Skip to Main Content
Recently, research on space-time trellis coded system in frequency selective fading channels have received increased attention because of its achievable capacity and diversity. It has been shown that the maximum likelihood (ML) detector can achieve spatial, temporal and frequency diversity in such channels. However, it assumes the channel parameters are known perfectly at the receiver. Moreover, as the ML detector complexity grows exponentially with the number of channel taps (channel order), estimation of the channel order must be performed to reduce necessary computations. In [So, D.K.C. and Cheng, R.S., 2002], we proposed an iterative receiver derived from the expectation maximization (EM) algorithm, which performs channel order, channel gain and sequence estimation. The conditional model order estimator is employed for channel order estimation and is shown to have good performance. In this paper, we have 3 major contributions in improving the performance of this receiver: 1) a generalized CME criterion for MIMO systems, 2) a modified CME criterion for MMSE based EM receiver, and 3) an order estimate adjustment scheme to avoid overestimation. Simulation results show that the new modifications improve the order estimate significantly; hence reducing unnecessary computation due to overestimation.