Joint data detection and channel estimation for two dominant users in the asynchronous GMAC with memory

In this paper we consider the situation of two strongly interfering signals, superimposed over a background of AWGN. We consider the situation where User 2psilas packet arrives initially and we are able to estimate accurately this userpsilas channel impulse response. Then a packet from User 1 arrives during the data portion of User 2psilas packet. We propose an algorithm for simultaneously estimating the channel impulse response (CIR) of User 2 and detecting the data of User 1, i.e. joint data detection and channel estimation for two dominant users in the asynchronous scalar Gaussian multiple access channel (GMAC) with complex-valued intersymbol interference (ISI). This approach proves to be useful for reestimation of User 2psilas CIR, after cancellation of interference from User 1. We demonstrate the algorithm with simulation results. We give an extension of the algorithm where the estimate of User 1psilas data is used for interference cancellation to refine the estimate of User 2psilas channel. In simulations, this is shown to improve the channel estimate. We also give reduced rank approximations of the algorithm, using the conjugate gradient method, which are more suited to real time applications and which implement noise-reduction for an over-modeled system.