A Physical Layer Key Generation Scheme Based on Deep Learning Compensation and Balanced Vector Quantization

Algorithm 1

Balanced vector quantization algorithm based on ILP.

	Input:
	Alice’s and Bob’s CSI sequences and .
	Number of quantified regions .
	Output:
	Quantification results (Alice and Bob’s initial keys).
(1)	Alice uses the kmeans++ algorithm to get initial cluster centers: ;
(2)	repeat
(3)	Sample assignment: using the simplex algorithm to solve (6) to assign evenly to the cluster centers;
(4)	Cluster center update: Alice updates the cluster center according to (7) and the sample assignment result;
(5)	until Alice’s new cluster centers are the same as in previous iterations;
(6)	Alice sends the final clustering centers to Bob;
(7)	Bob assigns evenly according to the received cluster centers by using the simplex algorithm to solve (6);
(8)	Alice and Bob get the region index of the sample points according to the final assignment matrix and then quantize the index to get the original key sequence.