Joint Mixed Semantic Communication and Computing Offloading Framework for Mobile Edge Computing

The rapid growth of data-intensive applications has significantly intensified the scarcity of wireless communication resources, necessitating innovative approaches for efficient bandwidth utilization. Semantic communication, which transmits essential semantic meanings rather than raw bits, emerges as a promising paradigm to alleviate bandwidth constraints. However, semantic processing often introduces additional computational overhead, limiting its universal applicability. To balance the trade-off between bandwidth savings and computational complexity, this paper proposes a hybrid semantic communication and computation offloading framework for edge computing environments. Specifically, computationally intensive tasks are offloaded from user devices to edge servers via three flexible communication modes: traditional raw-data transmission, semanticaware compression, and full semantic communication. Specifically, the transmission mode adopted by each task is dynamically determined based on the currently available computing resources, bandwidth conditions, and the task’s own requirements. We develop an efficient resource management algorithm to solve the optimization problem. Simulation results show that the proposed framework significantly improves the resource utilization and overall system performance compared with traditional methods.