DETECTION OF SIGNALS CORRUPTED BY NONSTATIONARY RANDOMNOISE VIA KALMAN FILTER-BASED STATIONARIZATION APPROACH

摘要

In this paper, a method of stationarization of nonsta-rntionary data is proposed in the signal detection problem.rnThe signal to be detected is corrupted in a nonstationaryrnrandom noise whose model is given by an ARMA(p, q)rnmodel. The time-varying coefficient parameters of thernARMA model are estimated by the Kalman filter. Thernstationalization of nonstationary observation data basedrnon the estimated coefficient parameters leads us to thernconventional binary hypothesis-testing for signals in sta-rntionary random noise.