论文部分内容阅读
在认知无线电网络中,传输层端到端(TCP)吞吐率是衡量网络性能的重要指标.前期相关研究大都具有以下两方面缺点:第一,大部分研究只考虑了协议底层参数来优化物理链路性能,对传输层性能有所忽略;第二,目前的研究大都基于马尔可夫决策过程建模,这需要网络具有完全知识,使得这类模型的应用受到很大限制.针对以上问题,本文提出一种新的算法:网络中每个节点通过联合配置物理层调制方式、发射功率、链路层信道接入和TCP拥塞控制因子来找到传输层端到端近似最优吞吐率.由于无线设备对环境感知存在误差,本文将网络模型建模为部分可观测马尔可夫决策过程,并将其转换成信念状态马尔可夫决策过程,采用Q值迭代找到近似最优策略.仿真分析表明,提出的算法能在动态无线环境下以一定的误差限收敛于最优策略,能在功率受限条件下,有效提高传输层端到端吞吐率.
In the cognitive radio network, the throughput of the transport layer (TCP) is an important measure of network performance.Previous studies mostly have the following two shortcomings: First, most of the studies consider only the underlying parameters of the protocol to optimize the physical Link performance, which neglects the performance of the transport layer.Secondly, the current researches are mainly based on Markov decision process modeling, which requires that the network has complete knowledge to make the application of such models greatly restricted.Aiming at the above problems, In this paper, a new algorithm is proposed: each node in the network can find the optimal end-to-end throughput of the transport layer by jointly configuring the physical layer modulation mode, transmit power, link layer channel access and TCP congestion control factors, This paper presents a network model as partially observable Markov decision making process and transforms it into belief state Markov decision making process using Q-value iteration to find the approximate optimal strategy.The simulation results show that, The proposed algorithm can converge to the optimal strategy with certain error limits under dynamic wireless environment and can effectively improve the transport layer end-to-end Throughput.