Load sharing for mobile agents using genetic algorithms

With the growth of service types and users in the mobile networks, the dynamic load sharing for mobile agents is required to improve service provision and throughput. To achieve the load sharing, effective resource allocation for mobile agents is necessary. The resource allocation for mobile agents is usually a combinatorial optimization problem. Because it is NP-hard, many different heuristics schemes were proposed to yield near optimal solutions. This study presents a new dynamic load sharing system that evaluates the agent host workload with a genetic algorithm (GA). The load sharing system contains two agents: load index agents (LIA), and load transfer agents (LTA). The performance results show that the proposed system outperforms other existing techniques for task throughput, transfer overhead, and communications overhead under most conditions.