论文部分内容阅读
由于多媒体业务需要更大的带宽和更高的实时性,所以对服务系统和接入控制提出了更高的要求。文章针对分布式媒体服务系统提出了一种新颖的接入控制方法,与以往方法不同之处在于将请求调度融合在接入控制之中,从而提高了系统的服务性能。我们先为系统建立部分可观Markov决策过程(POMDP)模型,并将请求调度融合在决策中,然后使用基于观测的随机接入控制策略,通过策略梯度优化算法仿真求解模型的最优策略。仿真结果表明,与其他分布式接入控制方法相比,该文所提方法在有效利用系统资源的同时,提高了系统性能。
As multimedia services require more bandwidth and higher real-time nature, so the service system and access control put forward higher requirements. In this paper, a novel access control method is proposed for distributed media service system. The difference between this method and the previous method lies in that the request scheduling is integrated into the access control, which improves the service performance of the system. We first establish some observable Markov decision process (POMDP) models for the system, and integrate the request scheduling into the decision-making. Then we use the observation-based random access control strategy to solve the optimal strategy of the model through the strategy gradient optimization algorithm. The simulation results show that compared with other distributed access control methods, the proposed method improves system performance while effectively utilizing system resources.