Sphere decoding in the presence of channel uncertainty

The performance of maximum likelihood (ML) and sphere decoding (SD) are investigated when the channel is uncertain. In this work, a modified low complexity approach is proposed to deal with the effects of the channel uncertainty and obtain robust ML and SD. A higher complexity approach is also suggested but is not investigated here as it appears to be unsuitable for real time applications. The approach adopted involves linearly preprocessing the received signal to suppress the interference caused by channel uncertainty before applying a detector. Parallels of this approach may be found in the channel shortening literature where it is suggested to linearly preprocess the received signal with a channel shortener to reduce the effective delay spread of the intersymbol interference (ISI) channel before applying a reduced state Viterbi decoder to bring down the complexity of ML decoding. In our case, however, the preprocessing is applied to reduce the effect of the interference caused by channel uncertainty.