Fast Multichannel Blind System Identification using Laguerre Filters

Multiple channel blind system identification (MBSI) is often used in applications where the input signal cannot be measured and its statistical properties are unknown. Traditionally, the channel dynamics are modeled using finite impulse response filters. The number of model parameters can be significantly reduced if the filters are expanded onto a suitably chosen expansion basis, such as the discrete Laguerre filters, but several tuning parameters must be chosen correctly. This paper describes an efficient implementation of the Laguerre MBSI technique that allows for the rapid evaluation of many possible basis expansions, and hence tuning parameters. Monte-Carlo simulations compare the performances of the traditional and fast implementations.