Shrinking the Insecure Area for Satellite Downlink using Multi-Beam Intersection

Satellite communication faces a great risk of eavesdropping due to the its broadcasting nature and vast coverage. Conventional beamforming-based physical-layer security technology encounters difficulty in preventing eavesdropping within the main-lobe of the satellite, which causes a large insecure area. In this paper, we propose a cooperative multi-beam transmission framework, which elaborately superimposes multiple beams to shrink the intersected coverage area, and exploits the additional design degrees of freedom to enhance the security. The designs of symbol mapping and beamforming are jointly studied at multiple beams to achieve transparent communication in the intersected area while randomizing the signals otherwise. Information-theoretic performance metric is identified to characterize the information leakage. A tight upper bound of the intractable metric is proved using variational methods, and a deep learning-based approximation and optimization scheme is then proposed to efficiently design the multi-beam transmit signals. To further distort the leaked signal, a convergence-guaranteed cross-entropy algorithm is developed for joint antenna subset selection at multiple beams. Simulations show that the proposed scheme can shrink the insecure area by an order of magnitude compared with the conventional methods.