Space/time/frequency methods in adaptive radar - New Jersey ...

69The scalar beamformed noise estimator iswhere w is the N dimensional weight vector. The array output is3.2.5 EigencancelerThe sample covariance matrix may be decomposed into two orthogonal subspacesdenoting the interference subspace, Q r , and the noise subspace, Q. This is shownin Equation 3.15. The eigencanceler selects its weight vector to lie in the noisesubspace as this space is orthogonal to the interference subspace. It is shown in [17]that the weight vector is written aswhere k is a complex constant and I is the identity matrix of dimension N.Using the Eigencanceler to select the weight vector produces higher values of thesignal-to-noise ratio compared to the SMI. The Eigencanceler requires only K 2rsamples to obtain the sample covariance matrix and achieve a signal-to-noise ratiowithin 3 dB of the optimal.3.2.6 Fixed TransformsA technique based on a fixed transformation is also used with maximum-likelihoodadaptive nulling[15]. The basis of this technique is 'Two-Step Nulling' by Marshall[50]. In this technique, the steering vector and sample covariance matrix are transformedto a new domain using a fixed transform T,