基于马尔可夫决策过程的网络吞吐量最优化算法

Network Throughput Optimization Algorithm Based on Markov Decision Process

以马尔可夫决策过程作为理论工具,将终端直通选择问题与有限阶段折扣MDP模型问题相结合,研究了网络吞吐量最优化问题。首先,利用MDP对终端直通选择进行建模;其次,利用有限阶段后向迭代算法给出最优模式选择策略;最后,通过进行大量的仿真实验,对给出的模式选择策略进行评估。结果表明,本研究提出的基于MDP的模式选择方法在最大化吞吐量方面拥有更好的性能,能得出更优的模式选择策略,具有获得更多系统吞吐量的优势。

The Markov decision process was used as the theoretical tool to study the network throughput optimization algorithm combined with the D2D mode selection with the finite phase discount MDP model.MDP was used to establish a model of D2D mode selection,the iterative algorithm was used to find the best policy of mode selection,and the strategy given by extensive simulation experiments was evaluated.The results showed that the MDP-based model selection method could make a better choice with better performance in maximizing system throughput.