A new method for blind source separation of nonstationary signals

Many algorithms for blind source separation have been in-troduced in the past few years, most of which assume sta-tistically stationary sources. In many applications, such as separation of speech or fading communications signals, the sources are nonstationary. We present a new adaptive al-gorithm for blind source separation of nonstationary signals which relies only on the nonstationary nature of the sources to achieve separation. The algorithm is an efficient, on-line, stochastic gradient update based on minimizing the av-erage squared cross-output-channel-correlations along with deviation from unity average energy in each output chan-nel. Advantages of this algorithm over existing methods in-clude increased computational efficiency, a simple on-line, adaptive implementation requiring only multiplications and additions, and the ability to blindly separate nonstationary sources regardless of their detailed statistical structure. 1