Algorithms for passive SAR imaging

Passive synthetic aperture radar (SAR) imaging using digital broadcast signals of opportunity presents several signal processing problems. The transmitted signals are continuous rather than pulsed and are interspersed with training sequences. In addition passive SAR systems that exploit such broadcast signals will need to operate in conditions of extremely low signal to noise ratio. In this paper passive SAR imaging has been formulated as a linear regression problem for which convolution back projection can be modified to provide an approximate solution. The performance of this solution is limited by the point spread function of the random signature waveform associated with each scatterer on the ground. A forward regression technique based on the othogonalised least squares algorithm is proposed as a relatively efficient method of solution both in terms of computational and statistical efficiency.