VOD系统在最优平均性能下的过载问题研究

由于视频点播(VOD)系统打破了以往传统影音服务的局限,可以给用户提供更多的主动性和交互性,因此近年来得到了广泛关注和应用。但在视频点播(VOD)系统中,当媒体服务器处于过载状态时,服务质量和性能都会下降。本文采用Markov决策过程(MDP)中的bias优化方法研究了VOD系统在最优平均性能下的过载问题,使用bias优化算法可在平均性能最优策略集中找到使服务器处于过载状态的稳态概率最小的策略。该策略能够使服务器在长期运行平均服务请求数达到最大的同时,处于过载状态的稳态概率最小。最后,通过一个仿真实例,说明了该算法的有效性。

Overload of the VOD Systems Under Optimal Average Performance

Nowadays the VOD systems attract much attention and get widely applied, as they break the limitations of traditional video service and therefore can provide the users with more initiative and interaction. However, the VOD systems would experience a decrease in the service quality and performance when in the state of overload. The overload of the VOD systems under the optimal average performance is studied by applying the bias optimality in Markov decision processes （MDP）. The bias optimality can find a policy in the set of all the gain optimal policies that minimizes the steady - state probability of the state of overload. The policy gained by the proposed method can maximize the long - run average number of user requests served by the media server, and at the same time minimize the steady - state probability of the state of overload. At last, a simulation example is provided to demonstrate the effectiveness of the proposed method.