THIRD ANNUAL
ASAP '95 WORKSHOP

Efficient Multichannel
Space-Time Whitening
Filters

S. Unnikrishna Pillai
Polytechnic University
Department of Electrical and Computer Engineering
Brooklyn, NY 11201

Abstract New multichannel whitening filter design methodologies are introduced with the purpose of providing more "efficient'' space-time interference suppression than traditional full-degree-of-freedom adaptive strategies, such as the batch sample matrix inverse (SMI) approach. Efficiency is measured in terms of the degree of interference suppression achieved versus both computational complexity and statistical rate of convergence. This latter design consideration stems from the need to minimize the number of adjacent range-bin samples used to form the requisite statistics, thereby relaxing the interference "stationarity'' assumption implicit in all space-time adaptive processing (STAP) methodologies.

This presentation introduces two distinct multichannel space-time design methodologies. The first technique is based on incorporating ideas from multichannel statistical data compression, viz., multichannel "predictive'' and "transform'' coding. The resulting "predictive-transform'' (PT) space-time array processor is presented which optimally integrates two distinct array processing strategies, viz., optimum multichannel sidelobe cancellation ("prediction'') and optimum beamforming ("transformation''). The second method is based on a recently developed, comprehensive multichannel system identification technique which guarantees stable and rational approximations. The technique is based on the powerful concepts of positive functions and bounded functions. The efficiency of these approaches is illustrated via an application to adaptive airborne MTI radar where it is demonstrated that better (and near optimal) interference suppression is achieved than the full-degree-of-freedom sample matrix approach, with less implementation complexity, when reasonable limitations on the interference "stationarity'' assumption are imposed.