A Cache-assisted Computing Offloading Strategy Based on Deep Q Network

Mobile edge computing (MEC) provides users with abundant wireless resources and cloud computing capabilities to meet their computing demand. Existing works tend to consider caching and computing offloading separately, so it is difficult to achieve overall optimization of system performance. To further improve system performance in smart home scenario, a novel collaborative caching and computing offloading scheme (CCCO) was proposed in this paper. First, a new collaborative caching strategy is designed in this paper to improve the cache hit rate, i.e., smart devices cache the task's computation results and edge servers collaboratively cache the related data of the sub-tasks after task division, Then, sub-tasks are collaboratively offloaded to servers for processing. Finally, Deep Q Network algorithm is used to obtain the optimal offloading and caching decisions for minimizing system latency. Simulation results show that the proposed algorithm significantly outperforms the traditional computing offloading scheme in terms of latency.