Probabilistic Semantic Communication Enabled MEC Systems for Low-Altitude Economy

Abstract

The burgeoning low-altitude economy has triggered a surge in computation-intensive applications, posing challenges for resource-constrained ground devices and bandwidth-limited low-altitude wireless networks. To address these bottlenecks, we investigate an unmanned aerial vehicle (UAV)-enabled mobile edge computing (MEC) system that is enhanced by probabilistic semantic communication to alleviate transmission burdens. We formulate the joint semantic offloading and resource allocation problem, aiming to minimize the average system energy consumption while ensuring the resource capacity and task latency constraints. To solve the problem, we first reformulate it into Markov decision process. Then, we propose the adversarial branching dueling double Q-network (ABD) algorithm with a constraintaware action masking mechanism to find the optimized solutions from action space with high-dimensional and non-convex nature. Numerical results demonstrate that the proposed ABD algorithm is capable of achieving stable convergence. Meanwhile, the ABD algorithm outperforms conventional baseline schemes in terms of energy efficiency, and task completion throughput.