A stochastic analysis of an iterative semi-blind beamformer for TDMA systems

Maximum Likelihood Optimization of Adaptive Asynchronous Interference Mitigation Beamformer

“…This means that in addition to the training signals and known a priori or estimated with the help of the training signals channel propagation matrix H, we can estimate the signal-free interference plus noise covariance matrix Rt on the training interval and the covariance matrix Rd of the mixture of interference and signal on the data interval. Thus, if we confine our adaptive search of the optimal mixing factor δ in (13) only by the second order statistics, no much more measurable parameters are left in our disposal. The main reason is that in terms of their correlation properties, the useful signals on the data interval are indistinguishable from CCI, which is often created by very similar signals.…”

“…1) the basic one as in (13) without any "cleaning" marked as "Initial"; 2) the one with "cleaning" of the data interval with the initially estimated desired signal…”

“…At the same time, no significant difference in LR t behavior between different types of "cleaning " in (37) -(39) could be observed. This is because for high enough values of δ in (13), the presence of the useful signals in Rd in (11) leads to rejection of the training data, but even the soft "cleaning" in (37) reduces the signal component in the covariance matrix estimate to the levels that do not significantly affect the extracted training signal accuracy.…”

“…A possibility to improve performance in such scenarios using constant modulus (CM) algorithms is pointed out in [10], [13]. Similar improvements can be achieved using algorithms with projections to the finite alphabet (FA) [22].…”

“…The asynchronous CCI scenario is addressed in time division multiple access (TDMA) networks [9] - [13], wireless local area networks (WLAN) [14], [15], long term evolution (LTE) networks on the downlink [16] and on the uplink [17]. Significant performance degradation for the conventional training based IRC in the asynchronous CCI scenario is reported in these papers.…”