Cross correlation and energy detection in multiarray processing

Multiple arrays can be modeled as a single array that is composed of separate subarrays. Likelihood ratio testing, maximum likelihood delay-Doppler estimation, and maximum likelihood estimation of signal and noise parameters are applied to such a model in order to determine whether cross correlation should be used alone or in conjunction with other operations. The results indicate that cross correlation alone is usually suboptimum. Detection or combined detection-estimation should use a weighted sum of pairwise cross correlations between subarrays and energy detection at each subarray. Subarray cross correlation alone is only optimum for certain estimation problems and time windows. Cross-correlator and energy detector outputs at different frequencies should be linearly combined using frequency-dependent weighting functions. These functions are obtained for multiarray processing. For the special case of two-array cross correlation, the weighting function is the same as the one obtained by Knapp and Carter.