A Novel Grouping-Based Rate-Splitting Multiple Access Scheme Against Internal Eavesdropping

In this paper, we propose a novel grouping-based rate splitting multiple access (G-RSMA) against internal eavesdroppers (Eves). Specifically, after users are grouped, their messages are split and encoded into inner-group common and private streams. The inner-group common streams could achieve the dual use: they deliver designated information to legitimate users and serve as artificial noise (AN) to Eves. To ensure the security among users in the presence of internal Eves, our approach maximizes the minimum secrecy rate (MSR) of all legitimate users in the system, which is formulated as a non-convex problem. We tackle it with a novel algorithm leveraging successive convex approximation (SCA), enabling efficient optimization of the transmit precoding scheme. Numerical results show that the G-RSMA approach enhances the MSR performance compared with the one-layer rate splitting multiple access (RSMA) or space division multiple access (SDMA), validating its effectiveness in securing RSMA systems against internal eavesdropping.