A Novel Classification and Identification Scheme of Emitter Signals Based on Ward’s Clustering and Probabilistic Neural Networks with Correlation Analysis

Algorithm 2

Classification Algorithm 2.

	Input: The original signal vectors that need to be classified.
	Output: The classified label vector .
	Compute by using self-adaptive filtering for ;
	Compute frequency spectrum of ;
	Compute Ward’s clustering dendrogram;
	Compute CH(K), Silh(K), DB(K) and Gap(K); Compute Kmin, Kmax;
	if Kmin = Kmax, then
	Compute center of each class ;
	Select training samples around and record their labels ;
	Create the PNN classifier by using and ;
	Determine the class of the remaining samples;
	Output the classified label of ;
	else for K = Kmin : Kmax
	for k = 1 : K
	Compute center of each class ;
	Select samples around and record their labels ;
	Create the PNN classifier by using and ;
	Determine the class of the remaining samples;
	Compute matrix and for each K;
	Compute and the optimal number of classifications ;
	Output the classified label of ;
	end
	end
	End: Classification Algorithm 2.