CGWSA: A Novel Strategy for Task-Dependent Load Balancing in Distributed Systems

Distributed systems form the critical infrastructure supporting high-performance computing and complex simulations, with effectiveness heavily dependent on load balancing strategies. Within these systems, distributed simulation tasks present unique challenges through strict temporal dependencies and sequential constraints that traditional methods struggle to address. This paper proposes CGWSA - a Color-Graph Grey Wolf-Simulated Annealing hybrid algorithm - that fundamentally advances load balancing for distributed simulation workloads. Our methodology introduces a comprehensive dual-layer framework that considers both node heterogeneity and intricate task dependencies: 1) Color-Graph preprocessing that categorizes tasks by resource dominance patterns, enabling efficient parallelization while preserving execution priorities; and 2) A bio-inspired optimization engine that combines Grey Wolf Optimizer's hierarchical search capabilities with Simulated Annealing's probabilistic acceptance mechanism to prevent local optima trapping. Experimental results on simulated computing environments demonstrate CGWSA's superiority with the lowest load balance degree of 1.678 and optimal makespan of 9.31 seconds - 10.6% faster than the second-best approach. The algorithm's dependency-aware scheduling architecture establishes new performance standards for time-sensitive simulation computing while maintaining applicability across diverse distributed environments including cloud computing and smart manufacturing systems.