Studying of Initial Weight Vector Selection Impact on Transient Performance of Coherent Sidelobe Cancellation System with Gradient Adaptation Algorithm

Coherent sidelobe canceller is considered as a special case of adaptive antenna arrays. Utilization of the steepest descent algorithm in such a system is studied. It is shown that eigenvectors of the auxiliary channels correlation matrix create two subspaces with so-called primed coordinates, which comprise the principal axes of the quadratic performance surface. The first subspace is formed by the eigenvectors corresponding to the singular eigenvalues, which equal channels' internal noise power, while the second is created by those corresponding to the distinct eigenvalues. Dimension of the second subspace equals the number of the jamming sources. It is proven that, if the initial weight vector is chosen zero, the trajectory of the transient process totally belongs to the second subspace. In this case, transient performance is good. If, for whatever reason, initial weight vector finds itself in the first subspace, the transient performance can be compromised. Computer simulation of the adaptive coherent sidelobe canceller has shown that this transient performance degradation can be substantial. Still, it lessens greatly if the number of the auxiliary antennas is much higher than that of the jammers. As a measure to improve performance, it is proposed to set to zero the initial weight vector in the systems with the small number of the auxiliary channels any time interference environment changes.