Hybrid Discrete Particle Swarm Optimization for Energy-Aware Scheduling on Test and Control Systems

In this paper, a hybrid discrete particle swarm optimization algorithm (HDPSO-VNS) is proposed for energy-constrained performance optimization in test and control systems (TCSs). Task scheduling among multiple processors needs to balance energy consumption and scheduling quality, which exhibits the characteristics of NP-hard. Considering the periodicity and dependency of test tasks, a directed acyclic graph is utilized to depict the priority relationship between real-time tasks. The proposed hybrid metaheuristic algorithm initiates with particle swarm optimization to obtain an approximate optimal processor allocation strategy. A variable neighborhood search strategy is applied to expand search space through three different neighborhood actions, refining local optimization ability of HDPSOVNS based on the current optimal solution. The algorithm is evaluated in a heterogeneous processor environment of CloudSim, and comparative experimental results confirm the feasibility and effectiveness.