Communication system and method for performing fast symbol estimation for multiple access disperse channels

A method for estimating at least a subset of symbol sequences from received signals at multiple receivers in a multiple receiver communication system wherein each of said multiple receivers receives a sum of signals from a plurality of transmitters sharing the same communication medium, wherein said signals are formed by a sequence of contiguous symbols, each of which occupies a fixed time duration, and samples said received signals with a time interval that is the inverse of an integer multiple of said fixed time duration, wherein said integer multiple is no smaller than the number of said signals divided by the number of said receivers, wherein the channels of said signals have finite impulse response and are different from one another, said symbol estimation method comprises:

• (a) sampling said received signals at multiple receivers with a time interval equal to a
• 1
• F
• fraction of said fixed time duration, wherein F, an integer is no smaller than
• J
• M
• ⁢
• ⁢
• M
• t
• ,
• wherein J is the number of said transmitters, M is the number of said receivers, Mt is a positive integer, and generating Q sampled received signals from each of said received signals, wherein Q is an integer no larger than Mt;
• (b) constructing data vectors from sampled received signals by filtering them with channel responses between said transmitters and said receivers and multiplexing outputs of said filtering;
• (c) obtaining the maximum length of the channel responses of users at multiple receivers, which is denoted as d+1, and constructing a channel response matrix for each receivers;
• (d) constructing space-time correlation matrices with said channel response matrices;
• (e) constructing a second set of matrices and vectors computed from said space-time correlation matrices and said data vectors;
• (f) applying a recursive bisection approach on said second set of matrices and vectors to generate an output sequence, wherein said recursive bisection approach comprises:
• forward division steps which recursively compute a first series of matrices and vectors initially based on said second set of matrices and vectors and yield vectors in said first series of vectors and matrices with dJ└Li/2┘ rows where dJLi is the number of rows of the last element in said first series of vectors and matrices, until a pre-defined stopping criterion is satisfied, where └x┘ denotes the floor function evaluated on x, wherein Li is an integer variable which is updated in each of said forward division steps,
• backward substitution steps which first construct an intermediate solution with the last of the first series of matrices and vectors and recursively compute, in a reverse order of said forward division steps, a second series of vectors with said first set of matrices and vectors until exhausting all of said first set of matrices and vectors;
• and
• (g) demultiplexing the vector from the last in said second series of vectors;
• whereby said symbols of said signals are found from said output sequence.