GB2515570A - Data transmission optimisation method and apparatus - Google Patents

Abstract

The invention applies to a multiple input multiple output (MIMO) transmission system or a system involving a plurality of parallel single input single output (SISO) channels. A transmitter divides an input signal into K streams and spreads each stream using a different spreading code. The streams are then transmitted using multiple transmit antennas. Claim 1 requires that symbols are spread on a symbol by symbol basis every symbol period such that each signal is spread using a respective spreading sequence. One embodiment is concerned with determining, at a receiver, the signal which was transmitted based on the received signal, a channel matrix and the spreading sequences used at the transmitter. This may involve multiplying the received signal with the inverse of a covariance matrix, the Hermitian of the channel matrix and the Hermitian of a normalization matrix. The covariance matrix and the channel normalization matrix are calculated from the channel matrix and a matrix of spreading sequences. In a second embodiment, for each received symbol a system value indicative of a SNR associated with the symbols respective spreading sequence is determined. A data rate for transmission of symbols is determined by averaging the system values of a plurality of symbols.

Description

Data transmission optimisation method and apparatus

Field of Invention

The present invention relates to the field of mobile radio system data transmission.

More specifically, but not exclusively, a method for determining a transmitted signal received at a receiver in a radio transmission system method is disclosed. In addition, a method for determining data transmission rates for transmitting a signal over a radio transmission system is disclosed.

Background to the Invention

Mobile radio system technologies are continuously advancing with the aim of increasing data rates. The third generation mobile radio system uses a code division multiple access (CDMA) transmission scheme and it has been extensively adopted worldwide. The third generation partnership project (3GPP) has developed the high speed down link packet access (HSDPA) system in the Release 5 specification of the Universal Mobile Telecommunications System (UMTS) as a multi-code wide-band CDMA system.

The success of the third generation wireless cellular systems is based largely on the efficient resource allocation scheme used by the HSDPA system to improve the downlink throughput. Multiple parallel channels are used to transmit data symbols.

The downlink throughput optimization for the HSDPA multi-code MIMO CDMA system involves the use of the mean-square-error (MSE) minimizing receiver despreading filter coefficients. The data rate for the symbol transmitted in each channel is adjusted using a measure of the signal-to-noise ratio at the output of the MMSE despreading unit. Once the symbol rate is determined, adaptive modulation and coding methods are used to determine the type of modulation to be used over each parallel channel. The incoming data bits are then converted to the appropriate symbols for the chosen modulation type. Each symbol for each parallel channel is spread using spreading sequences and spread signals are added before being transmitted over the MIMO antennas using spatial multiplexing methods. The total throughput of the HSDPA MIMO system is given by the summation of the data rates transmitted in each parallel channel. Spatial multiplexing methods are then used to maximize the throughput in the HSDPA downlink MIMO transmission.

The number of chip sequences used to spread symbols determines the processing gain, N, for the spread sequence systems. The current standards for the HSDFA system uses a processing gain of 16. For the processing gain of Y=16, the maximum number of orthogonal spreading sequences is 16. When using spatial multiplexing methods the number of spreading sequences is increased to Nxmin(N,,Nr) where N, and Nr aie the number of transmit and receive antennas. The upper bound of the spatial multiplexing gain of the HSDPA MIMO system is referred to as min(A,N.). When the total throughput, R2., for the HSDPA MIMO is normalized with respect to the log (SYR) at high S)VR values the resultant measure is upper bounded by min(N,,Nj, that is tog, SNRj «= min(N,NJ The spatial multiplexing gain for the HSDPA MIMO log. (SNR) system R can be obtained by plotting the total rate against tog, (SNR) and log, (SNR) measuring the slope of the capacity curve at high signal to noise ratios.

An approach was considered by 3GPP and a method was standardized to use a given fixed set size of Orthogonal Variable Spreading Factor (OVSF) spreading sequences for the processing gain of IV=16. To achieve the full multiplexing gain of the HSDPA MIMO system, a signature sequence set size equal to Nxmin(N,N,,) is required. 3GPP standardized a method which increases the OVSF set size by multiplying the given set with precoding weights and then concatenating the weighted sets of spreading sequences to increase the signature sequence set size to Nxmin(N,N).

Hence, a unique pre-coded spreading sequence is produced by concatenating the spreading sequences used at each antenna for each transmission symbol. At the transmitter, each transmission symbol is then spread before transmission with a different channelization spreading sequence at each MIMO antenna These spreading sequences are known as the channelization spreading sequences. Each concatenated channelization spreading sequence is orthogonal to the remaining set of channelization spreading sequences, available at the transmitter for other transmission symbols. In addition to the channelization spreading sequences, a set of scrambling spreading sequences, specific to each downlink MIMO transceiver, is used to further scramble the spread symbols, which have already been spread using the channelization spreading sequences. The scrambling spreading sequence is unique to each HSDPA downlink base-station and is organized to change at every symbol period.

The spreading sequence orthogonality is lost at the receiving end after transmission over frequency selective multipath channels. Therefore, it has been proposed that a linear MMSE equalizer followed by a de-spreader could be used to restore the orthogonality of the spreading sequences at each receiver and to recover the transmitted symbols after transmission over a multipath channel.

Recent developments have considered a self interference (SI) problem, which is present in linear MMSE equalizers, when operating over multipath channels. To improve the transmission throughput it is essential to reduce the large gap between the currently practical achievable rates and the theoretical upper bound for the HSDPA throughput. A receiver with an independent symbol level MMSE equalizer followed by a symbol level successive-interference-cancellation (SIC) scheme deals with the intra-cell self interference. It has been proposed that a hybrid linear equalizer/interference cancelling receiver tailored to the HSDPA standard could be utilized. Furthermore, it has been proposed that a SIC receiver with either a chip or a symbol level MMSE equalizer for the HSDPA downlink throughput optimization is used.

Instead of using a SIC based MMSE equalizer and de-spreader, if only a MMSE based equalizer and de-spreader is used at the MIMO receiver, the spatial multiplexing gain of the HSDPA MIMO can be much smaller than the log, (SIVR) spatial multiplexing upper bound min(N,,Nr).

Introduction of a chip level MMSE linear equalizer followed by a de-spreader and a symbol level SIC is considered to suppress the inter-chip interference (ICI) and also all inter-stream interference. Furthermore, a channel matched filter (CMF) as a linear chip level MMSE equalizer has been shown to maximize the signal-to-noise ratio by collecting the energy at the multipath channel central tap. The chip level equalizer is used to produce an estimate of the transmitted chip sequence, which is then de-spread by one of the transmitter spreading sequences to detect one of the transmitted symbol streams. The recovered symbol is then used to remove the interference iteratively at chip level using a Successive Interference Cancellation j receiver. This work is reported in a paper entitled "Shenoy, Shakti Prasad, Irfan Ghauri, and Dirk TM Slock Optimal Precoding and MMSE receiver designs for WCDMA, In IEEE Vehicular Technology Conference, 2008, VTC Spring 2008, pp 893-897, IEEE 2008 However the throughput performance when expressed in terms of the spatial multiplexing gain could be significantly lower than the log. (SNR) spatial multiplexing upper bound min(Aç,tL) at high signal noise ratios.

The use of a receiver with the linear MMSE equalizer and a single stage SIC detector requires the joint optimization of the transmitter and receiver to improve the spatial multiplexing gain R7. and hence maximize the downlink throughput. Various log, (SNR) transmission power allocation schemes can be used for different data streams for a two stage successive interference cancellation scheme in multi-code MIMO systems.

A two stage SIC detection scheme with the transmitter power optimization can improve the throughput performance for multi-code downlink transmission. However, for each iteration of the SIC, the equalizer coefficient and the power allocation calculations require an inversion of a covariance matrix for the received signal. The dimension of the covariance matrix is usually large and each of the iterative power allocation, the linear MMSE equalizer and the SIC implementations at the receiver therefore become computationally expensive. Simplifications for the inversion of large matrices has been examined to make the implementation of the linear MMSE equalizers followed by the symbol level SIC practically feasible.

It is also imperative to eliminate individual power allocations to each spreading sequence at the transmitter. It is important to use the equal energy and the equal rate allocation method for each symbol before spreading each symbol by using channelization and scrambling codes to maximize the spatial multiplexing gain of the HSDPA MIMO transceiver.

Various attempts have been made to optimize transceiver design. Usually, different optimization criteria are used when allocating powers for the multi-code downlink throughput optimization. Some techniques focus on the transceiver design optimization criteria and others concentrate on criteria for the joint rate and power allocation. One attempt to deal with these problems has been to generalise the joint rate and power adaptation methods as discussed in L. Zhao and J. Mark, Joint rate and power adaptation for radio resource management in uplink wideband code division multiple access systems," lET Communications, vol. 2, no. 4, pp. 562 -572, April 2008, which generalises the joint rate and power adaptation method in three ways as follows: 1. The first criterion includes the systems which optimize the transmission power to maximize the rate for a given realization of channel gains. A typical example is L. Y. Hoon and K. S. Wu, Generalized joint power and rate adaptation in ds-cdma communications over fading channels," IEEE Transactions on Vehicular Technology, vol. 57, no. 1, pp. 603 -608, Jan. 2008 which optimizes the number of symbols and the number of bits per symbol. The aim is to maximize the total rate by iteratively adjusting the transmission powers and spreading sequences whilst satisfying a target signal-to-interference-noise (SINR) ratio at each receiver.

2. The second method aims to maintain the received power at a target level, whilst maximizing the total rate by jointly optimizing the transmission power, rate and signature sequences and also the linear MMSE equalizers at the receiver.

One example of such a method is S. Ulukus and A. Yener, "Iterative transmitter and receiver optimization for cdma networks," IEEE Transactions on Wireless Communications, vol.3, no.6, pp. 1879-1884, Nov. 2004 which jointly optimizes a set of transmission spreading sequences and receivers with linear MMSE equalizers.

3. The third method, one example of which is described in L. Zhao and U. Mark, "Joint rate and power adaptation for radio resource management in uplink wideband code division multiple access systems," lET Communications, vol. 2, no.4, pp. 562 -572, April 2008, uses the average system performance as an evaluation criterion which requires the distribution of the received and the interference signal powers.

In the current HSDPA system specification, an equal energy allocation scheme is used to load each channel with either a single rate or two discrete rates. Parameters of the MMSE receivers are usually optimized using either the max-mm weighted SNR criterion or the total MSE minimization criterion. Recently, an iterative power adaptation method known as the two-group resource allocation scheme has been developed as described in the paper: "The interference-reduced energy loading for multi-code HSDPA systems" , Gurcan MK, Ma I, Ab Ghani H, EURASIP Journal on Wireless Communications and Networking 2012/ 127 March 2012 for 5150 HSDPA system. This method was extended to MIMO HSDPA system in the paper: "System Value based Optimum Spreading Sequence Selection for High Speed Packet Access (HSDPA) MIMO" , Gurcan MK, Ma I, Chungtragarn A, Joudeh H, EURASIP Journal on Wireless Communications and Networking, 2013, 20131:74. Mar 2013. Imperial Innovations Ltd have filed the following patent applications related to these two papers: UK patent applications GB1207546.1 and GB1115566.0, and International patent applications PCT/GB20131000185 and WO20131034875.

HSDPA systems have channel throughputs, which are usually plotted as a function of the channel signal to interference plus noise ratio. Usually HSDPA transmissions are over frequency selective multipath channels causing Inter Symbol Interference (ISI) and there are typical channels specified by the standardization organizations.

Two well known specified channels are the Pedestrian A and the Pedestrian B channels.

The Pedestrian A channels cause small 151 for the HSDPA system, and the Pedestrian B channels cause severe ISI for the HSDPA system.

For the Pedestrian B channels with severe ISI the existing practical systems achieve a performance below the theoretical performance. The patents filed by Imperial Innovations Ltd referred to above have the objectives to control the number and the ordering of signature sequences to achieve a throughput upper bound close to the theoretical throughput upper bound for the HSDPA receiver for the Pedestrian B channels with severe ISI when using: * successive interference cancellation receivers; and * deterministic signature sequences without scrambling codes.

The paper: "System Value based Optimum Spreading Sequence Selection for High Speed Packet Access (HSDPA) MIMO" , Gurcan MK, Ma I, Chungtragarn A, Joudeh H, EURASIP Journal on Wireless Communications and Networking, 2013, 20131:74.

Mar 2013 describes a MIMO HSDPA transceivers systems, whereinthese transceivers operate with the channelization codes as the signature sequences.

These schemes have not used scrambling codes as part of the signature sequences.

The system disclosed in this paper achieves a much improved throughput at high signal to noise ratios by bringing the spatial multiplexing gain RT close to the log, (SNR) spatial multiplexing upper bound min(N, N) A measurement known as the System Value measurement is used to control the rate and the power allocation when using only channelization codes as the spreading sequences. An equal SNR allocation scheme for the rate and the power allocation is considered. The System Value based allocation delivers impressive results providing throughput close to the theoretical upper bound. However, since the method requires the channel side information or the data rate for each channel to be transmitted from the receiver MIMO to the transmitter MIMO further optomisation could be possible.

Summary of Invention

In accordance with an aspect of the invention there is provided a method for determining a transmitted signal received at a receiver in a radio transmission system having a plurality of parallel single-input single-output or multiple-input multiple-output channels over which the signal is transmitted, the signal comprising one or more symbols, the symbols being spread prior to transmission by a plurality of spreading sequences on a symbol-by-symbol basis every symbol period p. The method comprises receiving a signal r10, (p)) corresponding to a signal transmitted over a channel, wherein each symbol of the transmitted signal is spread using a respective spreading sequence S(p). The method also comprises determining an estimated channel characteristic H indicative of effects of the channel over which the received signal r,rg(p)has been transmitted on the transmitted signal. Furthermore, the method comprises determining the transmitted signal r,ort(p)from the received signal r10 (p)in accordance with the estimated channel characteristic H and the spreading sequences used to spread the symbols.

The method may further comprise determining a combined matrix Z of all spreading sequences s(p)used to spread the symbols of the transmitted signal. The method may further comprise determining a covariance matrix C and a normalization matrix C in accordance with the estimated channel characteristic H, and a multiplication of the combined matrix of the spreading sequences Z and the hermitian of the combined matrix of the spreading sequences Z'. The transmitted signal may be determined from the received signal rb,,g(p) by multiplying the received signal r, (p)with the inverse of the covariance matrix C, the hermitian of the channel matrix H, and the hermitian of the normalization matrix C. The covariance matrix C may be defined as follows: = () HZZHHH + H1Z IZIH + H,ZPIZIH + NONK wherein ET is a total transmission energy, N = N3 is a total number of MIMO transmission antennas, 7'/3K is a total number of spreading sequences, 7V is a total number of transmitted symbols used to calculate the covariance matrix C. N0 is a noise power spectral density, N(N*L_l) is an identity matrix, the previous symbol period channel interference matrix H is produced from the estimated channel matrix H using: = (i o('f)n c and wherein the channel matrix for the next symbol period is produced using: H, = (INr ® EE wherein the (iY+L-1)x(N+L-1)-dimensional matrix is given as = L 2) 0 and also the operation JN refers to the Wh power of (iv*i-:) Q( if 2) N*L_l further the operator 0 corresponds to Kroneker product, ZZ" are transmitted signature sequence matrices and and also are the transmitted signature sequence matrices for the current previous and next symbol periods.

The covariance matrices of the transmitted signature sequence matrices ZZ and Z, 1Z and for the current, previous and next symbol periods may be determined according to the following: z = [s(2..., s(p), s(N') + = [so),..., s(p),. s(rv)j Z1 = [S(3),.-.,S(p),.. .S(NM + 2)j wherein a spreading sequence matrix S(p) of all spreading symbols for each symbol period p is determined accoiding to: S(p) = duig (p))SCPW.

wherein Shw is an extended channelization matlix and s(p)is a scrambling sequence vector for symbol peiiod p wheiein the scrambling sequence vector (p) is determined according to: = (d/ag(I)® s3) = (n),. .(N(x) + 2)1 and wherein the extended channelization matrix is determined accoiding to: SCh. = ( ®SChK) [Iwc;,i "wch.fr w:h.K I wherein a total of K channelization spreading sequences SUkK are generated from a total of N Hadamaid codes with spieading sequence processing gain N, and a signature sequence matrix s s Ni wherein a channelization spreading sequence vector s assocaited with the signature sequence matrix S has the property: 1 i=j

H = o

and wherein a transmitter precoding matrix W is determined according to: W/ = J±',7.WN, J wherein the precoding vectors n', forn, =1,...,N with the property: =1

H o

are the beam stearing vectors used at the transmitter antennas.

For a Successive Interference Cancellation, SIC, receiver the signature sequence covariance matrices z()z"(), z 1()z j) and zjE)z'1() may be updated using: Z()Z"()= z()z"()-Sk()Sft() kk1 wherein Sk()=ESk1(c1)v) +19 k')y(:) +P9 and z21()z, () = z2 1()z1()-sk(Pl)($k(Pl)(c) Ic-k1 wherein Sk(2 1)()_k_1) y(xJ skH -1) + p-1 and z21()z () = z1 ( )z1 ()-($() k=k1 wherein = Ek _1)N + 2J kL + p +iJI for the updated signature sequence matrices, and c() and c,() are the received signal covariance matrix and the normalization matrix respectively for use in the implemenation of successive interference cancelation receivers and wherein is the epoch number, N and NH are the number of symbols transmitted in a block and a set of the blocks respectively and wherein the covariance matrix c() is determined according to: c()= hTH, + u1z 1()z'1()Hf' + + A0A,KA and the normalization matrix is determined according to: c()= (Diag(Diag(H"c-'ç)H)))' For the Successive Interference Cancellation, SIC, receiver the received long matrix Rk may be iteratively updated for the correctly detected symbol sequences in the channels updated using: R,01 = R,0 -Ek Ic-k1 = [Ckng (i), r, (p),* . £;o (NC))] wherein the long matrix Gkfor the correctly detected symbols is determined according to: = b(1), k' cb(iV)j wherein the kth channel's chip sequence vector for the symbol period p and spreading seqeunce sk(p) is determined according to: _k)L1(+ Hlsfr(p-1)xkfl(p-1)+ H2sk(p+1)xkD(p+1)j for successive interference cancellation receivers.

The transmitted signal may be determined according to the following: = ((-, f10p) wherein C is an inverse of the covariance matrix, r,0(p) is the received signal, and the normalization matrix C. is determined according to: = (Diag(Diag(H"c'H))) 1 The method may further comprise estimating a received symbol vector using the following relationship: j'(p) = SW (Diag (sH (p))f r (p) = wherein rort(p) is the transmitted signal, S. is an extended channelization matrix and s(p)is a scrambling sequence vector for symbol period p wherein the scrambling sequence vector s(p) is determined according to: = (diag(I A)® s5) = p),...s(N(d +2)J and wherein the extended channelization matrix SChW is determined according to: = (\v®sChK)= wherein a total of K channelization spreading sequences SChK are generated from a total of N Hadamard codes with spreading sequence processing gain N, and a signature sequence matrix S = sj wherein a channelization spreading sequence vector Syk associated with the signature sequence matrix S33 has the property: 1 1=1

H

t_ssJt_ss,j 0 i!=j and wherein a transmitter precoding matrix \V is determined according to: W/ = J±',7.WN, J wherein the precoding vectors n', forn, =1,...,N with the property: =1

H

0 z!=j are the beam stearing vectors used at the transmitter antennas.

The estimated channel characteristic may be defined by: jj(11) . . A) JEll 1) wherein the channel convolution matrix H"'' between the transmit antenna n1 and the receive antenna n1 is defined by: 0 0 i r/ ) : [J 2p fl L-1 wherein a channel impulse response vector between a transmit antenna n and a receive antenna r is defined by channel impulse response function: = [h'r'"t... h' The estimated channel characteristic may be a channel impulse response function.

The received signal may be defined by: i,. (p) = Hz(p)+ H1z(p -i) + H,z(p -i)+ n(p) wherein the vector n(p) corresponds to the received noise samples at the output of receiver chip matched filter and z(p) is the transmit chip vector for the current symbol period which is determined according to: z(p) ETS(p)y(p) wherein z(p -i) is the transmit chip vector for the previous symbol causing intersymbol interference which is determined according to: z(p -i) ETS(p1)y(p -i) wherein the transmitted chip vector z(p +i) for the next (p//T +i) symbol period is: z(p +1) = S(p + i)y(p +1) wherein p)= [yI(p),,yk(p),,yK(p)tis the transmitted symbols vector which contains the symbols, over the symbol period p = 1,.. , wherein NIX) is the number of transmitted symbols in a transmission block wherein the entire block of transmission is represented as an (N x N1K) dimensional transmit symbol matrix using the following relationship: X = [X1, _Xk,***2AK] = [x(1),...

wherein X is the matrix conatining transmitted symbols, Xk is the vector containg symbols over channel k, NK is the total number of spreading sequences used in the downlink MIMO system, and wherein the (Nx1)-dimensional symbol vector is constructed using: = [xk(1),.

for each channel k = 1 iV,K In accordance with a further aspect of the invention there is provided a method for determining data transmission rates for transmitting a signal over a radio transmission system having a pluiality of paiallel single-input single-output oi multiple-input multiple output channels over which the signal is to be transmitted! the signal comprising one or more symbols, the symbols being spread prior to transmission by a plurality of spreading sequences on a symbol-by-symbol basis.

The method comprises receiving a signal comprising a pluiality of symbols, each symbol having been spread by a spreading sequence of a plurality of spreading sequences, determining, for each symbol of the plurality of symbols, a system value Ak indicative of a signal-to-noise iatio associated with the respective spreading sequence used to spread the symbol, and averaging the plurality of system values associated with the plurality of spreading sequences to determine an averaged system value for determining a data rate for transmission of symbols using the plurality of spieading sequences.

The method may fuithei comprise determining a data rate to be associated with the number of signature sequences for transmission in accordance with the average system value.

The data rate for each stream may be determined by finding the discrete data rate value which satisfies the following equation: [1] «= J)< [A] wherein 2(nl)K] (4) is the system value, which is for a streaming group number n.

and for symbol epoch and a target system value [A*], , which is determined I 1 F(2°-1) according to [2 = for p = I,..., P for a chosen discrete bit rate of -1(2 " -1) wherein F is the gap value.

The method may further comprise estimating the symbols in the received signal in accordance with the following equation: :(p)= s. (Djag(s(p))fr(p) Wherein orj(p) is the transmitted signal, is an extended channelization matrix and s(p)is a scrambling sequence vector for symbol period p wherein the scrambling sequence vector s(p) is determined according to: s. = (ciiagi)® Sy:) = [. (i), (n). s (y(x) + 4J and wherein the extended channelization matrix Shw is determined according to: = (t ® Sd,K)-[L?chI' "vvch.k' £i*vh.K] wherein a total of K channelization spreading sequences SK are generated from a total of N Hadamard codes with spreading sequence processing gain N, and a signature sequence matrix s j wherein a channelization spreading sequence vector s associated with the signature sequence matrix S has the property: 1 1=1

II =

0 i!=j and wherein a transmitter precoding matrix W is determined according to: w = 12 Ji?N wherein the precoding vectors w forn, ,N. with the properly: =1 II j L° i!=i are the beam stearing vectors used at the transmitter antennas.

The system value may be defined by: k ltk = Trace(Si'DZZ"D"S,) (N) (1-) -Trace(S7DZ'iZ D'S,) /() pVC) I'- -Trace(s"DZ" Z IJVS) N,IcVN k p11 p1 -Jr -N0 Trace(BB") wherein D is the estimated channel characteristic converted to a squaie inatlix representative of intersymbol interference, which is defined as follows: D=C HHC H-dtag(C HRC LH) and a signature sequence matrix for channel k is defined as: Sk = dzag( 1.kc = wherein an intersymbol matrix D1 for the interference from the previous symbol is determined according to: D1 = (cf' H"C IH) and an intersymbol interference matrix D, for the interference from the next symbol is determined according to: = (c H"C 114 and B is a noise vector despreading matrix, which is determined according to: B=SCU H"C.

The plurality of spreading sequences with which the average system value is associated may correspond to a first combination of receiver and antenna channels, wherein the method further comprises determining an average system value for one or more other combinations of receiver and antenna channels, and using the receiver and antenna channel having the highest system value for transmission.

Any of the different methods disclosed herein may be combined with one another, where appropriate, to form a combined process. For example, the method for determining a transmifted signal received at a receiver may be combined with the method for determining data rates, and vice versa.

Also disclosed is an apparatus arranged to perform any of the methods disclosed herein. The apparatus may be a receiver device. The apparatus may be a radio transmission system comprising one or more transmitters and one or more receivers.

Also disclosed is a computer program product arranged, in use, to instruct a computer to perform any method disclosed herein.

MMSE linear equalisers for MIMO transceivers with signature sequences constructed using scrambling and channelization codes are disclosed. Furthermore, a system model for the HSDPA MIMO transceivers which use signature sequences constructed using channelization and scrambling codes is discussed herein.

The System Value (SV) based rate allocation process may be used to allocate equal energy, equal SNR and equal rate for HSDPA MIMO downlink transmissions for a single user case. In the downlink transmission both the channelization and the scrambling codes are considered. Although the rate allocations may be based on the extended System Value calculations, known covariance matrix calculations may not be applicable to the system value calculations disclosed herein.

Covariance matrix calculations are disclosed herein to deal with spreading sequences which incorporate both channelization and scrambling codes.

A MIMO transmitter and receiver and multiple spreading sequences when operating with spreading sequences which change every symbol period when transmitting over frequency selective channels is disclosed.

A method for data transmission for a mobile radio system having a single input and single output channel or a plurality of Mulitple-lnput-Multiple-Output channels with a total number of N, transmit and)Vr receive antennas where N, is the minimum of N, and N., over which data is transmitted, data is represented by a plurality of data symbols, data symbols being spread prior to transmission by a plurality of spreading sequences. The method comprises determining a system value Ak for each signature sequence k of plurality of signature sequences AT.K, wherein the total number K of orthogonal signature sequences is the base number for spreading sequences and the system value A is indicative of a signal-to-noise ratio of the associated signature sequence constructed using a combination of channelizing and scrambling sequences wherein the method determines the number K of signature sequences to be used for spreading the data symbols in accordance with the system values A associated with a plurality of signature sequences N,K wherein the number of signature sequences selected corresponds to the determined number of signature sequences K* and spreading the data symbols using the selected signature sequences matrix S wherein the selected number spreading sequences are used to spread data symbols to be transmitted over a subset of transmitter antennas selected from N, transmitter antennas if N, is equal to N, alternatively the number of selected signature sequences are transmitted from N, transmitter antennas and at the receiver the system values 2k associated with a plurality of signature sequences N,K are used to select a group of N,. receiver antennas out of N, N1 receiver antennas if N, is equal to N,, the signals at the output of MIMO N1 receivers are combined to maximize the total signal to noise ratio and hence the total transmission data rate.

A number of non-deterministic channelization sequences, each with a set of scrambling sequences, may be into the detection methods, where the system values Ak associated with a plurality of signature sequences N,K are used to produce a measurement method. This System Value (SV) based measurement method provides improvements to the HSDFA throughput without requiring any significant changes to current operational HSDPA systems. The proposed method uses the System Value based measurement method to select a sub-set of MIMO receiver and transmitter antennas to achieve throughput improvement.

A system model for the High Speed Downlink Packet Access (HSDPA) system when using spreading sequences which change at every symbol period is disclosed. This simplifies the system implementation for the HSDPA down link operations which can be described using the system model for the HSDPA downlink transmission for known systems under the assumption that N, = ]V. All N,K spreading sequences are used to transmit data from N, downlink antennas to A/, receiver antennas at a single user receiver.

Initially, the signature sequences may be constructed by combining the scrambling and the channelization codes. These signature sequences may be used to generate transmitted chip sequences for a given number of QAM symbols transmitted over KY, parallel channels. In the MIMO system, a total of N, MIMO receiver antennas may be considered and N may be the processing gain of the spreading sequence.

Channelization codes may be generated using Hadamard OVSF codes as specified by the 3GPP standardization organization. For a spreading sequence processing gain, N, the total number of Hadamard codes is N, the signature sequence matrix S. has the dimension = c The channelization spreading sequence vector s e CN has the property: 1 1=1 If

-o

Atotal of K channelization codes ShK = rh,1c'" rh,Kj ECAXA for K «=N can be extracted from the channelization codes S53. C defines the two dimensional complex space with dimension sizes given as the superscript values. The spreading sequences may be real instead of complex ones.

The number of channelization spreading sequences may be increased to N5K and the length of each spreading sequence may be increased to N11V by using the precoding matrix of W = [w1,...,w, w. je The precoding vectors have the property: 1 i=j

H

o i!=j A total of N3K channelization codes are generated using a precoding matrix to produce an extended channelization signature sequence matrix: = (w' ® ) 31K]E CI\X\SK.

Scrambling sequences may be generated using a long binary sequence of m-sequences or any other data sequence with very low autocorrelation sidelobes. In this part of the system modeling, m-sequences of length 216_I and 2' -I may be used to construct scrambling sequences. The m-sequence vector rn = (i),... , m, (/),..., m, (26 -c C2k elements are tn. E {± i}. The scrambling sequences are used to generate scrambling sequence matrix Nx( N * *t = *.. s (iv + 2)] E C where p symbol period epoch number and NH is the total number of symbols to be transmitted over each channelization code. Each scrambling sequence vector sjp)= [s31(p), s3(p),.. *s5. (p) may have complex elements constructed using the m-sequence elements: ,n5((p -1)N+n)+pn5(2' +(p -1)N+ ,) sc,n\Pi The size of the scrambling code matrix may be increased using the following operation: s. = (diag(i)® N Nx( £v(X) I 2 = (0 (p) (N(t) +2)]e C For each symbol period a signature sequence matrix may be generated as: S(p) = d/ag(s1 (p))s12.

The signature sequence matrix S(p) for p=1,...,NH may be concatenated to generate the extended signature sequence matrix as: s = [s(i),.. ., . . . s(vH)] CNtNXAWN(X) For each channelization code a signature sequence matrix may be generated for different symbol periods as follows: = diag( = [s(1),...sfr(p),...s(NH)]e CA1fork = 1,**,NK.

The transmitted chip sequence matrix for different symbol periods may be generated using: Y YxY KV') z = [s(2),.. .,s(p),.. .s(iv +i)] c and z, = [s(i).... S(p),... s(AT(1))] E and also = [s(3), ... , s(p),. . . + 41 e cw, KV(X) The transmitted spreading sequence matrix may be incorporated with a novel covariance matrix calculation method to obtain the inverse of the covariance matrix.

The covariance matrix inversion may involve the use of transmission spreading sequence matrices Z, Z, and Z1 and their covariance matrices. The spreading sequence matrices may be available at the transmitter and receiver in advance of transmitting blocks of symbols. The use of preset transmission spreading sequence covariance matrix eliminates the requirement to use computationally intensive matrix calculations and inversions at the receiver each time a block of data is received.

The matrix inversion scheme disclosed herein can be incorporated with a symbol level MMSE receiver equalizer operating with short blocks of transmission symbols.

The covariance matrix inversions and also the receiver equalizer designs may involve the use of link level MIMO receiver modeling. In the link level MIMO modelling of the received signals, the spread transmitted symbols may be allocated equal transmission energies with the transmission energy constraint that the total transmission energy per symbol period is ET and each spreading sequence may be E1 allocated equal energy such that Efr = N K for k = 1,. N3K.

A process is disclosed herein based on the principle of utilizing a system value to determine which data rates should be allocated to increase the achievable data rates when spreading the transmitted symbols.

A symbol level equalizer may be followed by a successive interference cancellation scheme. Also, a method to use a constrained power level at the transmitter of the HSDPA downlink to adjust the transmission data rates for different groups of spreading sequences when using the proposed System Value and a measurement method and the simplified covariance matrix calculation and inversion is disclosed.

A novel method of calculating transmission data rates for different groups of transmission spreading sequences is disclosed that provides a total transmission data rate close to the theoretical upper bound.

Systems disclosed herein allocate Discrete Data Rates to maximize the total rate using System Values. Providing multiplexing versus diversity gain trade off when selecting a sub-set of receiver antennas from a large set.

Brief Description of the Drawings

Exemplary embodiments of the invention shall now be described with reference to the drawings in which: Figure 1 provides a schematic illustration of an HSDPA MIMO transmitter and receiver arrangement; and Figure 2 provides a schematic illustration of a Successive Interference Cancelling receiver.

Throughout the description and the drawings, like reference numerals refer to like pa rts.

Specific Description

Figure 1 shows a multi-code CDMA downlink system. The system has a total of iV and N,. transmitter and receiver antennas and also N,K spreading sequences, where N. = N, is taken for this specific example, each of which is realizable with a bit rate of b, bits per symbol from a set of bit rates, 1b1, }) , for a given total k Ic pl transmission energy ET and p = 1,2,, P. The basic principle of operation of this system is that by excluding the weak channels corresponding to a specific set of spreading sequences, the number of parallel transmission channels is reduced to N,K spreading sequences to transmit a symbol per channel. The data for the intended symbol for each channel is placed in an (N xl)-dimensional vector u for k=1 N,K.

The system shall now be described in detail.

In Figure 1, a transmitter 100 receives input vectors of u for k = I N,K. The input vectors represent input data to be transmitted by the tiansmitter 100. This input data is then input into encoding unit 101. Each of these data packets is then channel encoded by the encoding unit 101 to produce a (Bxl)-dimensional vector 4. The encoded data k for k =1,...,NK produced by the encoding unit 101 is then processed by the adaptive modulation and coding unit 102 to transform the encoded data into symbol vectois using a quadrature amplitude modulation scheme (QAM) with Al constellations to transmit data at a rate b = 1og214 bits per symbol.

The channel encoder rate is and the realizable discrete rates are given by = ç,,. Iog,iI for p = P where P is the number of available disciete data rates.

Data is tiansmitted in packets at a transmission-time-inteival (TTT) and the number of symbols transmitted pei packet is denoted as N wheie = TTI and N is

NT

the spieading sequence length, I. is the chip period and N1. is the symbol period.

The modulation and coding unit 102 transforms the encoded data into the (Nx1)-dimensional symbol vector Xk = [xk(1), for each channel k = 1 N1K. The entire block of transmission can be iepiesented as an (N x N1K) dimensional transmit symbol matrix defined as: x=Lxl,...,xk,...,xvKj = [. . The transmitted vector y(p)= [yjp),...,y(p),...,y (p)j' contains the symbols, over the symbol period p = 1, **,N', with the unit aveiage eneigy ELVk(p)y;(p)j= 1 for k = 1 N,K where y;(p) is the complex conjugate of the symbol y,(p).

The transmission symbol energies are then adjusted using the power control unit 103. Power allocation is performed on the symbols before spreading. The energies for all N,K channels are stored in an amplitude matrix A = diag(J,. ..,J,. ..JEJ<) subject to the total energy ET such that = E1.

The energy weighted data symbols are then transformed into a transmit vector using the vector generation unit 104. After assigning energies, the amplitude weighted symbols are spread in the spreading unit 105 by a plurality of NxN,K dimensional spreading sequences s(p)= CNXIK, for n, =1,**,N1 where C defines the two dimensional complex space with dimension sizes gives as the superscript values, the signature matrices are generated from s(p)= sf(p),". s(p),... s(p)1. At every symbol period p = i,.", NH the length N transmit chip vector: z(p) = E, S(p)y(p) E Cfor p = 2,**, (N" + i) N, K is generated at the input of n antenna unit for ii, =l,'",Nç. where Y(P)ECNtK is the transmitted signal symbol vector for the p" symbol period. The transmitted chip vector for the previous (th -i) symbol period is: -i) = E S(p -1)y(p -i) E N, K The transmitted chip vector for the next (h +i) symbol period is: z(p + ET s(p+ 1)y(p +i)e CN The spread signals are next filtered using the pulse shaping filter 106 to produce transmission signals for transmission from the MIMO transmitters 107a, lO7b,...l 07 NT.

The signal is then transmitted from the transmitter to a receiver 200 across a channel. Since characteristics of the channel affect the received signal it is necessary to model the channel in order for the transmitted signal to be derived from the received signal. The modelling of the channel shall firstly be discussed before considering how the received signal is then dealt with.

At each TTI, the transmission procedure described above is used to transmit the pilot signals which are used in order to estimate the channel condition at each receiver to calculate the data rate to be transmitted over the km channel. The rate is calculated at the receiver and is fedback to the transmitter. It is assumed that the channel condition does not change for that TTI. All symbols in a block of length N in all spread sequence channels when transmitted from the n transmitter antenna to the n receiver antenna experience the same channel condition in a multipath environment with L resolvable paths. This channel condition [ACCEPTED] can be represented by a channel impulse response function = ...

corresponding ((N+L-1)xN)-dimensional channel convolution matrix can be represented as follows: h't) o o = . . . 0 ? 1z7 n,) The overall (N(A/+L_1)xAçN)-dimensional MIMO channel convolution matrix can be formed as: H'' H = . e C_L_1tN where L is the transmission channel impulse response length. The channel matrix for the previous symbol period is: H = (i, ®(jrf)n e The channel matrix for the next symbol period is: H, = (INr ® E 0N(+T_I)xAAT The (N--L-1)x(ZV+L-1)-dimensional matrix, i.e. the amount by which one OL. 0 symbol affects the next symbol, is defined as VLl = --. For (NiL 2) QNIL 2) simplicity the subscript will be dropped from the J matrix notation. When the matrix jN operates on a column vector, it downshifts the column by N chips filling the top of the column with N zeros. J is therefore used to convert H to H1 and H2.

Now that the channel has been defined we shall return to the description of Figure 1 and specifically examine the processes carried out at the receiver 200.

The transmitted pilot signal is then received at the receiver 200 by the MIMO receivers 201a, 201b, ...201 N,. The received signal long chip vecto1rk(p), which is the signal received at the receiver 200 has been affected by the channel and therefore does not directly correspond to the signal that was transmitted. It is therefore necessary to determine the signal that was transmitted by processing the received signal accordingly, as will now be discussed.

Assuming the clocks at the transmitter and receiver are fully synchronized, the received signal is then frequency down converted, filtered and sampled as chip period intervals by the chip matched filter unit 202. The received signal long chip vector rfl,,,, (p)eCNT1L 1) is received as: Lioug (p) = HS(p)y(p) +v 1)) (P i) +ETH2s(p+1)y(p+1) = Hz(p)+H1z(p -1)+H2z(p -1)+n(p) where n(p) E c''' is the noise vector at the output of the receiver chip matched filter.

From the received signal r10,(p) the received long signal matrix N (w+r-i N) Rj,,g c C produced as follows: R1 = [r (i),... ,r1 () (N)]e CN_L_O The received long signal matrix is obtained in order to increase the processing efficiency.

Using a simplified covariance matrix inversion method and also with a minimum mean square error equalizer implementation, the signal that was transmitted can be obtained from the received long signal matrix, as shall now be described.

Firstly, the derivation of a covariance matrix used to map the long vector rg(p) at the output of the chip matched filter unit 202 to produce the short vector r,(p) at the output of the vector mapping unit 203 shall be considered. At the receiver chip matched filter output the long vector iong(p)eC4J_1) has dimension Nr(N + L -i) and to depsread the received vector all the interferences from the N (N*L-1lxN N. multipath MIMO channels H cC r / must be minimized by multiplying H at / \u -V VxK(V_T_I) the receiver with the vector mapping matrix C 11Cr) cC with the / -1 VI dimension such that the multiplication C HCr) He C / has the dimension N1N x N1N and its diagonal terms are unity. Here I (N-L-] ixN (* -i-li C = Er (p)r,0, (p)c C is the covaraince matrix of the received N (N+l-1) . NrN A long vector r(p)cc r. The matrix C cC / is the amplitude normalization matrix. The non-diagonal terms of (C-HC,, )1' 11 identify the residual interferences between specific transmitter and receiver antennas. When multiplying -. (v i. i) the long vector r20,, (p) c C with the vector mapping matnx ( -] Vt NNxN (N-il-I) C HC4 cC I r at the receiver the short (p) cC / is produced using r(p)= (: H(J'r (p)c CN. The resultant short vector r01(p)e CN is the best estimate of the transmitted chip vector z(p) c CN produced by the MMSE equalization unit at the receiver. An estimate of the transmitted symbol is obtained at the receiver by descrambling the received r1,0 c C vector with the scrambling sequence and then despreading it using the channelization signature sequence for a specific channel.

The received signal's covariance matrix averaged over a given number of symbols, corresponding to a given block length of the symbols observed in the demodulation process, is calculated using the following formulation: I i \ H I N (N*L-ffrN (N*L-1) C = Ek/_:iwzgP)?::io,,g P))E C Equation I E V VKV1 = +HIZ1Z1Hff +H2ZQ1Z1H + 1 In the above equation during the averaging process the channel matrices H, 11 and H, are taken to be constant as the channels are considered to be stationary during one TTI period. The identity matrix V(V±[_l) is used to deal with noise covariance matrix which has noise power spectral density of N0. The averaging will be taken by using the term which changes from symbol to symbol when observing the long vector r1 (p) E cN_). As the only term which changes from symbol to symbol is the scrambling sequence combined with the channelization spreading / \ N (N--I-fl sequence, the received signal vector r10,.ip)e C r and its complex conjugate transpose vector r,0,,"(p) will be averaged. The averaging process involves multiplying r (p) with r1 (p) resulting in the following relationship: C = N(x) (p)r,H ()

-I

This averaging process involves the calculation of covariance matrices for every symbol period and summing it over symbol periods and dividing the result by As the matrix calculation involves the use of transmitted chip vectors z(p)z" (p), z(p -1)z" (p -1) and z(p + 1)z" (p + i) in the following form: Liong (pkiong" (p) = (p)" (p) H' ) +nj p=I + H2z(P+1)zH(P+1)IJH211 + i\Q N I!V (!V -i-i) As the matrix summations in the above equation are expressed as Zr and:z(plY(pl) ZpiZpi and also r V( +1zM( T z Z" / NxT -" P' the received signal covariance calculations zzIJ involve the use of trasnmitted signature sequence covariance matrices and Z1ZHI Z1Z, . Z Z1 Z1 " -and also " where signature sequence and and also matrices are defined earlier. In the covariance matrix calculations the terms and ZT1 Z " and also " are the conlex conjugate transposes of and " and also Z1,4 matrices respectively. In equation (6) the received signal covariance matrix zzu z z" z V' calculations matrices and " and also " are the most

V KAT

computationally heavy calculations as the number of columns of matrices (x) A AxA AA, N AxA AN A AxA AN I S Z eC Z eCt and " and " is proportional to the \T() number of symbols over which that averaging is performed. However, as both the transmitter and receiver know over which symbol period which scramling and chanelization signature sequences are used for transmission for a given number of ZZH z Z", symbol observation periods the covrainces matrices and p p and also

ZZH

-P-1 can be calculated in advance at the transmitter and receiver and may be used in conjunction with Equation 1 to calculate the covariance matrix ji) C = N when the channel matrices H, H1 and H, are estimated at the MIMO receivers using the transmitted pilot signals.

By using a given number of symbols which 1< NX) «= N, the minimum mean square error equalizer can be implemented for different periods of covariance matrix v (x) averaging. There will be a total of different epochs with each epoch having its covariance matrix c() for the epoch number = 1,,' . Keeping N' low

VS

gives a covariance matrix which is a better representative of the signature sequence over a shorter period, however, this results in a higher computational complexity at the receiver. A compromise between the computational complexity and the averaging period of the covariance matrix calculations over a shorter number of symbols will need to be made. The most computationally heavy matrix multiplication involved in calculating the covariance matrix C is the calculation of matrices ZZ" and ZJ1Z1 and also Z1Z. If the covariance matrix averaging is undertaken for different number of symbol periods N, as both the transmitter and receiver know the transmission signature sequence matrices Z, Z1, and also at the receiver the matrix dimensions can be fixed to be Z e and z e and N NxN KV(X) . . . . . also e C S s if the average covariance matrix is estimated using Pv; symbol periods. Once the the dimensions are fixed for the transmission signature sequence matrices Z, Z and also Z, covariance matrices z()z"), z,, 1()z1() and z,, 1()z1ç) for the signature sequence matrices for the c" observation epoch can be calculated in advance to produce the averaged covariance matrix c() using: cç)= NKN(t [llz(5)zH()HH ÷n1z_1()z1()Hf + NNflT Iv(VLI)J Equation 2 for observation periods 4 = . It is also assumed that for the averaging process the channel matrices H, H1 and H, are produced at the start of every TTI period and are assumed to be stationary for the duration of TTI period.

The covariance matrix of Equation 2 is then used to map the long vector (p) at the output of the chip matched filter unit 202 to produce the short vector r0(p) at the output of the vector mapping unit 203, as will now be discussed.

A minimum mean square error (MMSE) equalizer is firstly used to reduce the length Pv (y_f_) YrT+1) of the long received signal vector rfr,rg(p)EC r to size a N,N vector and and also to de-scramble the shortened vector elements to produce the short received vector r1,0(p) E CISIA. A vector mapping matrix (c' HCj e CA, with the dimension N,Nx Nr(N +L -i) is used to shorten the vector to size a N,N vector to produce the short received vector r,hOfl(p)CC' . For the receiver signal amplitude normalization the chip sequence normalization matrix Cr E CAINXI,A is generated as follows: ii = (Diag(Diag(H"c IH))) e The shortened and de-scrambled received signal chip vector r510,1(p) E CV is produced at the output of vector mapping unit 203 using: = HC r (p) CN In a matrix form the mapping operation between the received long matrix Rjong e cr L and the matrix are used to de-scramble the short vector and produce the elements of the shortened received signal matrix N NxN R0 e C is given as = )* [Lh07I (U!_shc,rz, Lc;20ri (1\Tc)] = (s)*((CIHC)R10J The shorted received signal vector rskor,(p) is used together with the channelization codes to estimate the received symbol vector by the despreading unit 204 using the relationship 2(r) = sZ. (Diag( (p)))i''jyy () = k' (n)' 5' (n)' 5' (p)j The estimated data sequences are then reorganized to produce the estimated data for each spreading sequence using the receiver mapping unit 205 and the decision unit 206.

Each of the above mentioned units of the transmitter and receiver apart from the actual MIMO transmitters 107a, 107b 107N, and receivers 201a, 201b 201 N. are implemented in software.

Above, the operation of the system for determining the transmitted signal from the received signal has been discussed. Now the determination of the data rate for transmission using a system value based method shall be discussed. The following procedure may be used in combination with the above process or with other processes.

Firstly, the system value calculation shall be defined for a system in which signature sequences are assigned on a symbol-by-symbol basis. Then, use of the System Value in order to determine which sub-set of transmit and receive antenna groups should be used when a large group of transmit and receive antennas are available shall be discussed.

It will be appreciated that by determining the System Values in the above mentioned data transmission apparatus the overall data rate achievable by the system can be improved.

The system value is a variable which is indicative of the characteristics of the channel over which data is transmitted. The system value is the normalized signal energy at the output of the despreading unit. The difference between normalized total energy of unity and the system value gives the mean square error at the output of the despreading unit. The ratio of the normalized energy, the system value, to the mean square error gives the signal to noise ratio at the output of the despreading unit.

The mean square error c for the kth symbol is: / \ 1 k = Ei'k(p)-Yk(pf)= = I A Yk +1 where y is the signal-to-noise ratio for the kth symbol and E(.) identifies the expectation operation. The system value 2,, = iLk is related to the signal-to-interference-noise ratio by:

-____ 1-2

A modified version of the System Value optimization criterion is used to maximize the total rate when transmitting with spreading sequences which change every symbol period. In current HSDPA standards the spreading sequence is changed every symbol period. The MMSE receivers can be configured to achieve average SNRs, which are equal at their output, when using equal energy transmission allocation based on the system value measurements. As the equal SNR scheme only requires equal rates to be allocated at the transmitter, the data rates can be estimated at the receiver when the transmission channel impulse response is available at the receiver. As groups of channels use the same transmitter data rate for each channel, the receiver only needs to report a reduced number of data rates to the transmitter by using the System Value calculations.

The System Value formulation involves the use of interference matrix for the current symbol defined as: D=C' H"CH_dtag(C HJJCHH) The channel impulse response H is converted to the residual channel interference D 0N,NxNN matrix C, which is a square matrix for more efficient processing. In particular, as will be seen, by converting the channel impulse response H to matrix D later processing steps are optimised.

H H -] N NxN (N*L-1) . . . Here when the matrix Cr H C CC r is multiplied with the channel matrix H the resultant matrix C' HHCIH e CNtNXN is a square matrix with diagonal elements equal to unity. When the diagonal terms diag(C' H"C H) are subtracted from C' HHCIH the resultant matrix D identifies the Multiple Access Interference (MAI) between chip elements of the transmitted chip sequences. Hence the matrix H 1] 1 NxW (WH-T-i) . . . C H C cC t r is multiplied with any matrix it changes the number of rows from Nr (AT + t -i) to NN.

When the transmission signal chip vector z(p)eC"t" is multiplied with the matrix O e CXIVtA the resultant vector ZAIAJ(P)E CNt identifies the residual interference for each chip before despreading as follows: (p) = Diag(s = (p))" Dz(p) When the vector z1(p) is de-spread using the channelization spreading sequence at the output of the Ic despreading unit, the error signal due to the multiple access interference is given as: = s (p)Dz(p) Equation 3 N N<N The matrix C H C e C r can also be used to change the number of rows of the intersymbol interference matrix H1 for the previous symbol period from V NxN N Nr(N+L_1) to Ay to produce the interference matrix DeC / l for the intersymbol interference from the previous symbols as follows: = (C:;' H"C1 (lAr ®(j' ))H) =(c' HHc1H1)eC When the transmission signal chip vector z(p -1)e C' is multiplied with the matrix D1 the resultant vector z1i(p)eCtA identities the residual ISI interference from the previous symbols as follows for each chip before despreading: (p)= Diag(Jp)fD1z(p-l) When the vector 11(p) is de-spread using the channelization spreading sequence at the output of the k despreading unit, the error signal due to the previous symbol intersymbol interference is given as: / \ Hf \ I /sr,,,P)Thk p)Dzp-1 Equation 4 The interference matrix for the intersymbol interference from the next symbols is also given in the following form: D2 = (c' HHCi(INr ØJN)[t) = (cf' HHC1H,).

When the transmission signal chip vector is multiplied with the matrix 0, the resultant vector identifies the residual interference vector for the next symbol period intersymbol interference as follows for each chip before despreading: z, (p) = Diag fl)H D,z(p + i) When the vector z72(p) is de-spread using the channelization spreading sequence at the output of the kM despreading unit, the error signal due to the next symbol intersymbol interference is given as: lSJ,k(P) = (p)o2(p + i) Equation 5 ill H 1 N NxN When using the matrix C H C. e C r at the receiver, the noise vector n(p)e at the output of chip matched filters is also converted to the short noise vector 5b0, (p) e C*t\I by the following operation: shofl (p) = Diagj (p)f ct,' u"c n(p at the output of the de-scrambling unit. When the noise vector is de-spread with the channelization spreading sequence the noise signal at the output of the channelization despreading unit is given as: noise.k(P) = H "C Ln(p) Equation 6 The error signal k for the k/h channelization despreading unit due to Multiple Access Interference and the Inter Symbol Interferences (ISI) from the previous and next symbols and also the noise component is obtained by adding the error signals (3) (4), (5), (6) to produce the error at the output of the de-spreading filter for the k channel as follows: (p (j ( fl tMAI,k + Jsfl Ic Ph + tISI2k P + tnoisek = (PXD_Z(P) D1z(p -i)+ D2z(p + + C' HHC1n(p)) The Mean-Square-Error due to the MAI interference for the signal at the output of the de-spreader for the kt' channel is given as: tMAJk = E( Jkk)= KNNfr)N(X) Trace(SfDZZ"D"Sk) Equation 7 where E(.) refers to the expectation operation Here the covariance matrix ZZ" is the signature sequence matrix Sfr is the signature sequence matrix for channel k.

The Mean-Square-Error due to the Inter-Symbol-Interference from the previous symbol is given as: 815T.I.k = E( ji,c(pcji,c(p))= KNNA Trace(SDIZIZ/,IDUSfr) Equation 8 The Mean-Square-Error due to the Inter-Symbol-Interference from the next symbol is given as: tJSJ2k = E(IM2k(p)2(p))= KN3N1X)N) Trace(S'D2ZlZPlD'Sk) Equation 9 The following noise vector despreading matrix B e matrix is formulated as follows: B=s7c: H'C' cCAx*_ to calculate the Mean-Square-Error due to noise as follows: = E(ro;%.ek (Pl:-Lk(P))= N' Trace(BBH) Equation 10 The normalized Mean-Square-Error due to multiple access interference, previous symbol and next symbol ISIs and also noise is obtained by adding equations (7), (8), (9) and (10 in the following form: = E(kk* ) MAJ.k + + + = Trace(S7DZZ"D"Sfr) 7TJ(yk) AT + rrace(S"D Z" Z D"S) N,KN'N k 1 p1 p1 1 A + Trace(SffD,Z Z D'S) jVKNt4 NC) A -p-i p4-i + 0 Trace(BB")

N

Hence the System Value, averaged over N1'symbol period using spreading sequences, which change symbol by symbol basis is: Ak ILk = -ET rrace(S'DZZ"n"S) N1KN'N" -E rrace(S"DZ" Z D"S) N,KN' N(T) A pi pl 1 A -rrace(s"Dz" Z Di'S) N,ICN/ y(r) k p*l p-i -A -N0 rrace(BB") Now that it can be seen how the System Value for use in a system that applies signature sequences on a symbol-by-symbol basis has been described the determination of the data rate based on the determined System Value shall be considered.

The following procedure is based on the principle that the number N of transmitter antennas is less than the number N, of receiver antennas. The spatial multiplexing Ac number N is hence equal to N. At the MIMO receivers, a total of Ac transceivers can be used independently. Each MIMO receiver has a different channel matrix H that is uniquely associated with the MIMO transmitter and receiver antennas belonging to the sub-set of the receiver antennas selected from the full set of receiver antennas.

At a given MIMO transmitter-receiver antenna combination, the transmission scrambling and channelization spreading sequences are known to both the transmitter and receiver. Before the packets are transmitted from the transmitter to the receiver, the transmission packet duration is expressed using the number Nt of y(x) symbols transmitted in a given packet. Hence at the receiver different sub-N; (x) packets, with an epoch number = 1,***, may be formed each with a packet length At the receiver the signature sequence covariance matrices z()z"(), and are calculated in advance for the signature sequence matrices and the observation epoch and for the equal energy allocation Ek = to produce the covariance matrix c() averaged over a total of N symbol periods using: F V \TKV) c()= VKu + Hz,()Z_ç)ll + H2Z,÷ç)z_()H + U' Equation 11 for observation epochs 4 = Using the covariance matrix c() for epoch , the interference matrices averaged over iV symbol periods may be calculated using: c()= (Diag(Diag(H"c ()H)))' Equation 12 D() = C () H"C ( )u -diag(c' () H"C 1 ( )u) and D1()= (cU ()i-i"c'()H1) and also D2()=(C ()HHc1()H,) and also the noise matrix ()W'c'() The System Value for the th epoch and the ktL channel is then given by: Ak () =1-bk () Equation 13 = N,KN N1 -IK N n?ice(S ()D, ( )z, ()z,1( )D' ()sk()) -AcK(1) Nr Trace(S7 ()D2()z1 ()zP+I()D)sk())

-___

If the number N of symbols is a small number, the System Value Afr() for each channel for epoch will be almost identical for each group of K signature sequences out of N different groups of streams. Hence, the streams will have the following System Values.

) () fort «= k «= K and the stream number flç = 1 AKi () forK + 1 «= k «= 2K and the stream number n5 = 2 for(n3 -UK +1 «= k«= n5Kandthestreamnumbern5 N,-])k--l () for(N, -UK + I «= k «= N, Kand the stream number ti = N, On this basis, the discrete data rate for each stream is allocated as follows. For the set of bit rates {bJ_1, the corresponding target system value is determined according to the following equation: 1' [1 [(2-1) forp=1,...,P 1 -[(2 -1) For the stream group number ii, find h,,(n.) that satisfies [At],, «=A,_ij,44()<[*],,_i This determines the symbol rate allocation for stream group n,. In the receiver implementation, the stream with the largest system value is selected. Symbol rates are allocated by finding the b value that satisfies the equation [Lt] «= ()< [A],,11 I'(p) = s. (D/ag(s (p))fr30, (p) Equation 14 for the symbols received during epoch. From the estimated signal's vector 5'(p), the received signal matrix X is generated as follows: = Equation 15 From the estimated signals matrix X the decision matrix XD channel decodeis is generated in the following form: = L1D' k,D' _IK.DJ Equation 16 and = [xkflO),.. X)J)(N)f The system values can be used to implement sucessive interference cancellation receiver as shall now be described.

In the successive interference cancellation receiver implementation, the estimated symbols for the stream group with the largest value of are considered. The channel decoders are used to decide which symbols are detected v (x) correctly for epochs = 1,..., If up to n, channels are detected correctly in the stream, the channel numbers ki**kr are identified for the channels which have all the symbols detected coiiectly foi p = 1,***,N. The signature sequence covaliance matlices z( )ZH (a), z_, ()z_, (2) are updated and z1( () using: z()z"()= z()z"()-Sk()Sf() k=k1 where SI(E)=[siH1) +i} and it,, z21Q)z1 () = z2 1()z1Q)-Sk(Pl)(sk(PUc) where S;(PH)()= -)}...s-')NH +p-i9 and z21()z± () = z21L)z1()-Sk(P_U()S'(9±Uc) k-/c1 where S (pIt) = -i)1 + 29... 4( _0N + p + i99 for the updated signature sequence matrices, the received signal covariance matrix cç) and c(e) are calculated using Equations 11 and 12. The received long matrix R;g is iteratively updated for the correctly detected symbol sequences in the channels updated using: "ic Rjoj,g = R,3, -E4 k=k' Equation 16 = (i), ... , (n),. , (v)] where = k(0, . Equation 17 with -() [Uk(P)X/:/)(P)+ H]Ifr(P_1)X/:.p(P_1)+ u2sk(p+1),)(p+1)j after updating the receiver long matrix using Equation 16 the vector for the estimated symbols are generated using Equation 14 The successive interference cancellation is iteratively run for the remaining data streams in the order with the highest remaining values of A1,2 i:J)R']p A Successive Interference Cancellation (SIC) receiver shall now be described with reference to Figure 2. The SIC receiver uses the System Value discrete rate allocation scheme just described. In Figure 2, the signal transmitted from MIMO transmitter, as shown in Figure 1, is received at the receiver 300 by the MIMO antennas 301a, 301b 301 1v. Assuming the clocks at the transmitter and receiver are fully synchronized, the received signal is then frequency down converted, filtered and sampled at chip period intervals by the chip matched filter unit v (v-ii-1) 302. The received signal long chip vector rj0,(p)e c r is concatenated to produce the long received signal matrix R by the concatenation unit 303. The successive interference cancellation matrix handling unit 304, processes the long received matrix R10 to remove signals iteratively using Equation 16 and the signal removal matrix Fk given in Equation 17 The generation of the signal removal matrix 1'k is handled by units 304, 305 and 307. The estimated signal vector generator unit 305 handles the generation of v(p) using Equation 14 and produces the estimated signals matrix X using Equation 15 The decision unit 307 produces the decisions matrix XD given in Equation 16 using the estimated symbols matrix X. The decisions matrix XD is passed to the unit 304 to produce the removal matrix tFk using Equation 17 and hence the long received matrix R,0,,. for the iterative SIC operations. The estimated symbols matrix X is also passed from unit 305 to produce the estimated symbols vector y(p). The binary data unit 308 takes the symbol decision matrix XD from unit 307 and produces the (N xi)-dimensional binary decisions vector ut,.

In alternative arrangement, the system values given in Equation 13 are used to choose a subset of MIMO receiver antennas with the subset cardinality is equal to N if the number of receiver antennas is N and N = N. In other words, if there are more transmit antennas than reciever anternnas then the antennas to be used, i.e. those with the best channels, can be selected for use in transmitting to the receiver.

The system values are used to determine which subset of the receiver antennas may be used over the epoch given that the number of bits to be transmitted over each channelization code in a stream group is already determined using the rate allocation algorithm described above. The idea of using different set of receiver antennas is based on the fact that System Value obtained over a given large number of symbols N will be constant for a given stream. However the System Value will vary from channelization code to channelization code and also from symbol period to symbol period over a short number of symbols If the system value is low for a channelization code for a given stream for a given subset of MIMO receiver antennas, the system value for the same channelization code and epoch for a different subset of MIMO receiver antennas. Hence, the System Value calculations described above may be used to identify how different subset of MIMO receiver antennas can be combined at the receiver.

The various methods described above may be implemented by a computer program.

The computer program may include computer code arranged to instruct a computer to perform the functions of one or more of the various methods described above.

The computer program and/or the code for performing such methods may be provided to an apparatus, such as a computer, on a computer readable medium or computer program product. The computer readable medium could be, for example, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, or a propagation medium for data transmission, for example for downloading the code over the Internet. Alternatively, the computer readable medium could take the form of a physical computer readable medium such as semiconductor or solid state memory, magnetic tape, a removable computer diskette, a random access memory (RAM), a read-only memory (ROM), a rigid magnetic disc, and an optical disk, such as a CD-ROM, CD-R/W or DVD.

An apparatus such as a computer may be configured in accordance with such code to perform one or more processes in accordance with the various methods discussed herein. Such an apparatus may take the form of a data processing system. Such a data processing system may be a distributed system. For example, such a data processing system may be distributed across a network.

Claims (22)

1. Claims: 1. A method for determining a transmitted signal received at a receiver in a radio transmission system having a plurality of parallel single-input single-output or multiple-input multiple-output channels over which the signal is transmitted, the signal comprising one or more symbols, the symbols being spread prior to transmission by a plurality of spreading sequences on a symbol-by-symbol basis every symbol period p, the method comprising: receiving a signal r (p)) corresponding to a signal transmitted over a channel, wherein each symbol of the transmitted signal is spread using a respective spreading sequence determining an estimated channel characteristic H indicative of effects of the channel over which the received signal r,orjp)has been transmitted on the transmitted signal; and determining the transmitted signal rh,,(p)from the received signal rrg(p)in accordance with the estimated channel characteristic H and the spreading sequences used to spread the symbols.
2. 2. The method according to claim 1, wherein the method further comprises: determining a combined matrix Z of all spreading sequences S(p) used to spread the symbols of the transmitted signal; and determining a covariance matrix C and a normalization matrix Cin accordance with: the estimated channel characteristic H; and a multiplication of the combined matrix of the spreading sequences Z and the hermitian of the combined matrix of the spreading sequences Z"; wherein the transmitted signal is determined from the received signal r,(p) by multiplying the received signal r10(p)with: the inverse of the covariance matrix C; the hermitian of the channel matrix H; and the hermitian of the normalization matrix Cr.
3. 3. The method according to claim 1 or claim 2, wherein the covariance matrix C is defined as follows: C = HZZ"H + HIZ IZIH + H2Z0÷IZ±IH + N0KVH IN(N-T-l)/J wherein FT is a total transmission energy! N = Nc is a total number of MIMO transmission antennas, NK is a total number of spreading sequences, NN) is a total number of transmitted symbols used to calculate the covariance matrix C, N0 is a noise power spectral density, N(N*L-l) is an identity matrix, the previous symbol period channel interference matrix H is produced from the estimated channel matrix H using: H1 = (N ® Jrf)H e and wherein the channel matrix for the next symbol period is produced using: H, = (INr ®,jN)11 e wherein the (N+L-1)x(N+L-1)-dimensional matrix NiL is given as = and also the operation J refers to the W" power of (NiL 2) Q(NL 2) N+L-i further the operator ® corresponds to Kroneker product, ZZH are transmitted signature sequence matrices and ZJ_1Z_1 and also are the transmitted signature sequence matrices for the current previous and next symbol periods.
4. 4. The method according claim 3, wherein the covariance matrices of the transmitted signature sequence matrices ZZH and Z 1Z1 and ZIZI for the current, previous and next symbol periods are determined according to the following: z= [s),...,s(p),...s(AT(1) +i)] = [s(i. ..,S(p. ..s(1v)j z,,1 = [S(3),. . . , S(p),.. . + 2)j wherein a spreading sequence matrix s(p) of all spreading symbols for each symbol period p is determined according to: S(p) = (p))S1.wherein S1.hw is an extended channelization matrix and s(p)is a scrambling sequence vector for symbol period p wherein the scrambling sequence vector (p) is determined according to: = (dIag(I)® S) = [. (i),.. () (y(x) + 2)1 and wherein the extended channelization matrix S. is determined according to: = (t ®SChK)= kwd.1 wch.fr"irh.K I wherein a total of K channelization spreading sequences SUbK are generated from a total of N Hadamard codes with spreading sequence processing gain N, and a signature sequence matrix S = j wherein a channelization spreading sequence vector 5yk assocaited with the signature sequence matrix S has the property: 1 i=jH =0 i!=j and wherein a transmitter precoding matrix W is determined according to: w = Ii?N j wherein the precoding vectors w forn =1,... ,N5 with the property: 1 1=1H0 i!=j are the beam stearing vectors used at the transmitter antennas.
5. 5. The method according to claim 4, wherein! for a Successive Interference Cancellation, SIC, receiver the signature sequence covariance matrices Z(E)Z"(), z,,_()z_() and z,, ()z÷() are updated using: k z()z)= z()z)-S($) k=k1 wherein Sk() = Ek 1)N() 9' "k 1) + and z 1()z ()= z j)z ()-Sk(P 1)($k l)(c) kk1 wherein Sk(PI)() = -i) 9... S[( _l)N() + 99 and z1()z, ()= z1()z1()-Sk(P k-/c1 wherein = -i) + 2, s1c[( _l)N(X) + p +199 for the updated signature sequence matrices, and c(4) and c() are the received signal covariance matrix and the normalization matrix respectively for use in the implemenation of successive interference cancelation receivers and wherein is the epoch number, N and NH are the number of symbols transmitted in a block and a set of the blocks respectively and wherein the covariance matrix c(e) is determined according to: c()= NKT() + H1Z 1()z 1()Hff +H2z1()z;1()H + NN'L 1)9 and the normalization matrix is determined according to: 1
6. 6. The method according to claim 5, wherein, for the Successive Interference Cancellation, SIC, receiver the received long matrix R10 is iteratively updated for the correctly detected symbol sequences in the channels updated using: tt/Qflg = Rj0jg -h,, q) Ic-k1 = k09 (N)] wherein the long matrix Gfor the correctly detected symbols is determined according to: = k'' k (N(1))j wherein the k channel's chip sequence vector q5(p) for the symbol period p and spreading seqeunce sk(p) is determined according to: _k)Lk()+ H1k(p-1kD(p-1)+ H2k(p+1)xkD(p+1)j for successive interference cancellation receivers.
7. 7. The method according to any one of claims 2 to 6, wherein the transmitted signal e51011(p) is determined according to the following: = ((c HC,, Wherein C1 is an inverse of the covariance matrix, £(P) is the received signal, and the normalization matrix C,, is determined according to: = (Diag(Diag(H"c'H))) 1
8. 8. The method according to claim 7, further comprising estimating a received symbol vector using the following relationship: 2(P) = S. (Diag (p)))" (p) = L; ()** Yk ()** ;AK ()j Wherein r.hGr,(p) is the transmitted signal, SChW is an extended channelization matrix and s(p)is a scrambling sequence vector for symbol period p wherein the scrambling sequence vector s(p) is determined according to: = (ciiag(i)® s) = [IHSC (i), ... (n)' + 2)J and wherein the extended channelization matrix SCbW is determined according to: Srhw = ® S42) = vch.k Lrh.K] wherein a total of K channelization spreading sequences SK are generated from a total of N Hadamard codes with spreading sequence processing gain N, and a signature sequence matrix s = [55 "S35 j, wherein a channelization spreading sequence vector s associated with the signature sequence matrix S55 has the property: 1 i=jHys,j = 0 i!=j and wherein a transmitter precoding matrix W is determined according to: w = Ji?N wherein the precoding vectors w forn, =1,***,ZV. with the property: ri II j L° i!=i are the beam stearing vectors used at the transmitter antennas.
9. 9. The method according to any preceding claim, wherein the estimated channel characteristic is defined by: jj(11) . . A,) JEll 1) wherein the channel convolution matrix H"'' between the transmit antenna ii, and the receive antenna n1 is defined by: o o i 7/ : [J fl L-1 0 0 wherein a channel impulse response vector between a transmit antenna n, and a receive antenna nr is defined by channel impulse response function: = [h'r'"t... h'
10. 10. The method according to any preceding claim, wherein the estimated channel characteristic is a channel impulse response function.
11. 11. The method according to any preceding claim, wherein the received signal is defined by: (p) = Hz(p)+ H1z(p -i) + H,z(p -i)+ n(p) wherein the vector n(p) corresponds to the received noise samples at the output of receiver chip matched filter and z(p) is the transmit chip vector for the current symbol period which is determined according to: z(p) S(p)y(p) wherein z(p -i) is the transmit chip vector for the previous symbol causing intersymbol interference which is determined according to: z(p -i)= s(p-')y(p -i) wherein the transmitted chip vector z(p + i) for the next (p//T + i) symbol period is: z(p +1) = ET S(p +1)y(p +1) wherein p)= [yI(p),,yk(p),,yK(p)tis the transmitted symbols vector which contains the symbols, over the symbol period p = 1,.. (x) , wherein NIX) is the number of transmitted symbols in a transmission block wherein the entire block of transmission is represented as an (NH x N,K) dimensional transmit symbol matrix using the following relationship: x = [x tk,",aN,Kj = [y(1),...,v(p),...v(NH)f.wherein X is the matrix conatining transmitted symbols, Xfr is the vector containg symbols over channel k, NK is the total number of spreading sequences used in the downlink MIMO system, and wherein the (NHx1)dimensional symbol vector is constructed using: = [x(1),. . . . . for each channel k = I AT7K.
12. 12. A method for determining data transmission rates for transmitting a signal over a radio transmission system having a plurality of parallel single-input single-output or multiple-input multiple output channels over which the signal is to be transmitted, the signal comprising one or more symbols, the symbols being spread prior to transmission by a plurality of spreading sequences on a symbol-by-symbol basis, the method comprising: receiving a signal comprising a plurality of symbols, each symbol having been spread by a spreading sequence of a plurality of spreading sequences; determining, for each symbol of the plurality of symbols, a system value Ak indicative of a signal-to-noise ratio associated with the respective spreading sequence used to spread the symbol; and averaging the plurality of system values associated with the plurality of spreading sequences to determine an averaged system value for determining a data rate for transmission of symbols using the plurality of spreading sequences.
13. 13. The method according to claim 12, further comprising: determining a data rate to be associated with the number of signature sequences for transmission in accordance with the average system value.
14. 14. The method according to claim 13, wherein the data late for each stream is determined by finding the discrete data rate value h,, which satisfies the following equation: [*] * )K1(4)< IpI1.wherein is the system value, which is for a streaming group number n and for symbol epoch and a target system value [A*],, , which is determined F(21'-i) according to [A j, = for p = 1,..., P for a chosen discrete bit rate of 1-F(2 P1) wherein F is the gap value.
15. 15. The method according to claim 13 or claim 14, further comprising estimating the symbols in the received signal in accordance with the following equation: 2(P) = S (Diag tV9C ( DY' £shon (P) Wherein rh(,V,(p) is the transmitted signal, SCkW is an extended channelization matrix and s.(p)is a scrambling sequence vector for symbol period p wherein the scrambling sequence vector s. (p) is determined according to: = (c/tag (i)® s) = [IHSC (i), ... (n)' + 2)1 and wherein the extended channelization matrix SCbW is determined according to: Srhw = ( ® s2)= [e,I wherein a total of K channelization spreading sequences SK are generated from a total of N Hadamard codes with spreading sequence processing gain N, and a signature sequence matrix s = j wherein a channelization spreading sequence vector s associated with the signature sequence matrix S55 has the property: 1 1=1H =0 i!=j and wherein a transmitter precoding matrix W is determined according to: w = Ji?N wherein the precoding vectors w,, forn, =1,... ,N with the property: =1 II j i!=I are the beam stearing vectors used at the transmitter antennas.
16. 16. The method according to any one of claims 12 to 15, wherein the system value is defined by: = - = -ET rrace(S'DZZ"n"S) N1KN'N" -E rrace(S"DZ" Z D"S) N,KN' N(T) k pi pl 1 k -Trace(S'ftZ' Z D'S) viciv'' yy(x) k 1l 1 A -N0 TraceBB")Nwherein D is the estimated channel characteristic converted to a square matrix representative of intersymbol interference, which is defined as follows: D=C' H"C H-diag(C.' H"C Lu) and a signature sequence matrix for channel k is defined as: Sk = fr WC = (p wherein an intersymbol matrix D1 for the interference from the previous symbol is determined according to: D1 = (cf' H"C'H1) and an intersymbol interference matrix D. for the interference from the next symbol is determined according to: D, = (c i-i11c i-i4 and B is a noise vector despreading matrix, which is determined according to: B=SC H"C.
17. 17. The method according to any one of claims 12 to 16, wherein the plurality of spreading sequences with which the average system value is associated corresponds to a first combination of receiver and antenna channels, wherein the method further comprises: determining an average system value for one or more other combinations of receiver and antenna channels; and using the receiver and antenna channel having the highest system value for transmission.
18. 18. The method according to any one of claims ito 11, further comprising the method of any one of claims 12 to 17.
19. 19. Apparatus arranged to perform the method according to any preceding claim.
20. 20. The apparatus according to claim 19, wherein the apparatus is a receiver device.
21. 21. The apparatus according to claim 19, wherein the apparatus is a radio transmission system complising one or more transmitters and one or more receivers.
22. 22. A computer program product arranged, in use, to instruct a computer to perform the method according to any one of claims 1 to 18.