Particle filtering algorithm for detecting a maneuvering target

Particle filter on track-before-detect algorithms are widely used for detecting of the radar targets at low signal-to-noise ratio. A track-before-detect algorithm based on the information about target´ distance, velocity and acceleration is considered. We introduce a likelihood relation that takes into account the multi-channel with respect to acceleration structure of the preprocessing system. We show that the proposed algorithm enables the detection of a non-uniformly moving target at SNR 0.5 to 1.0 dB.