System and method of a MIMO transmitter / receiver

1. A method of obtaining a plurality of received signals in a Multiple Input Multiple Output (MIMO) system with K transmission elements and N reception elements, wherein N≧K (N>2, K>1), said method comprising:

obtaining a channel coefficients matrix H;

obtaining a plurality of K transmission channels;

decomposing the channel coefficients matrix H using WB factorization, wherein W is a partial equalization matrix, chosen such that decomposing the channel coefficients matrix H results in a decomposed matrix B that is supplemented with a plurality of non zero diagonals exhibiting a bandwidth b; and

running an iterative process for obtaining K received signals while applying a Maximum Likelihood Estimation (MLE) algorithm on all sub-sequence combinations of K transmitted signals according to a transmitted signal vector propagated through the decomposed channel coefficients matrix B and wherein at each iteration stage, i (i={0 . . . K}), a likelihood value is calculated and updated in accordance with a single reduced received signal of index K−i calculated from combination of b transmitted signals of indices i to i+b−1 and a sub-sequence of reduced received signals of indices K−i+1 to K calculated from the combination of transmitted signals of matching indices.

2. The method according to claim 1 wherein the MIMO system uses a wireless connection.

3. A method of obtaining a plurality of received signals in a Multiple Input Multiple Output (MIMO) system with K transmission elements and N reception elements wherein N≧K (N>2, K>1), said method comprising:

A. obtaining a channel coefficients matrix H;

B. obtaining a plurality of K transmission channels and sorting thereof according to a first set of criteria;

C. rearranging said channel coefficient matrix according to the obtained order of said transmission channels;

D. decomposing said channel coefficients matrix H using a decomposition process using WB factorization, wherein W is a partial equalization matrix, chosen such that decomposing the channel coefficients matrix H results in a decomposed matrix B that is supplemented with a plurality of non zero diagonals exhibiting a bandwidth b;

E. defining first partition matrices obtained from said decomposed channel coefficients matrix; and

F. obtaining K received signals according to best combination of M estimated transmitted signals (M<K, M>1) of said decomposed channel coefficients matrix by applying an iterative process, wherein at each iteration is determined best combination of L out of M (L≦M) estimated transmitted signals of said partition matrices are determined according to second set of criteria and subtracting the effect of estimated signal from the received signal r and re-partitioning matrices according to determined L transmitted signals.

4. A method of simultaneously decoding a plurality of transmitted signals in a multiple input multiple output (MIMO) system having K transmission elements and N reception elements wherein N≧K (N>2, K>1) thereby providing a reduced complexity approximation of optimal maximum likelihood estimation of MIMO decoding, the method comprising:

receiving a plurality of transmit signals;

sampling the plurality of the received signals resulting in sampled received vectors;

obtaining a channel coefficients matrix H from the sampled received vectors;

obtaining, through computation of algebraic reduction on the matrix H, a reduced channel matrix B having a bandwidth b, such that B=W′H, wherein B is supplemented with a plurality of non zero diagonals, and wherein W is a partial equalization matrix, and wherein W′ is a Hermitian conjugate of the matrix W;

computing a reduced received vector z for each sampled received vector, wherein the sampled received vectors are reduced according to the bandwidth b of the reduced channel matrix B, such that z=W′x=Bt+W′n, wherein x is the sampled received signal vector, and wherein t is the transmitted signal vector and wherein n is a random additive noise vector;

setting i=N−b;

unless i=0, defining all possible combinations of the transmitted signals (ti+1, . . . , ti+b) in a trellis as survivor branches with corresponding metrics;

repeatedly decrementing i by 1 and for each possible value of transmitted signal (ti, . . . ti+b−1) and for each one of the survivor branches (ti+1, . . . , tN) that contain the subsequence of transmitted symbols (ti+1, . . . , ti+b−1), computing a corresponding branch metric (z1, . . . , zNti, . . . , tN) such that L(z1, . . . , zNti, . . . , tN)=L(zi+1, . . . zNti+1, . . . , tN)+L(ziti, ti+1, ti+b−1), wherein L is a log function;

when i reaches 0, choosing the transmitted signals along the survivor branch with the highest value of branch metric.