准入控制
马尔可夫决策过程
预订
限制
计算机科学
马尔可夫过程
计算机网络
服务(商务)
电信
工程类
服务质量
业务
数学
统计
机械工程
营销
作者
Pak Kay Tang,Yong Huat Chew,Wai-Leong Yeow,Ling Chuen Ong
出处
期刊:IEEE Transactions on Vehicular Technology
[Institute of Electrical and Electronics Engineers]
日期:2009-02-13
卷期号:58 (7): 3674-3683
被引量:21
标识
DOI:10.1109/tvt.2009.2014873
摘要
When spectrum is shared among multiple radio systems, spectrum admission control (SAC) can be performed via a centralized spectrum manager to meet their respective traffic demands. The adopted policy will determine the admission of service requests and, in turn, affect the overall spectrum utilization efficiency. The design of SAC policies becomes more challenging when each radio system provides a grade of service (GoS) in the form of a blocked service guarantee to its users. Given such constraints, we study the admission region, which indicates the maximum amount of supported traffic for the given resources. We first study the performance based on a simple first-come-first-serve (FCFS) policy. Our studies show that the admission region is, in general, limited by the radio system that first violates its prescribed GoS guarantee (i.e., the performance-limiting radio). We propose three SAC policies to enhance the total offered traffic and compare them against the FCFS policy. The first is a random discard (RD) policy, where requests from other systems are discarded with some predetermined probabilities so that a larger portion of the spectrum is made available to the performance-limiting radio. The second SAC uses a reservation (RES) policy, where a suitable amount of spectrum is reserved for the performance-limiting radio. In the third policy, SAC is formulated as a discrete-time constrained Markov decision process (MDP). We analyzed the performance of each policy at low and high request rates and show how the additional design parameters in each policy can be optimized to enhance the offered traffic. As admission decisions are jointly made with traffic predictions, the MDP admission policy shows superior performance over the other policies due to its flexibility to operate both radios at the bound of the respective GoS constraint.
科研通智能强力驱动
Strongly Powered by AbleSci AI