Swarm Intelligence Algorithms for Resource Optimization: A Comparative Experimental Analysis and Hybrid Approach

Abstract

Resource optimization is an essential factor in many areas, such as manufacturing, logistics, cloud computing, wireless sensor network and energy systems. Conventional optimization methods can be impractical when utilized in large, nonlinear, or dynamic optimization problems. Swarm Intelligence (SI) algorithms are based on collective behavior in nature, which provide alternatives, which are decentralized, flexible, and efficient. This paper provides a detailed experimental analysis of three traditional swarm algorithms Particle Swarm Optimization (PSO), Ant Colony Optimization (ACO), and Artificial Bee Colony (ABC) to an allocation of resources and scheduling. The performance of each algorithm was determined using a benchmark problem, which modeled the problem of minimizing costs, resource utilization, convergence behavior, the complexity of computation and robustness to dynamic problems. Furthermore, a hybrid PSO-ACO algorithm is suggested with the utilization of the fast convergence of PSO and the global search of ACO. The observed experimental results can confirm that the hybrid model is more effective in cost optimization and utilization by 12-18 percent than other algorithms. The paper ends by projecting the research directions in the future such as adaptive hybridization, multi-objective optimization, and real-world applications.