Research On Frequency Conversion Task Offloading And Cache Incremental Optimization Under Edge-to-side Collaboration

Edge computing is a new field of cross-development and integrated applications such as cloud computing,Internet of Things,and 5G/6G mobile communications.It relies on edge resources deployed close to users and terminals to provide elastic resources and on-demand services on the business side,which can effectively reduce communication delay and interaction overhead,and is widely used in industrial Internet,smart city and other scenarios with high practical requirements.Different from the centralized resource management and standardized task processing mode of traditional cloud computing,edge computing faces the requirements and challenges of multi-domain and multi-layer distribution of resources and multi-party and multi-level collaboration of tasks.It is urgent to carry out task offloading and cache optimization for edge computing environments.Academia and industry have carried out a lot of theoretical research and practical application work on the topic of task offloading and cache optimization of cloud-edge-device collaboration in edge computing mode,but the existing solutions do not fully consider the mobility characteristics of device-side devices and user entities,there are problems such as the rigidity of the task offloading strategy and the passive lag of the cache scheduling mechanism.To this end,this thesis focuses on the collaborative task offloading and cache scheduling between the terminal layer and the edge layer in edge computing.Then based on the active perception and prediction of the entity’s motion trajectory and position on the terminal side,a side-to-end collaborative task frequency offloading scheme and cache incremental optimization method are proposed.Aiming at the problem of task offloading delay and jitter caused by the variable speed motion of device-side entities and the movement of cross-edge servers,this thesis studies the endto-end collaborative sensing of the device-side mobile entity co-location algorithm,and designs a fine-grained task allocation model that adapts to the entity’s motion trend.Then a task-variable-frequency unloading scheme based on entity movement trajectory is proposed.Theoretical analysis and experimental tests show that the task frequency offloading scheme proposed in this thesis can effectively smooth the service delay jitter during the offloading process of mobile entity tasks,reduce the delay effect by about 30% compared with the comparative scheme,and effectively improve service quality and user experience.Aiming at the problem of low edge cache hit rate caused by the dynamic movement of device-side entities across edge servers,this thesis studies a multi-entity location prediction algorithm for device-side movement based on co-location.relationship attribute,a cache scheduling update model based on batch file priority is designed,and an adaptive incremental cache optimization scheme is proposed.Theoretical analysis and experimental tests show that the cache incremental optimization scheme in this thesis has the parallel processing capability of batch entity cache scheduling update.Compared with the comparison scheme,the hit rate of cache files in the scenario of batch moving entities is improved by about 14%.Based on the above research,this thesis uses virtualization technologies such as containers to design and develop an edge computing prototype verification system based on task frequency conversion offloading and cache incremental optimization,which realizes the task and resource management and control of peer devices and edge servers.The characteristics of task offloading and cache file request of batch terminal equipment verify the practical feasibility and effectiveness of the research scheme in this thesis,and the related work has successfully supported practical applications such as avionics systems and community platforms.