Results on Canceller Convergence in Nonstationary Noise

摘要: Abstract : Convergence results for the Sampled Matrix Inversion (SMI)/Gram- Schmidt (GS) canceller algorithm in nonstationary noise is investigated by using Gram-Schmidt as an analysis tool. Lower and upper bounds convergence rate of canceller's average output power residue normalized to quiescent are derived. These a function number independent samples processed per channel (main auxiliary), auxiliary input channels, external environment. The environment was modeled Gaussian, with level specified at each sampling time instant. Furthermore, this model generalized sense that joint probability distribution defined levels over processing batch. This leads capability modeling evaluating SMI/GS variety interference scenarios such continuous or discrete processes mix these.