Georgios B. Giannakis
Sergio Barbarossa
Dept. of Elec. Eng., Univ. of Virginia, Charlottesville, VA 22903-2442
The analysis of polynomial phase signals using the recently developed polynomial phase transform suffers from spurious harmonics when dealing with multiple components having identical high-order coefficients. It is illustrated that such a model appears in synthetic aperture radars (SAR) imagery and transmission of polynomial phase signals through multipath channels. Analytical expressions for the spurious peaks are derived first, followed by algorithms for estimating the parameters of multicomponent polynomial phase signals taking advantage of the overdeterminacy introduced by the so called multilag high-order ambiguity function. An algorithm will be presented based on the projection of the observed signal onto a subspace estimated as the intersection of signal subspaces obtained using different sets of lags. Fast suboptimal algorthms will also be presented for suppressing interference terms and removing ambiguities. Autofocusing of real SAR imagery and mitigation of multipath when acoustic signals arrive at moving receivers, will be used to illustrate the usefulness of the proposed techniques.