A Multiobjective Resource Allocation Strategy for Personal Healthcare Digital Twins in Dynamic Edge Environments

This paper proposes an adaptive multi-objective optimization approach based on the Multi-Objective Evolutionary Algorithm based on Decomposition (MOEA/D) to manage offloading decisions under dynamic conditions. By considering both environmental variability and task heterogeneity, the proposed strategy improves system robustness, responsiveness, and efficiency, enabling PHDT systems to deliver reliable and timely healthcare services in resource-constrained, real-time computing environments. MOEA/D is selected as the baseline due to its proven effectiveness in balancing conflicting objectives and its widespread adoption as a benchmark in multi-objective optimization research. Its decomposition-based framework also provides a solid foundation for integrating our adaptive mechanisms for dynamic PHDT scenarios. In our experiments, CAMO-MOEA/D reduced ACT by up to 13.26%, AEC by 23%, and the SRT average penalty by 67.03% over the worst-performing baseline, all while preserving balanced exploitation of low-latency end devices.