Research On Cache-assisted Collaborative Computing Offloading Strategy

With the development of the Internet,there has been an increasing number of complex applications that require low computing latency and energy consumption.Traditional mobile cloud computing attempts to offload these computation-intensive tasks to the cloud center for processing.However,due to the distance from the smart devices,the quality of experience for latency-sensitive applications cannot be guaranteed.Mobile edge computing is considered a feasible solution to address this issue.Existing research primarily focuses on optimizing collaborative offloading strategies,but didn’t address how to effectively utilize the caching resources of servers to further reduce the response latency of offloading,thus providing a better quality of users’ service.This thesis proposes a collaborative computation offloading scheme designed for smart home scenario,aiming to optimize performance by minimizing system latency.Firstly,the scheme introduces a cache-assisted collaborative computation offloading algorithm,which effectively improves the utilization of cache resources through the proposed cache update strategy and centralized cache management mechanism.Secondly,by jointly considering cache and offloading decisions,unnecessary data transmission is reduced,further reducing system latency.Lastly,the DQN algorithm is used to solve optimization problems in high-dimensional state spaces,enhancing decision accuracy and efficiency.Based on the smart city with subtasks dependencies,a cache-assisted computing offloading scheme based on task queue sorting algorithm is proposed.And it can reduce waiting latency by making dependent applications independent of each other through a multi-application decoupling strategy,and by making the independent sub-tasks execute in parallel through a subtask parallel offloading scheme.Secondly,it combines genetic algorithms with deep reinforcement learning algorithms to solve optimization problems step by step,resulting in optimal offloading and caching decisions.Simulation results demonstrate that this approach effectively reduces the average response latency of tasks in the system.Through theoretical analysis and experimental simulations,we have demonstrated that the proposed cache-assisted collaborative computing offloading schemes for both scenarios can effectively reduce the computation latency of the system.