: rD2 wherein: n is the index of the unselected receiver antenna under consideration for which the SINR is calculated; h is matrix quantifying the channel between a desired user arid any previously selected receiver antennae as well as the unselected receiver antenna n; Hui is the norm of n =[a' a... a]T [o_1 a]T wherein a., a1... a are row vectors quantifying the interference and noise power contributions received at antennae i1,i2.. . n of the receiver respectively from all of a number of interfering power sources, wherein G-1 comprises the row vectors of all previously selected receiver antennae; and B = Ak1 -Ak_Ia (i + aflAk_Iafl) aflA/_1 wherein: H denotes a Hermitian transpose; and Ak_I is the matrix B used in determining the SINR of the previous, (k-1)-th, receiver antenna.

As discussed above, the method can be applied so as to determine receiver antennae recursively/iteratively. In doing so, the matrix Ak_2 can automatically determined when selecting the last but second receiver and be retained for selecting the (k-1)-th receiver. This is true for all preceding receiver selection steps. The matrix A0 required for selecting the second receiver antenna can be an identity matrix comprising a number of rows and columns corresponding to the number of interferers.

From the above it is clear that the above described antenna selection method takes the relationship between the channel vector of the desired user and an interference matrix quantifying the interferences received by the available receiver antenna in to account to maximize the SINR of the IRC receiver. This is different from prior art techniques that simple select, as the next receiver antenna the receiver antenna providing the highest SINR, while ignoring the relation between channel vector of the desired user and the correlation matrix of noise plus interference. The above described methods have a much lower complexity compared with the above mentioned optimal exhaustive search. The performance of the above described methods are nevertheless near optimal as the examples will show later.

Ii,, can be normalised by the square root of the noise power of each channel.

Alternatively, the noise power by which is normalised may be considered to be the same for all receiver antenna and/or all resource blocks. This is advantageous if the channel's noise power is low. In this case the SINR of the channel could be unduly emphasised. Such an undue emphasis is avoided by using the same noise power for all channels.

Received interference signals can further be measured at times when no signals are received from a desired user. Based on such a measurement an estimate of a noise and interference power can be generated for each of the receiver antennae based on the measured signals. The row vector a of one or more of the receiver antennae under consideration can be determined based on the estimated noise and interference power and on an estimate of the received noise power.

It may not be possible, to connected each of the receiver antennae to each of the receiver chains. In this case the antenna selection method can further be simplified limiting a set of unselected receiver antennae from which the next receiver antenna is chosen to exclude the one or more receiver antennae that cannot be connected to a receiver chain for which the receiver antenna is chosen. The next receiver antenna is then chosen from the limited set.

The method may further comprise, after all receiver antennae have been selected, determining a receiver combining vector according to: c =(i' _1GBGHIc1)h wherein *denotes a complex conjugate operation, I is a diagonal matrix comprising the noise powers of the selected receiver antennae, G0 is a matrix comprising long its rows the row vectors of all of the selected receiver antennae, wherein Bk is the matrix B calculated when selecting he last receiver antenna, h is a channel vector of desired user comprising channels of selected antennae.

The methods described thus far have operated based on the assumption that no receiver antenna is selected and that receiver antenna are then consecutively added to a set of selected receiver antennae one by one in accordance a selection criterion, as discussed above. The present invention is, however, not limited to this incremental' antenna selection mechanism. According to another aspect of the present invention there is provided a method of selecting a receiver antenna of an interference rejection receiver comprising a set with a number of receiver antennae that is larger than a number of receiver chains of the receiver. The method comprises determining for one or more or each receiver antenna of the set a loss in signal to interference plus noise power ratio of the receiver, hereinafter referred to as SINR, that would be incurred if the receiver were operated using all of the receiver antennae of the set but excluding the receiver antenna under consideration. The receiver antenna that minimally contributes to the SINR, that is the receiver antenna that, if de-selected, leaves a set of selected antennae that has a rnaximised SINR, is then excluded from the set of receiver antennae. Further antenna may be excluded from the set of selected receiver antennae by applying these steps again repetitively until the number of antenna in the set corresponds to the number of available receiver chains.

It may moreover not be desirable to operate the receiver as an interference rejection receiver under all circumstances. It can, for example, be envisaged that the receiver may be operated in a different mode if the interference power experienced by the receiver is acceptably small. This has been recognised as being advantageous in its own right and according to another aspect of the present invention there is provided a method of operating a receiver. The method comprises determining a power of an interference or of interferences received through receive antennae of a receiver and comparing the determined power to a threshold power. If the interference power is above the threshold the receiver is operated as an interference rejection receiver using any of the methods described herein. Otherwise the receiver may be operated in a different operating mode, for example as an MRC receiver.

According to another aspect of the present invention there is provided a computer program product arranged to cause a processor of a receiver to execute any of the methods described herein when executed by the processor.

According to another aspect of the present invention there is provided an interference rejection receiver comprising a number of receiver antennae and a number of receiver chains. The number of receiver chains is smaller than the number of receiver antennae. The receiver comprises a detector that is arranged to determining for each of more than one unselected receiver antennae a signal to interference plus noise power ratio of the receiver, hereinafter referred to as SINR, when the said unselected receiver antenna is used in combination with all previously selected receiver antennae. The receiver is arranged to select from among the more than one unselected receiver antennae as a next, k-th, receiver antenna the receiver antenna with the highest SINR. The receiver can be any type of wireless receiver, such as femtocell base stations, macrocell base stations, mobile devices, mobile terminals, picocell base stations or cognitive radio receivers.

BRIEF DESCRIPTION OF THE DRAWINGS

Embodiments of the present invention will be described in the following by way of example only and with reference to the accompanying drawings, in which: Figure 1 shows a system model comprising two base stations and two user, giving rise to co-channel interference; Figure 2 shows a baseband diagram of a communication system with receivers comprising smart antenna selectors; Figure 3 shows a flowchart of a low complexity algorithm for selecting receiver antennae in an IRC receiver; Figure 4 shows an antenna selection algorithm for selecting antennae for receipt of data of a desired user of an OFDMA system using the MaxSum selection criterion, wherein the user data is transmitted using more than one resource block; Figure 5 shows an antenna selection algorithm for selecting antennae for receipt of data of a desired user of an OFDMA system using the MaxMin selection criterion, wherein the user data is transmitted using more than one resource block; Figure 6 is a flow chart illustrating the operation of an IRC/MRC receiver with antenna selection; Figure 7 compares the performance of IRC rejection receivers using antenna selection methods of described embodiments to select four receiver antennae from among eight available receiver antennae in the presence of one interfering user with the performance of other IRC rejection receivers and MRC receivers; and Figure 8 compares the performance of 1RC rejection receivers using antenna selection methods of described embodiments to select four receiver antennae from among eight available receiver antennae in the presence of two interfering users with the performance of other IRC rejection receivers and MRC receivers.

DETAILED DESCRIPTION OF EMBODIMENTS

AN ALTERNATIVE EXPRESSION FOR SINR

In the following an alternative expression of the signal-to-interference plus noise power ratio (SINR) of interference rejection combining (IRC) receivers will be derived.

This alternative expression will be shown to enable low-complexity computation of the SINR. Following this derivation search methods for selecting antennae to optimise performance of the receiver will be proposed. Let

G=[g1 g2... g] (9) be the channel matrix of interferers. For the purpose of the following discussion it is assumed that the number of antennae N is the same as the number of RF receiver chains K, hence N = K. The SINR expression can be derived as follows: p = hTRh h" (R')' h

M

h (10) =hI+GG')1 h. The matrix inversion lemma states that for a nxn positive definite matrix A and a nxl vector a: (A + aa")' = A' _A'a(l + a"A'a)1 aHA (11) Using the matrix inversion lemma for the term (i + GG11)1 provides: p = h' [i -I'G(IM + GIi1G) 1 G'']h K h2 (12) +G"'G)' GtE'h. k=1 Let

and (13) (14) then hk _H_( Gh. (15) wherein Gmay be considered the noise normalised channel matrix of afl of the interferers and linked to all available receive antennae N. As thus far it has been assumed that the number of available RE receiver chains is equal to the number of available antennae, the fact that equation (15) comprises a sum with K summands is correct. The rows of G may also be considered as quantifying the interference to noise ratio created by all of the M interferers in the N antennae. hmay be considered as quantifying the signal to noise ratio of the normalised channel matrix of the desired transmitter.

The main complexity of computing the SINR based on equation (15) is the need to compute the matrix inversion (M +W' (16) Again, using the matrix inversion lemma (equation (11)), the matrix inversion of equation (16) can be successively computed in the following manner. Let

ak(k=1,2,...,N) (17) be the rows of G. Each row of Gcomprises the noise normalised contributions the M interfering transmitters make to the signals received at one of the antennae. G can hence be expressed as: =[a a... a]T (18) then the matrix inversion of equation (16) required for computing the SINR can be expressed in the following manner: AK =(1M +GG) =(A1 4-aKaK) (19) where AKI = (IM +GK-IGK-1) and (20) GK-I =[ar a... (21) Put in other words, the matrix GK1 is obtained by removing the K -th row of G. Using the matrix inversion lemma (equation (11)) on equation (19) provides: AK (A1 +aKaK) 1 (22) AKI -AK_laZ (i + aKAK IaK) aKAK_l Thus, once the matrix AK_I is known, the matrix AK can be computed without explicit matrix inversion. It is similarly possible to calculate AK_2 from AK_I, and so on.

This means that the calculation can start from A0 = I to compute A1, then compute A2 using A1, and so on, until AK has been calculated. As mentioned above, the computation of Ak(k=1,2,...K) involve the rows of matrix G, which correspond to the receive antenna index k.

Low COMPLEXITY ANTENNA SELECTION ALGORITHM To maximise SINR the receiver antenna indices that maximize SINR p, as defined in equation (12) need to be identified. A low complexity algorithm 300 that can be employed for this purpose is discussed in the following with reference to Figure 3.

In a first step 310, the receive antenna that maximises p in accordance with equation (15) is selected as a first receive antenna and given the index i, using: arg max h 2 (i -aBa') (23) ne{I,2 N) where h is an element of h, B = (A0 + aa)1, A0 M The inversion of matrix B can be computed by matrix update according to equation (22) as: B =(A0 +a,'a)1 M Mafl(1--afl1Mafl)afl1M =M _a1(1+IIantI2)_I a (24) Let Q = {i, 2,.. ., N} , the selection criteria then reads: arg 2 [i -a (IM -a' (1 + IIa 12)_I a ac'] --2 2 hall 1 -argrnaxh 1.-llall +11112] (25) _________ lhl2 argmax -2-=argmax 2 flQ 1+a flC1 o +o hlaIl The firstly selected antenna is consequently the one that has the largest SINR.

Once the first antenna is selected the selection of the second receive antenna can start (step 320 in Figure 3) and equation (15) can again be used to select the second antenna as: 2 H--H' i2 zargmax h -h GBG h (26) necl\,1 where h.. =[h JnIT G =[a a]T. h thus comprises the properties h1of the channel linking the desired user with the already selected antenna i1 as well as the properties h of the channel linking the desired user with the receiver antenna n currently under investigation/consideration. G comprises the row of matrix G relating to the already selected antenna i as well a the rowof matrix G relating to the receiver antenna n currently under investigation/consideration. Moreover, B = (A1 + aa)1 and A1 = (IM + a'a1)'. It will be appreciated that A1 corresponds to the one matrix B used in selecting the first receiver antenna that had the index n of the eventually selected first receiver antenna. This matrix is thus already known and does not have to be computed a second time. The matrices B, h, are determined in step 330 of Figure 3 for all elements of the set Q with the exception of element i1, before equation (26) is used (for k=2) to identify the receiver antenna that provides the maximum SINR, when taking account of signal contributions as well as interference from the already selected receiver antenna i1. A set Co containing the indices of selected antennae can be defined. Before the second antenna has been selected w = {i1}.

Similarly, subsequent, k-th antennae can be selected such that: 2 jj... _-jI_ 1k argmax h -he GBG h (27) ----iT r --iT where the matrix h I h, h1, ... h h I = I d,, I I-J L --again comprises the properties of the channels linking the desired user with the respective ones of the already selected receiver antennae as well as the properties of the channel linking the n desired user with a receiver antenna currently under consideration. The matrix = [ar' aT a]T aT] . . . I 2 7 again comprises the rows of the matrix

G

relating to the already selected receiver antennae as well as the row of the matrix relating to the receiver antenna currently under consideration. The matrix 7 is B = (A;', +a?!a) in conformity with and generalised from the above definedaS 7 -definition used for selecting the second receiver antenna. B can be written as (taking into account equation (22) above): B = (A;'1 + a'a)1 = A, A'1a' (i + aA'1a)' aA'. (28) Again in conformity with the above discussion relating to the selection of the second receiver antenna, Ak_I has been found when previously selecting the antenna with index k-i and consequently does not have to be calculated a further time when selecting the next antenna.

The antenna selection process illustrated by steps 330 and 340 of Figure 3 is repeated until K out of N antennae are selected, that is until one antenna is selected for connection to every available RF receiver chain, in each repetition increasing the index k of the of the receiver antenna to be selected by one (step 350 in Figure 3).

The complete algorithm is summarized in Table 1.

1 input h, G, K, N 2 Initialization: 3 -II2 = [ai' a... a 1T h = 4 i1 "En 6 A0 =M' A1 =A0-A0aj" (1÷a1A0a) a1A0, Q1 =a 7 fork=2:K B B = Ak! -Akla (1 + aflAk iafl) aflAkl Vn E Q 9 Ii,, [d1 i;]T, n[Q1 a]T VnE k =argmaxhn -h GBG h 11 updatea=aUi,=Q\i 12 update dk = h Qk = G,, Ak = B1 13 end 14 output: index set of selected antennas w

TABLE I -ALGORITHM I

Low COMPLEXITY ANTENNA SELECTION ALGORITHM WITH CONSTANT NOISE POWER When selecting receiver antenna by exhaustive search, the noise powers of the receive antennae are needed to compute the best combining vector. In the above discussed search algorithm, the channel coefficients of desired and interference users are in fact normalized by the noise powers of the same antennae. When the noise power is too small, the normalization may increase the value of the normalized channel, which might reduce the accuracy of the digital signal processing circuits. In the following a further antenna selection algorithm is proposed, wherein all the noise powers are set to a constant value can, for example be chosen to 1,. hence I, G = G, and Ii = h. For convenience, the thus modified Algorithm I with unit noise power is called Algorithm II, which is summarised in Table 2.

1 input h, G, K, N 2 Initialization: 3 G=[a a... a]T 4 i1 nc1 a)={i}, c=c\11, d=h, 6 A0 A1 =A0-A0a' (1+a1 A0a') a.A0,Q1 =a1 7 fork=2:K 8 B -Ak1 -AkIa' (i + aflAk 1a) aflAk Vn E un = [di', h:]T [Q1 a]T Vn E Q z=argmax h flEQ 11 update co=a)uik,Q=»=\ik 12 update dk =hik, Qk:GIA, Ak =B1 13 end 14 output: index set of selected antennas a)

TABLE 2-ALGORITHM II

A minimum SINR threshold could be defined, so that only those receiver antennae having a minimum SINR are considered for selection. Limiting the number of available receiver antennae in this manner reduces the complexity of the search determining the receiver antenna maxim ising SINR.

DECREMENTAL SEARCH ALGORITHM

Algorithms I and II select the receiver antennae one-by-one. It is also possible to select the receiver antenna in other ways. It is, for example, possible to initially select all antennae and to then deselect antennae one-by-one such that the remaining antennae yield highest SINR. Algorithms I and II in Tables 1 and 2 can be modified for this selection strategy in the following manner.

Assuming that all receive antennas are selected at the beginning, the SINR is given by equation (15).

= iir _iH(i Hj (29) Then one antenna is removed such that the remained N-i antennas have maximal SINR. Let AN = (IM + H (30) which can be computed by a series of matrix updates using the matrix inversion lemma. Moreover, let B =(IM +_aa)' =(A _a'a)' (31) then, using the matrix inversion lemma again, we have:

B_I H

Nflfl 1 (32) = AN +ANa (1-aflANa aflAN.

Let hc,, be a column vector containing the elements of h, whose indices belong to the set n. Similarly, let Gc1n be the matrix containing the rows of matrix G, whose indices belong to the set Then the SINR of N-i antennas after removing the antenna is = Oh r --ff -(IM + : G0) : (33) -h GQ\B G,, h. The algorithm to decrementaly remove one-by-one antennas is summarized in

Table 3.

1 input h, G, D, K, N 2 Initialization: )={1,2,...,N} 3 G = = [a a... a 1T h 4 AN=(IM+G11G) for k=1:N-K 6 B = AN_k÷l +AN k÷la(1-anAN-.k;Iafl) aflANk+l Vn E 7 k 8 update 9 update AN_k = B end 11 output: index set of selected antennas) Table 3: Algorithm I-A

HARDWARE CONSTRAIN BASED ANTENNA SELECTION ALGORITHMS

A RF chain in a receiver may not be connectable to each and every receiver antenna of the receiver. Instead, an RF chain may only be connectable to a sub-set of N out of the N antennae. In this case, the total number of possible antenna (N N! combinations is reduced to less than I I = . In the following a K) K!(N-K)! modification to above discussed Algorithms I and II implementing this constraint is discussed.

The first antenna can be selected as specified in equation (25). Following this selection, a search for the second antenna to be selected (and for any subsequent antennae) is executed in the above described manner but limited to those remaining antennae that are still connectable to one of the remaining receiver chains.

The above selection process is now illustrated by example. Assume that there are four antennae, antennae (Rxl, Rx2, Rx3, Rx4), and two RF chains. RF chain I can only be connected to antennae Rxland Rx2, while the second RF chain can only be connected to antennae Rx3 and Rx4. Because of this limitation in the possible connections between antennae and RF chains, there are only four possible ways in which the antennae can be selected. These are: (Rxl, Rx3), (Rxl, Rx4), (Rx2, Rx3), and (Rx2, Rx4), and.

As there is no physical limitation on the selection of the first antenna, the first antenna is selected according to equation (25). Suppose that the first selected antenna is antenna Rx3. Since the only physically possible combinations including antenna Rx3 are combinations (Rxl, Rx3) and (Rx2, Rx3), the search for the second antenna can be limited to the set {1,2} consisting of antennae RI and R2.

ANTENNA SELECTION BASED ON THE NOISE AND INTERFERENCE COVARIANCE MATRIX

In some situations, it may be difficult to obtain accurate information about the channel vectors of interferers. However, the covariance matrix of noise interference R can still be obtained by measuring a number of received interference signals over time. For example, the covariance matrix can be computed as: R=Ey,y' (34) where L is the number of received signal vectors at receive antennae when no signal from desired user is transmitted. We can write: R÷(R-)=I+X (35) X is a semi-definite positive matrix and can consequently be decomposed by a singular value decomposition as: X UDU" = UD'12D"2U11 = GGN (36) where U is a unitary matrix and D is a diagonal matrix. Thus we have: R=4-GG" (37) The noise covariance matrix bcan be measured in the absence of interfering signals and equation (37) can then be used to calculate the matrix G for use in Algorithms I and II respectively. This allows selecting receiver antennae even if accurate information regarding the channel vectors of the interferes is not available.

The above decomposition of matrix X is given as an example. Other matrix decompositions can be also employed, although the resulting performances could be different.

ANTENNA SELECTION FOR OFOMA RECEIVERS

The proposed antenna selection algorithms can be also be used in selecting receiver antenna in orthogonal frequency division multiple access (OFDMA) receivers.

In OFDMA systems, user data is carried on resource blocks (RB). Each resource block consists of a number of consecutive subcarriers Nand a number of OFDM symbols N5. In the following antenna selection algorithms for OFDM receivers are described.

First a description of an antenna selection algorithm is provided for the case where data for a desired user is transmitted in a single resource block only. Then a description will follow for the case where the data for the desired user is transmitted in more than one resource block.

ANTENNA SELECTION FOR OFDM SYSTEMS TRANSMITTING USER DATA IN ONE RESOURCE

BLOCK

If the data for a desired user is transmitted in a single resource block the above described antenna selection algorithms I and II can be directly applied for each subcarrier of a resource block. The following method is one of a number of possible solutions.

Because of high correlation, the channels of subcarriers are quite similar.

Therefore, representative channel vectors of one subcarrier may be used for antenna selection purposes. The position of this representative subcarrier within the resource block for the purpose of antenna selection can be arbitrary se'ected. The above described antenna selection Algorithms I and I can then be applied to select those antennae that, in combination and based on the representative channel vectors, provide maximum SINR for the desired user.

ANTENNA SELECTION FOR MULTIPLE RBs If the data of the desired user is transmitted on Nb resource blocks, then for the purpose of antenna selection, one representative subcarrier of each resource block can be employed. The antennae can be selected to satisfy selection criteria. One such selection criterion can be to maximize the sum of SINR of all resource blocks. This selection criterion will be referred to as the MaxSum criterion in the following. Another selection criterion could be to maximise the minimum SINR of all resource blocks. This selection criterion will be referred to as the MaxMin criterion in the following.

THE MAxSuM CRITERION To fulfil this criterion the receive antennae will be selected to maximise the sum of the SINR of all representative subcarrier channels. Algorithms I and II can be modified to accommodate the computation of multiple SINR as summarized in Table 4.

The flowchart of Algorithm Ill is shown in Figure 4. In Algorithm Ill, h1 and G, (j 1* .,Nb) indicate the channel vector and channel matrix of desired user and interferers, respectively, in resource block j.

Referring now to Figure 4 and Table 4, Algorithm Ill starts (step 410) with the initialisation of the channel vector h and matrix G1 for each resource block as well as of the noise covariance matrix (F, the number of available RE receiver chains Kand the number of available antennae N, as set out in row I of Table 3. Set ft comprising the indexes of all unselected receiver antennae, as well as set i, comprising the indexes of all of the resource blocks used for transmitting data for the desired user, are further defined, as indicated in row 2 of Table 5. The noise normalised channel matrix of the interferer G as well as the signal to noise ratio of the normalised channel matrix h1 of the desired transmitter are moreover calculated for each resource block, as indicated in row 3 of Table 5.

In step 410 of Figure 4 and row 4 of Table 4 the receiver antenna that provides the largest sum SINR for all resource blocks is then selected as the first receiver antenna. The receiver antenna fulfilling this selection criterion is identified by applying the modified version of equation (25) shown in row 4 of Table 4. Following the identification/selection of the first receiver antenna i1, the set Q is updated to indicate that the selected receiver antenna i1 is no longer available for selection. A channel vector d1 that comprises the channel coefficients h,,,1 of all the resource blocks in set A of the selected receiver antenna i1 is moreover defined and the matrix A0, needed as an initial matrix aiding the later selection of further receiver antenna in a manner described in more detail below is further initialised to M' wherein M is a MxM identity matrix.

The intermediate quantities required for selecting the next receiver antenna are calculated in step 430 of Algorithm Ill. These quantities include the calculation of the matrix A1 in accordance with equation (22) but individually for all of the resources blocks of the set. The subsequent steps are then performed in the loop indicated by rows 7 to 13 of Table 4 and include calculation of the matrix in conformity with equation (28) but for all thus far unselected receiver antenna, as detailed in the set, and for all Nb resource blocks. The matrix comprises the properties hJ,k-1 of the channel(s) linking the desired user with the already selected antennae as well as the properties hof the channel linking the desired user with the receiver antenna ncurrently under investigation is moreover computed for each thus far unselected antenna n as well as for each resource block (row 9 of Table 4).

The next receiver antenna is selected in step 440 of Figure 4 and row 10 of Table 4, using equation (27) but amended to select the receiver antenna that provides a maximum sum SINR for all of the resource blocks. Subsequent to the selection the sets a and of selected and of thus far unselected receiver antennae respectively are updated to include/exclude the newly selected receiver antenna ik. The channel matrix dfk comprising the channel vectors of all the resource blocks in set i of the selected receiver antennae is moreover updated. The matrices AJk are moreover assigned to hold the matrices B1 for use in the next iteration of the loop. The index k indicating the number of selected receiver antennae is further incremented by one in step 450 of Figure 4 and the loop is terminated if this number is not smaller than the number K of available RF chains, or repeated if k is smaller than K. I input G1 (j =l,...,Nb), b, K, N 2 Initialization: = {1,2,.. .,N} , A {1,2,.. .,P.Jb} 3 = I'2G1 =[a1 a2... aJN], = t112h V3 A 4 i1 w={i1},Q=Q\i1, d1 =h1, (j=l,...,Nb),Ao=IM 6 A1 = A0 -A0a'1 (i + a1A0a11)I a1A0 Q = a11 Vj A 7 fork=2:K 8 = -AJ,k_lafl (1+ af,flAJ,k_laLfl) aJ,flAf,_1 =[,_i 1T J,n =[Q,k_1 an]T (j=l,...,Nb) VnEQ Nb 2 H.-= argmax flE«=2 11 a-couik,Q=)\ik 12 dJk Q,k =G,1,, AJk =B11 VjA 13 end 14 output: index set of selected antennas w TABLe 4-ALGORITHM III When the number of subcarriers per RB is 1, Algorithm Ill reduces to a greedy algorithm in which the selected antennae have maximum SINR of all subcarriers that are being considered. This method is suitable for cases in which the subcarriers allocated to a desired user are not consecutive.

THE MAxMIN CRITERION The MaxSum criterion selects the receiver antennae so that those receiver antennae that maximise SINR across the resource blocks are preferred. It may, however, be that an antenna for which the SINR of an individual resource block is poor, possibly unacceptably so, is nevertheless selected because the SINR achievable for other resource blocks is sufficiently large to compensate for the poor SINR of any individual resource block. The MaxMin selection criterion aims to prefer those receiver antennae that have the best SINR for the resource blocks received at the antenna that exhibits the worst SINR performance out of all of the resource blocks. An algorithm, hereinafter referred to as Algorithm IV, for selecting receiver antennae using the MaxMin criterion is detailed in Table 5 as well as Figure 5.

As can be seen from a comparison of rows I and 2 of Tables 3 and 4 respectively, the initialisation steps in these two rows are identical for both selection criteria. In row 3 of Algorithms lV/Table 5 though, the noise normalised channel matrix G of the interferers is defined in a different manner from the manner used in Algorithm III. In particular, the components of the matrix G1 are defined as row vectors, wherein each row vector comprises the noise normalised contribution ajkm for all interferes m E 1,...,M and for all unselected receiver antennae k E 1,...,N. The matrix Gthus comprises information of the amount of interference each resource block j receives at each individual antenna k from individual interferers m.

In step 520 the first antenna is selected as the antenna that has the maximum SINR for the worst resource block, i.e. for the resource block that has the minimum SINR for the particular antenna. This includes individually determining the resource block jthat has the worst SINR for each of the thus far unselected receiver antennae n. Row 4 of Table 5 provides the required equation for this purpose. There the signal to noise ration hi,,,, of each resource block j for each antenna n is divided by the sum of all interference and noise contributions the M interferers make at the antenna n to the resource blockj. In row 4 of Table 5 the SINR of each individual resource block j is thus calculated for each individual antenna n. The mm operation then identifies the one resource block jWith the lowest SINR for each antennan. Step 4 of Algorithm IV thus provides the minimum SINR value for each antenna in vector 7. The receiver antenna i that has the maximum SI NR value stored in to the vector determined using the max operation in row 5 of Table 5 and selected as the first receiver antenna.

Following the selection of the first receive antenna i1 in the above described manner, the index k is increased for the selection of the second antenna in step 530 of Figure 4 and the intermediate quantities required for selecting the second antenna are determined in step 540 of Figure 4. Rows 6 to 10 of Table 5 detail the steps necessary for this purpose. These steps are the same as those of rows 5 to 9 of Table 4 and will consequently not be discussed in detail at this stage.

Once the intermediate quantities have been determined, the nextlk-th receiver antenna is selected in Step 550 of Figure 4. The steps involved in this process are those detailed in rows 11 and 12 of Table 5. In Row lithe SINR of each resource block j for each antenna n is calculated and the minimum SINR of all the resource blocks j for each antenna is provided as entry for vector y. It will be appreciated that this calculation is performed taking into account the previously selected receiver antennae. In step 12 the receiver antenna iproviding the maximum SINR from among the minimum SINRs of vector y is determined. The antenna 1k providing this maximum SINR is then selected as the k-th antenna. The index k is incremented by one in step 560 of Figure 4 and the loop comprising steps 540 to 570 repeated if it is determined that the index k is smaller than the number K of available receiver chains.

If the loop is repeated then the intermediate quantities are again computed. The required steps are those of rows 13, 14, 9 and 10 of Table 5. These steps correspond to the steps of rows 11, 12, 8 and 9 of Table 5 and will therefore not be discussed in detail at this stage.

1 input h, G.(j=l,...,Nb), F, K, N 2 Initialization: Q:(1,2,...,N} i\ {1,2,...,Nb} -r T T T -i Lf.1 a12... aJNJ h = a1,1 [aJ,k,l aJk2 aJkJ.A] , Q1 = a111, Vj E VkE1,...,N,VmE1,...,M = J, I/i l/( + a1fl,, 12), 11 arg max flEL 7 A = A -A0a (1+ a,1 A0a'11)1 a1 A0, Q1 a11 Vj E 8 fork=2:K 9 B = Af,k_I -AJk_lafl (1+ aJ,flAJ.k_laf,fl) aJ,flAJ,k_1 Ij.n =[d',k_I 1T (j=l,...,Nb) j.n =[Q,kl an]T, vnEc»= 2 -H- 11 = mm h1. ,VnE 12,1=argmaxyfl flE1 13 w=a)L)ikc=c\ik 14 dJk Q.k Alk =B VfEL end 16 output: index set of selected antennas o.

TABLE 5-ALGORITHM IV In one OFDM transmission, several users may occupy some resource blocks of the same OFDM symbol. In this case Algorithms Ill and IV can be employed without modification to select users from among multiple users.

CALCULATION OF RECEIVER COMBINING VECTORS

Low complexity algorithms for selecting receive antennae have been discussed above. The above selection approaches can be exploited for computing receiver combining vectors after the receiver antennae have been selected. To maintain generality, the receiver combining vectors are calculated for all subcarriers of the OFDMA resource blocks. Considering a specific subcarrier, the combining vector can be calculated as follows.

From the above referred to paper by Bladsjo et. al., it is known that: c =(R)'h =(W+GG)hQ 1 (38) _.(i -Iç'G(IM +GI'G) Gc1)h where c' is the conjugate of the desired receiver combing vector and wherein the subscript o. indicates that the matrices concerned consist of the relevant parameters of the selected antennae. The major computational complexity in determining the receiver combining vector c is due to the need to invert the matrix (IM + G4b'G,)'. Using equation (13) provides(IM +G G0)' =(IM +GG) . The matrix inversion lemma of equation (11) can again be used to compute this matrix inversion as in equations (19) and (22). Additionally, for the subcarriers which has been used for selecting antennae, this matrix inversion is the matrix AK obtained in line 12 of Algorithms I and II for a single resource block or line 12 of Algorithm Ill for multiple resource blocks.

Figure 6 illustrates a method 600 incorporating any of the above discussed channel selection methods. After the start 610 of the method the channel properties of the desired user as well as the interference power of the interferers are determined (step 620). In step 630 it is then decided whether or not the interferences are sufficiently large to warrant operation of the receiver as an interference rejection receiver. Should the interference power be above a predetermined threshold, the channel information of the interferers is obtained in step 640 and any one of the above discussed antenna selection processes may be used in step 650 to select the antennae for use as receiver antennae in the interference rejection receiver. Once the antennae have been selected the receiver is set up to operate as an interference rejection receiver in step 660 and data can be received in step 670. Should the interference power be below the predetermined threshold, then the receiver is instead set up to operate as an MRC receiver, as indicated in step 680.

EXAMPLES

Simulation results illustrating the performance of the proposed antenna selection algorithm are discussed in the following. A receiver has been simulated with N =8 receive antennae and K = 4 RE chains. The uncorrelated frequency-flat Rayleigh channel model is employed to generate channel vectors of desired users and interfering users. Each user was simulated to have one transmit antenna. All users employ 4QAM (quadrature amplitude modulation) signals. The signal of each interfering user was simulated to have the same power as that of the desired user. All receive antennae were simulated as having the same noise power.

Figure 7 and 8 illustrate the performance of the proposed antenna selection algorithms for an Interference Rejection Receiver (IRC receiver) for one and two interferers respectively. Both these figures plot the uncoded bit error rate over signal to noise ratio and compare the above mentioned maximal ratio combining receiver (MRC receiver) with an IRC receiver employing random receiver antennae selection (IRC, random selection line), an IRC receiver employing the conventional antenna selection technique described in the background section above (IRC, conventional selection line), IRC receivers employing the antenna selection techniques discussed above as Algorithms I and II (IRC, proposed algorithm I and IRC, proposed algorithm II respectively) and an IRC receiver employing the above discussed computationally expensive optimal selection process (IRC, optimal selection line).

The simulation results indicate that the two MRC receivers fail to achieve full diversity and have a high error floor when an interfering signal is present. The IRC receiver consistently outperform the MRC receivers, with the exception of the IRC receiver using random antenna selection, which has a higher bit error rate than the MRC receiver using optimal antenna selection at low signal to noise ratios.

Both of the IRC receivers using antenna selection methods according to the embodiments of the present invention achieve near optimal performance, in particular at low signal to noise ratios, despite the drastically reduced complexity associated with the antenna selection methods of the embodiments. The antenna selection methods of the embodiments of the invention also significantly outperform the conventional selection method, which relies on the SINR of individual antennae. It is worthwhile to note that antenna selection Algorithm II does not require the knowledge of noise powers. It nevertheless performs slightly better than Algorithm I.

Claims (15)

1. CLAIMS: 1. A method of selecting a receiver antenna of an interference rejection receiver comprising a number of receiver antennae that is larger than a number of receiver chains of the receiver, the method comprising: determining for each of more than one unselected receiver antennae under consideration a signal to interference plus noise power ratio of the receiver, hereinafter referred to as SINR, when the said unselected receiver antenna that is under consideration is used in combination with all previously selected receiver antenna; and selecting from among the more than one unselected receiver antennae as a next, k-th receiver antenna the receiver antenna with the highest SINR.
2. 2. A method according to Claim 1, wherein a first receiver antenna is selected by identifying and selecting the receiver antenna having the highest SINR.
3. 3. A method according to claim 1, wherein the receiver is an interference rejection receiver for receipt of OFDMA signals in which data for a desired user are transmitted in more than one resource blocks, the method further comprising: Individually determining, for each of the more than one unselected receiver antennae, said SINR for more than one of the more than one resource blocks.
4. 4. A method according to Claim 3, further comprising basing the determining of the SINR of each resource block on one or more representative subcarriers of each resource block.
5. 5. A method according to Claim 4, further comprising calculating the sum of the SINRs of all of the more than one resource blocks for each of the more than one unselected receiver antennae, wherein selecting the next receiver antenna comprises selecting as the next receiver the receiver having the largest sum.
6. 6. A method according to Claim 4, further comprising associating with each of the more than one unselected receiver antennae the lowest SINR for a resource block of the receiver antenna, wherein selecting the next receiver antenna comprises selecting as the next receiver antenna the unselected receiver antenna that has the highest SINR associated with it.
7. 7. A method according to Claim 1, wherein said SINR is determined for each of the more than one unselected receiver antennae by calculating: H2 wherein: n is the index of the unselected receiver antenna under consideration for which the SINR is calculated; h is matrix quantifying the channel between a desired user and any previously selected receiver antennae as well as the unselected receiver antenna n under consideration; is the norm of h; iL =[aT a' ... a]T =[ a] wherein a,, a. ... a are row vectors quantifying the interference and noise power contributions received at antennae i1,i2. .. n of the receiver respectively from all of a number of interfering power sources, wherein G_1 comprises the row vectors of all previously selected receiver antennae; and B = Ak1 -AkIa' (i + aflAk_Iafl) aflAk_l wherein: H denotes a Hermitian transpose; and Aklis the matrix B used in determining the SINR of the (k-1)-th selected receiver antenna.
8. 8. A method according to Claim 7, further comprising measuring received interference signals in the absence of a received signal from a desired user, estimating a noise and interference power for each of the receiver antennae based on the measured signals and determining the row vector a of one or more of the receiver antennae based on the estimated noise and interference power and on an estimate of the received noise power.
9. 9. A method according to any preceding claim, wherein the step of determining the SINR comprises setting a noise power value used in determining the SINR to the same value for all of the unselected receiver antennae under consideration.
10. 10. A method according to any preceding claim, wherein a said receiver chain cannot be connected to one or more of the unselected receiver antennae, the method further comprising limiting a set of unselected receiver antennae under consideration to exclude the said one or more receiver antennae and selecting a receiver antenna for connection to the receiver chain from the limited set.
11. 11. A method according to claim 7 or 8, further comprising, after all receiver antennae have been selected, determining a receiver combining vector according to: * _( -1 -I'-i 1-iH -1\j C - -n'tv k"a, wherein *denotes a complex conjugate operation, b is a diagonal matrix comprising the noise powers of the selected receiver antennae, G is a matrix comprising long its rows the row vectors of all of the selected receiver antennae, wherein Bk is the matrix B calculated when selecting he last receiver antenna and wherein h, is a channel vector of a desired user that comprises channels of selected antennae.
12. 12. A method of selecting a receiver antenna of an interference rejection receiver comprising a set with a number of receiver antennae that is larger than a number of receiver chains of the receiver, the method comprising: determining for one or more receiver antenna of the set a loss in signal to interference plus noise power ratio of the receiver, hereinafter referred to as SINR, if the receiver were operated using all the receiver antennae of the set but excluding the receiver antenna under consideration; and excluding the receiver antenna bringing about the smallest loss in SINR from the set of receiver antennae.
13. 13. A method of operating a receiver comprising determining a power of an interference or of interferences received through receive antennae of a receiver, comparing the determined power to a threshold power and, if the interference or interferences exceed the threshold, selecting receiver antennae for operation of the receiver as a interference rejection receiver using a method according to any of the preceding claims.
14. 14. A computer program product arranged to cause a processor of a receiver to execute any of the methods of any of the preceding claims, when executed by the processor.
15. 15. An interference rejection receiver comprising a number of receiver antennae and a number of receiver chains, wherein the number of receiver chains is smaller than the number of receiver antennae, the receiver further comprising: a detector arranged to determining for each of more than one unselected receiver antennae a signal to interference plus noise power ratio of the receiver, hereinafter referred to as SINR, when the said unselected receiver antenna is used in combination with all previously selected receiver antenna; wherein the receiver is arranged to select from among the more than one unselected receiver antennae as a next, k-th, receiver antenna the receiver antenna with the highest SINR.Amendments to the Claims have been filed as follows: CLAIMS: 1. A method of selecting a receiver antenna of an interference rejection receiver comprising a number of receiver antennae that is larger than a number of receiver chains of the receiver, the method comprising: determining for each of more than one unselected receiver antennae under consideration a signal to interference plus noise power ratio of the receiver, hereinafter referred to as SINR, when the said unselected receiver antenna that is under consideration is used only in combination with all previously selected receiver antenna; and selecting from among the more than one unselected receiver antennae as a next, k-th receiver antenna the receiver antenna with the highest SINR.2. A method according to Claim 1, wherein a first receiver antenna is selected by identifying and selecting the receiver antenna having the highest SINR.3. A method according to claim 1, wherein the receiver is an interference rejection receiver for receipt of OFDMA signals in which data for a desired user are transmitted S *S...* in more than one resource blocks, the method further comprising: individually determining, for each of the more than one unselected receiver antennae, said SINR for more than one of the more than one resource blocks.S..... * *4. A method according to Claim 3, further comprising basing the determining of the SINR of each resource block on one or more representative subcarriers of each resource block.5. A method according to Claim 4, further comprising calculating the sum of the SINRs of all of the more than one resource blocks for each of the more than one unselected receiver antennae, wherein selecting the next receiver antenna comprises selecting as the next receiver the receiver having the largest sum.6. A method according to Claim 4, further comprising associating with each of the more than one unselected receiver antennae the lowest SINR for a resource block of the receiver antenna, wherein selecting the next receiver antenna comprises selecting as the next receiver antenna the unselected receiver antenna that has the highest SINR associated with it.7. A method according to Claim 1, wherein said SINR is determined for each of the more than one unselected receiver antennae by calculating: wherein: n is the index of the unselected receiver antenna under consideration for which the SINR is calculated; h is matrix quantifying the channel between a desired user and any previously selected receiver antennae as well as the unselected receiver antenna n under consideration; is the norm of h; r T -IT r r =[a a aT] =1 n-I aT I 2 II * * L wherein a1, a, ... a are row vectors quantifying the interference and noise power contributions received at antennae i1,i2. . . n of the receiver **.i respectively from all of a number of interfering power sources, wherein comprises the row vectors of all previously selected receiver antennae; and B = Ak_I -Akla' (1+aflAk_Ia)' aflAkl wherein: H denotes a Hermitian transpose; and AkIis the matrix B used in determining the SINR of the (k-1)-th selected receiver antenna.8. A method according to Claim 7, further comprising measuring received interference signals in the absence of a received signal from a desired user, estimating a noise and interference power for each of the receiver antennae based on the measured signals and determining the row vector a of one or more of the receiver antennae based on the estimated noise and interference power and on an estimate of the received noise power.9. A method according to any preceding claim, wherein the step of determining the SINR comprises setting a noise power value used in determining the SINR to the same value for all of the unselected receiver antennae under consideration.10. A method according to any preceding claim, wherein a said receiver chain cannot be connected to one or more of the unselected receiver antennae, the method further comprising limiting a set of unselected receiver antennae under consideration to exclude the said one or more receiver antennae and selecting a receiver antenna for connection to the receiver chain from the limited set.11. A method according to claim 7 or 8, further comprising, after all receiver antennae have been selected, determining a receiver combining vector according to: c* (_i 1GJ3GF1Y1)h a.wherein *denotes a complex conjugate operation, I is a diagonal matrix * comprising the noise powers of the selected receiver antennae, G is a matrix comprising long its rows the row vectors of all of the selected receiver antennae, wherein Bk is the matrix B calculated when selecting he last receiver antenna and wherein h,is a channel vector of a desired user that comprises channels of selected *:*. antennae.12. A method of selecting a receiver antenna of an interference rejection receiver comprising a set with a number of receiver antennae that is larger than a number of receiver chains of the receiver, the method comprising: determining for one or more receiver antenna of the set a loss in signal to interference plus noise power ratio of the receiver, hereinafter referred to as SINR, if the receiver were operated using all the receiver antennae of the set but excluding the receiver antenna under consideration; and excluding the receiver antenna bringing about the smallest loss in SINR from the set of receiver antennae.13. A method of operating a receiver comprising determining a power of an interference or of interferences received through receive antennae of a receiver, comparing the determined power to a threshold power and, if the interference or interferences exceed the threshold, selecting receiver antennae for operation of the receiver as a interference rejection receiver using a method according to any of the preceding claims.14. A computer program product arranged to cause a processor of a receiver to execute any of the methods of any of the preceding claims, when executed by the processor.15. An interference rejection receiver comprising a number of receiver antennae and a number of receiver chains, wherein the number of receiver chains is smaller than the number of receiver antennae, the receiver further comprising: a detector arranged to determining for each of more than one unselected receiver antennae a signal to interference plus noise power ratio of the receiver, hereinafter referred to as SINR, when the said unselected receiver antenna is used only in combination with all previously selected receiver antenna; wherein the receiver is arranged to select from among the more than one unselected receiver antennae as a next, k-th, receiver antenna the receiver antenna * .*S..* : with the highest SINR.****.. * S S. 5* * SS S. S S S * S **