A Computation Offloading Strategy Based on Distributed Deep Learning

As innovative applications continue to emerge, numerous compute-intensive and latency-sensitive tasks are rapidly evolving. Computation offloading has become a key technology in the new paradigm of mobile edge computing, aimed at transferring tasks from mobile devices to edge servers or remote clouds to reduce system latency and energy consumption. In this context of industrial IoT terminal devices in edge computing scenarios, this paper formulates the problem of jointly optimizing offloading and resource allocation as a mixed-integer programming problem. Additionally, a distributed deep learning-based offloading algorithm (DDLOA) is introduced. This method employs multiple parallel deep neural networks (DNNs) to generate offloading decisions and resource allocation plans. Extensive experimental results demonstrate that the proposed DDLOA algorithm can improve the speed of task computation and reduce the energy consumption of terminal devices compared to comparative algorithms.