Waveform extraction from reference channels of passive multistatic radar systems

Multistatic passive radar systems utilize transmitters of opportunity such as television, radio broadcasters etc. or other non-cooperative illuminators such as military radars to remain covert and perform detection. They process data from two groups of channels, namely the reference and surveillance channels. These may in turn comprise one or more smaller sub- reference or sub-surveillance channels. Modern illuminators both civilian and military are digital and emit waveforms in finite repetition intervals. In this paper, our focus is estimating the waveforms present in the reference channels of passive radar systems, with an immediate goal of exploiting the finite waveform repetitions for detection, along with a long term goal of providing a waveform map to other cooperative radars in the immediate vicinity, in order to use this information for future surveillance passes. We consider three cases: (1) the waveform is unit rank, (2) the waveform for each transmission is randomly chosen from a P waveform ensemble, with each waveform being orthogonal to one another, and (3) that the waveform is an unknown linear combination of P unknown linearly independent vectors. A signal subspace framework is employed, maximum likelihood (ML) solutions are proposed. Simulations reveal, that the ML solution is biased for low SNRs. For some of the models used, the bias of the ML solution may be explained from random matrix theory.