Optimal and suboptimal approaches for training sequence based spatio-temporal channel identification in colored noise

In wireless communications (with particular regard to digital mobile radio), spatial (via antenna arrays) and temporal (excess bandwidth) diversity may be exploited to simultaneously equalize a user of interest while canceling or reducing (cochannel) interfering users. This can be done using the interference canceling matched filter (ICMF) which we introduced previously. The ICMF depends on the channel for the user of interest, to be estimated with a training sequence, and a blind interference cancellation part. The critical part is the channel estimation. The usual least-squares method may lead to poor estimates in high interference environments. Significant improvements may result from the maximum-likelihood (ML) and suboptimal techniques investigated here.