Optimal cooperative jamming for security

In this paper, based on a cooperative jamming scheme, the design of the optimal covariance matrix of the artificial noise produced by a helper that maximizes the secret rate between a source and legitimate destination pair in the presence of multiple eavesdroppers is addressed. It is shown that the original nonconvex design problem can be transformed into a sequence of convex optimization problems. The rank property of the optimal covariance matrix is investigated. In the presence of multiple eavesdroppers, the optimality of beamforming is established. Further, in the presence of two eavesdroppers, it is shown that generally the zero-forcing solution does not maximize the secret rate. Finally, the robust design under imperfect channel state information is considered.