【作者】 丁胜；

【导师】 蒋建国；

【作者基本信息】 合肥工业大学 ， 信号与信息处理， 2018， 博士

【摘要】 无线数据汇聚网络(WSNs,Ad hoc,Mesh,et al)作为一类新兴的、极具特色和发展前景的无线网络技术,正逐步在国民经济和社会生活的各个领域得到了广泛的应用,同时也成为学术界研究的热点领域。在无线数据汇聚网络中,如何进行多信道和TDMA时隙等网络资源的调度,从而最大化网络监测质量,最小化网络冲突,最优化网络数据汇聚性能是一个重要的研究课题。该文全面总结了该领域的理论和技术现状,重点研究了无线节点信道和时隙调度问题的数学模型和算法,并通过理论分析和不同网络条件下的大量实验证明了所提出算法的综合有效性。该文的主要研究工作及创新之处在于:(1)针对无线数据汇聚网络中监测节点多信道调度问题,提出了一种基于同步微扰随机近似(SPSA)的信道选择算法。设计了二维网格编码以构造解空间,算法以随机扰动策略得到目标函数的近似梯度,以引导搜索过程逐步逼近最优解。该算法的运行只需要已知监测节点及其邻居节点,以及邻居用户的工作信道信息(可通过全扫频的方式获得)。该算法适合于复杂的多维优化问题求解,复杂度低,收敛速度快。大量实验结果表明该算法可以实现无线数据汇聚网络中监测节点的信道优化选择,并可达到较高的网络监测质量(QoM)。(2)针对无线数据汇聚网络中无线节点多信道TDMA时隙调度问题,提出了一种基于静态非合作博弈论的分布式顶点着色方法。首先基于网络拓扑构建路由树,形成干扰图,从而将上述资源调度问题转化为干扰图中的顶点着色问题,其目标是最小化网络通信冲突数;然后以每个顶点的通信冲突作为博弈的收益函数以构建纳什均衡状态,通过概率方式选择顶点以调整其分配方案并通过全局通信冲突数判断方案是否达到帕累托有效,经过多次调整后即可完成对信道和时隙的优化选择。在不同网络规模情况下,本文通过多种对比实验结果表明所提算法能够有效降低网络冲突数,提高数据吞吐量,降低数据传输时延和最小调度长度,最终达到提升网络数据汇聚性能的目的。(3)设计了数据汇聚网络资源调度算法测试床"DAN-RSM testbed",包括软硬件系统的设计和实现,可以提供中等网络规模的资源调度算法的应用测试。基于该实验平台,对该文所提出的算法和策略进行了实际有效性测试,得到了一系列实验数据,进一步验证了所提出方法的综合有效性以及不足之处,并为下一步研究和相关科研工作奠定了基础。该文研究无线数据汇聚网络中资源调度方法,提出了相应的模型、算法和实验测试,构建了相应的理论方法体系,相关成果对无线网络技术的发展具有一定的推动作用和参考价值。更多还原

【Abstract】 As a new type of wireless network technology with unique characteristics and development prospects,wireless data aggregation networks(WSNs,Ad hoc,Mesh,et al)are gradually applied in various fields of national economic and social life,and it has also become a hot area for academic research.In a wireless data aggregation network,how to schedule network resources such as channels and TDMA time slots to maximize network monitoring quality,minimize network collisions,and optimize network data aggregation performance is an important research topic.The paper comprehensively summarizes the theoretical and technological status in this field,and focuses on the mathematical models and algorithms for wireless node channel and time slot scheduling problems.Theoretical analysis and extensive experiments under different network conditions demonstrate the comprehensive efficiency of the algorithm.Main research work and innovations of the paper lie in:(1)Aiming at the problem of multi-channel scheduling for monitoring nodes in wireless data aggregation networks,a channel selection algorithm based on Simultaneous Perturbation Stochastic Approximation(SPSA)is proposed.Two-dimensional grid coding is designed to construct the solution space.The algorithm uses a stochastic perturbation strategy to obtain an approximate gradient of the objective function,which can guide the search process to gradually approximate the optimal solution.The algorithm requires only known monitoring nodes and their neighbor nodes,as well as neighbor users’ working channel information(which can be obtained by full frequency sweep).The algorithm is suitable for solving complex multidimensional optimization problems with low complexity and fast convergence.A large number of experiments demonstrate that the algorithm can realize the optimal channel selection of monitoring nodes in the wireless data aggregation network with high Quality of Network(QoM).(2)Proposed a distributed vertex coloring algorithm based on static non-cooperative game theory,on the problem of multi-channel TDMA time-slot scheduling problem in the process of data collection in wireless monitoring networks.First of all,based on wireless monitoring network topology,build a routing tree to form an interference graph,so as to convert the resource scheduling problem into vertex of interference graph double-coloring problem.The goal is to minimize network traffic conflict.Then,the communication conflict of each vertex is used as the revenue function of the game to construct the Nash Equilibrium state,and the vertices are selected in a probabilistic manner to adjust the allocation scheme.The global communication collision number is used to determine whether the scheme has reached Pareto Efficiency,after multiple adjustments.The optimized selection of channels and time slots can be completed after several adjustments.A series of comparative experimental results under different network conditions demonstrate that the algorithm can effectively reduce network conflicts,improve network throughput,reduce network transmission delay and scheduling length,achieving a higher data gathering performance of the wireless monitoring networks.(3)The data aggregation network resource scheduling algorithm test bed"DAN-RSM testbed" is designed,including the design and implementation of software and hardware systems,which can provide application testing of medium-scale network resource scheduling algorithm.Based on the experimental platform,the practical validity tests of the algorithm and strategy proposed in the paper are carried out and a series of experimental data are obtained,which further validates the comprehensive validity and deficiencies of the algorithm proposed in the paper and lay the foundation for the future research and related scientific work.This paper studies resource scheduling methods in wireless data aggregation networks,proposes corresponding models,algorithms and experimental tests,and constructs a corresponding theoretical method system.Related achievements have a certain role and reference value in promoting the development of wireless network technology.更多还原