QoS-Aware Augmented Reality Task Offloading and Resource Allocation in Cloud-Edge Collaboration Environment

The integration of Augmented Reality (AR) into mobile devices has sparked a trend in the development of mobile AR applications across diverse sectors. Nevertheless, the execution of AR tasks necessitates substantial computational, memory, and storage resources, which poses a challenge for mobile terminals with limited hardware capabilities to run AR applications within a constrained time. To address this issue, we introduce a mobile AR offloading approach in the cloud-edge collaboration environment. Initially, we break down the AR task into a series of subtasks and gather characteristics related to hardware, software, configuration, and runtime environments from the edge servers designated for offloading. Utilizing these characteristics, we build an AR Subtask Execution Delay Prediction Bayesian Network (EPBN) to forecast the execution delays of various subtasks on different edge platforms. Following the predictions, we frame the task offloading as an NP-hard Traveling Salesman Problem (TSP) and propose a solution based on Particle Swarm Optimization (PSO) heuristic algorithm to encode the offloading strategy. Comprehensive experiments have demonstrated that the prediction performance of the EPBN surpasses the other baselines, and PSO approach can reduce offloading latency effectively.