Programmable canonical correlation analysis: a flexible framework for blind adaptive spatial filtering

We present a new framework known as the programmable canonical correlation analysis (PCCA) for the design of blind adaptive spatial filtering algorithms that attempt to separate one or more signals of interest from unknown cochannel interference and noise. Unlike many alternatives, PCCA does not require knowledge of the calibration data for the array, directions of arrival, training signals, or spatial autocorrelation matrices of the noise or interferers. A novel aspect of PCCA is the ease with which new algorithms, targeted at capturing all signals from particular classes of interest, can be developed within this framework. Several existing algorithms are unified within the PCCA framework, and new algorithms are derived as examples. Analysis for the infinite-collect case and simulation for the finite-collect case illustrate the operation of specific algorithms within the PCCA framework.