Secrecy capacity optimization in MIMO systems via optimal power allocation and artificial noise

Abstract

To overcome the fundamental limitation of conventional artificial noise (AN) — which requires strict orthogonality to the legitimate channel and thus wastes transmit power — this paper proposes a novel non-orthogonal AN scheme named Secrecy Capacity Optimization-Artificial Noise (SCO-AN). Unlike all existing AN methods that either enforce orthogonality or rely on heuristic power allocation, SCO-AN deliberately allows the AN to leak into the legitimate channel and then statistically optimizes this leakage. The key novelty lies in two aspects: (i) a rigorous proof that non-orthogonal AN can achieve higher secrecy capacity than orthogonal AN under the same total power constraint, and (ii) a closed-form, algorithm-free optimal power allocation coefficient derived directly from the law of large numbers, eliminating the need for iterative optimization. Under the MIMO scenario, an ergodic lower bound on the secrecy capacity is derived, and the optimal power allocation coefficient is obtained via statistical analysis without any additional algorithm. This work fundamentally reinterprets the role of AN: from a purely harmful interference to a controllable and optimizable resource.