A Generalized Particle Model and Algorithm for Dynamic Optimization of Bandwidth Allocation in ATM Networks
Shuai Dian · Chinese Journal of Computers · 2007
A novel generalized particle model(GPM) and its algorithm for dynamically optimizing both the VP bandwidth allocation and the VP negotiated QoS parameters in ATM networks are presented.The proposed approach transforms the dynamic bandwidth allocation problem among VP′s in ATM networks into the kinematics and dynamics problem of particles in two reciprocal dual-force fields,so that the evolution of particles states can eventually results in an optimal solution of the original bandwidth allocation problem.The basic properties of GPM,including the suitability,convergency and stability,are discussed.The bandwidth allocation problem in ATM networks is NP-complete.Based on the service categories,traffic characteristics and QoS requirements,dynamically allocating the virtual path bandwidth in ATM networks plays a significant role in enhancing the ATM network throughput and improving the QoS performance.The theoretical analysis and numerous simulations on ATM network bandwidth allocation have shown that the GPM approach has the higher parallelism,lower computation complexities,easy of hardware implementation and better availability for complex environment.