针对传统区分服务模型服务等级分类有限,结合策略管理的优势,提出了基于策略的细粒度区分服务概念,给出了不同类型区分服务的通用描述形式,提出了基于遗传算法的细粒度服务优化算法.当系统资源不足以满足多媒体流的传输要求或者用户需要细粒度的区分服务质量时,通过启用遗传算法对PHB进行组合优化,能够在较少的代数内求得满足用户QoS要求的最优解,实验表明该算法可行.
Present DiffServ can only provide rough granularity QoS guaranteed for clients since limited PHB classification. In this paper, a policy based fine granularity DiffServ mechanism as well as Universal Expression of Differentiated Service is proposed. The core idea of fine granularity differentiated service is that we can obtain different Per Domain Behavior (PDB) by composing multiple Per Hop Behaviors (PHB) in the path they passed. Then a GAs based Optimized PHB composition algorithm is designed. The algorithm is suitable for the cases such that system resources can not meet the requirements of multimedia stream transmission or the user need fine granularity QoS. The results show that the near global optimal solution can be easily and quickly obtained by this method.