一种在DS-TE网络环境中新的LSP抢占算法

现有的启发式抢占算法不易找到最优解,容易造成带宽的浪费。文中通过分析DS-TE（Differentiated Services Traffic Engineering）网络环境中的抢占策略,提出了一种优化的抢占算法GA-PREPT（Genetic Algorithm Preemption）。GA-PREPT利用遗传算法进行LSP（Label Switch Path）的优化选择,同时针对抢占问题的约束条件在进行选择操作时引入模拟退火选择算子,指导进化过程的选择操作。仿真结果表明GA-PREPT性能相比现有的启发式抢占算法有较大改善。

A New LSP Preemption Algorithm for DS-TE Network

The heuristic preemption currently used to select LSP could not easily find the best solutions and wastes the bandwidth. This paper, by analyzing the preemption policy in DS-TE networks, proposes an optimized preemption policy named GA-PREPT. This new preemption selects the preempted LSP via an improved genetic algorithm. The annealing selecting operator is also introduced to handle the constrained optimization problem and guide the evolutionary process. Simulation results show that the proposed algorithm significantly outperforms the recently-proposed heuristic.