Dependent Task Offloading of Cross-Applications and Service Caching for End-Edge-Cloud Computing in Consumer Electronics

As consumer electronics (CEs) continue to become more intelligent and interconnected, end-edge-cloud collaborative computing is an essential paradigm to meet the demands of cross-application dependent tasks. However, the limited memory resources and heterogeneous caching capabilities of edge servers intensify competition in multi-dependency task offloading and resource allocation, increasing the complexity of decision-making. To address the above issues, we first propose a Memory Resource Management Strategy (MRMS) that prevents system overload by maintaining server memory utilization within a safe threshold, thereby achieving efficient use of server resources. Then, we propose a Predictive model-based Optimal Edge Server selection (POES) algorithm, which improves offloading decision accuracy by predicting task completion times and selecting the optimal offloading location. Building on POES and the MRMS strategy, we propose a Multi-Application Dependent Task Offloading Strategy (MA-DTOS) algorithm. Our MA-DTOS algorithm incorporates the heterogeneous caching capabilities of edge nodes, determines the execution order based on task priorities, and formulates optimal offloading decisions for each task, thereby minimizing the total completion time of all tasks. Experimental results demonstrate that MA-DTOS reduces the average task completion time by approximately 45% and enhances server resource utilization by about 14.6% compared to existing benchmark algorithms.