DRL-Enabled RSMA-Assisted Task Offloading in Multi-Server Edge Computing

The growing demand for efficient and reliable wireless communication has fueled interest in Rate-Splitting Multiple Access (RSMA) as an advanced multiple access technique for future networks. Simultaneously, Multi-Access Edge Computing (MEC) has become a transformative solution for addressing emerging applications' latency and computing challenges. This study explores the integration of RSMA and MEC to enable simultaneous offloading of users' tasks to multiple MEC servers. We formulate a computation offloading problem to minimize the delay experienced by all users within the RSMA-aided multi-MEC server environment. To tackle this problem, we employ Deep Deterministic Policy Gradient (DDPG), a deep reinforcement learning technique known for its effectiveness in dynamic environments. Simulation results validate the superior performance of the DDPG-based approach compared to conventional methods.