Multi-Server Assisted Task Offloading and Resource Allocation for Latency Minimization in Thermal-Aware MEC Networks

As the 6G network advances, the integration of sophisticated data mining techniques within the Consumer Internet of Things (CIoT) intensifies challenges in CPU temperature management and power efficiency. The growing use of consumer electronics and applications is also expected to drastically increase data traffic. Consequently, the substantial computational demands of these technologies risk causing CPU overheating, particularly given the limited capacities of Mobile Edge Computing (MEC) servers and IoT devices. Addressing these issues requires innovative resource allocation strategies that effectively manage CPU temperatures while maintaining robust system performance. This paper introduces the Thermal-aware Offloading and Resource Allocation Strategy (TORAS), specifically designed to minimize total execution latency for users across multi-server MEC networks by tackling a complex mixed-integer nonlinear programming problem. We streamline this problem into manageable subproblems: offloading ratio selection and resource allocation optimization, which are alternately optimized. Our approach includes deriving closed-form solutions for offloading ratio selection and transmission power allocation, and employing the primal-dual method to determine the optimal resource allocation strategies. This methodology not only addresses the thermal and operational challenges in MEC environments but also enhances latency performance, achieving improvements of 5.2% and 9.7% compared to other schemes, respectively.