Adaptive Filtration of Radio Source Movement Parameters Based on Sensor Network TDOA Measurements in Presence of Anomalous Measurements

Abstract

The methods based on TDOA measurements find wide application for localization of radio sources using wireless sensor networks. The need of taking into account the presence of anomalous measurement results often occurs in real conditions. Their appearance means a significant malfunction of sensor network components that results in divergence of traditional algorithms of Kalman filtration of radio source movement parameters. Based on the mathematical tools of mixed Markov processes in discrete time domain, the optimal and quasioptimal algorithms of adaptive filtration of radio source movement parameters were synthesized on the basis of TDOA measurements of sensor network in the presence of anomalous measurements. The optimal algorithm describes the evolution of joint a posteriori probability density of the vector of movement parameters and switching variables determining the type of measurement errors of network sensors. The quasioptimal algorithm obtained by linearization of the measurement equation involves the implementation of sequential technique of incoming data processing and performance of the Gaussian approximation of a posteriori probability density of radio source movement parameters. For the case considered in this paper using the statistical simulation, the developed quasioptimal algorithm makes it possible to recognize the appearance of anomalous errors of measurements with probability close to unity and eliminate their impact on the accuracy of determining the radio source movement parameters.