Power Allocation and Beamforming Design for IRS-aided Secure Directional Modulation Network
Rongen Dong, Feng Shu, Fuhui Zhou, Yongpeng Wu, Jiangzhou Wang
TL;DR
The paper tackles secrecy-rate maximization in an IRS-aided secure directional modulation network by jointly optimizing the power allocation factor $α$, CM beamforming, AN beamforming, and IRS phase shifts $Θ$. It introduces two alternating-optimization schemes: Max-SR-HP, which uses derivative-based PA, generalized Rayleigh-Ritz for CM, generalized power iteration for AN, and SDR for IRS design; and Max-SR-LC, which uses Max-SLNR/Max-ANLNR for beamforming and SCA for IRS optimization to reduce complexity. Numerical results show both approaches significantly boost SR over baseline schemes such as equal PA, no IRS, and random-phase IRS, with Max-SR-HP offering higher performance at the cost of higher complexity. The findings highlight the effectiveness of jointly optimizing PA, beamforming, and IRS in enhancing physical-layer security for directional modulation networks. The work has practical implications for secure wireless networks employing IRS to control and distort signals in the presence of eavesdroppers.
Abstract
With the aim of boosting the security of the conventional directional modulation (DM) network, a secure DM network assisted by intelligent reflecting surface (IRS) is investigated in this paper. To maximize the secrecy rate (SR), we jointly optimize the power allocation (PA) factor, confidential message (CM) beamforming, artificial noise (AN) beamforming, and IRS reflected beamforming. To tackle the formulated problem, a maximizing SR with high-performance (Max-SR-HP) scheme is proposed, where the PA factor, CM beamforming, AN beamforming, and IRS phase shift matrix are derived by the derivative operation, generalized Rayleigh-Ritz, generalized power iteration, and semidefinite relaxation criteria, respectively. Given that the high complexity of the above scheme, a maximizing SR with low-complexity (Max-SR-LC) scheme is proposed, which employs the generalized leakage and successive convex approximation algorithms to derive the variables. Simulation results show that both the proposed schemes can significantly boost the SR performance, and are better than the equal PA, no IRS and random phase shift IRS schemes.