Iterative Convex Optimization of Multi-Beam Directional Modulation With Artificial Noise

A joint optimization algorithm for designing multi-beam directional modulation (MBDM) symbols with artificial noise (AN) is proposed for physical layer security in wireless communication systems. The proposed approach guarantees specific minimum error probabilities along the given eavesdropper directions. The algorithm is implemented by iterating between a convex program that optimizes symbols with AN and a convex (linear) program optimizing a set of weight vectors to guarantee specific error probabilities along the eavesdropper directions. The proposed method is compared against several MBDM algorithms, and is found to provide excellent bit error rate performance results. A hardware experimental test bed is described and is used to demonstrate the feasibility of the proposed solution.