Antenna array training and adaptation techniques in an unpredictable and uncontrolled interference environment

We investigate the performance of an array processing technique based on sample matrix inversion (SMI) and the effects of channel estimation (using various training schemes) to combat intermittent interference. Most current applications of broadband wireless communication systems use short data block lengths. Therefore, antenna weights estimation is usually done only once and these weights are used for the whole packet length. Through simulations, we show that, with an intermittent kind of interference, weights estimation based on a preamble or post-amble only scheme fails to track suddenly appearing interferers, resulting in a degradation to the output SINR. Two new, training-based, channel estimation techniques are presented which show superior performance over the training schemes in this type of environment. For simulating intermittent interference traffic streams, a batch Poisson traffic model is used.