无线SDN稳定性拥塞控制算法研究

发布时间：2018-03-30 19:05

本文选题：无线软件定义网络　切入点：拥塞控制　出处：《电子科技大学》2017年博士论文

【摘要】：无线网络中大量数据传输会造成网络拥塞,导致网络性能下降,无线软件定义网络(Wireless Software Defined Network,WSDN)稳定性拥塞控制技术是解决网络拥塞的技术手段。该技术通过优化在传播时延和外部干扰影响下WSDN中网络参数,保持网络性能稳定。本文分别从分布式和集中式拥塞控制技术的分析着手,对传播时延下稳定性算法以及传播时延和外部干扰下鲁棒性算法进行了深入研究,主要开展了WSDN稳定性拥塞控制技术的模型建立、控制算法和性能分析研究,本文主要贡献概括如下:(1)针对大量数据传输导致WSDN中拥塞控制的不稳定问题,即如何在稳定性拥塞控制下使得在传播时延下WSDN网络参数稳定在最大化全局吞吐量优化值处,以及在鲁棒性拥塞控制下使得在传播时延和外部干扰下网络参数长时间保持其稳定状态,提出了一种分层分布式WSDN的网络架构。该网络架构由WSDN核心网(Core-WSDN)和分簇无线传感器网络子网(Subnetwork Cluster Wireless Sensor Network,Sub-CWSN)两部分混合分层组成。通过混合分层方式,分别研究了在传播时延和外部干扰下WSDN中分布式Sub-CWSN和集中式Core-WSDN拥塞控制的稳定性。提出了相应的稳定性拥塞控制算法,经理论分析,得到了传播时延下无线网络参数稳定在最大化全局吞吐量优化值处的稳定充分条件;并提出了相应的鲁棒性拥塞控制算法,经理论分析,得到了传播时延和外部干扰下无线网络参数能保持在优化状态下的稳定充分条件。解决了WSDN拥塞控制的全局稳定性控制问题。(2)针对WSDN稳定性拥塞控制的全局吞吐量性能优化问题,提出了相应的最大化全局吞吐量的调度算法。基于WSDN的网络参数,分析网络中处理数据流的节点资源分配。该算法是一种网络资源公平分配的优化算法,为拥塞控制的稳定性算法提供了理想化的控制目标,为WSDN稳定性和鲁棒性拥塞控制算法的实现提供了支撑。(3)考虑分布式Sub-CWSN全局稳定性拥塞控制,通过对其网络参数和所需探测调整技术的分析,本文先提出了一种基于加性增加乘性减少(Additive-Increase Multiple-Decrease,AIMD)策略、随机早期检测(Random Early Detection,RED)和邻居节点信道竞争的调整算法。在此算法基础上,给出了Sub-CWSN簇同步概念,即同一簇中所有节点发送速率增减一致,再提出了一种Sub-CWSN全局稳定性拥塞控制簇同步算法,采用控制论中稳定性分析法,通过对Lyapunov-Krasovskii泛函的分析,将无线网络拥塞控制策略和控制论中稳定性分析理论相结合,使得Sub-CWSN全局拥塞控制算法转化为拥塞控制全局簇同步算法,最后得到了稳定性拥塞控制的稳定充分条件。数值结果表明该算法满足稳定充分条件,能将网络参数稳定工作在最大化吞吐量的优化值处,实现了稳定性拥塞控制,比传统RED算法具有优势。(4)针对传播时延和外部干扰对Sub-CWSN和Core-WSDN拥塞控制系统影响导致网络开销增大和不稳定等问题,建立了在传播时延和外部干扰下拥塞控制模型,提出了实现该模型相应的鲁棒性拥塞控制算法。该算法在无线网络稳定的前提下,能克服传播时延和外部干扰的影响,实现鲁棒性拥塞控制,经理论分析,得到了鲁棒性拥塞控制的稳定充分条件。数值结果表明该算法满足稳定充分条件,能将网络参数长时间稳定工作在最大化吞吐量的优化值处,解决了在传播时延和外部干扰下无线网络拥塞控制的全局鲁棒性控制问题,比传统拥塞控制算法具有优势。(5)为了简化Core-WSDN拥塞控制算法过程,通过中央控制器基于全局拥塞状态信息,给出了填充等待时间概念,建立了在传播时延和外部干扰下拥塞控制模型,提出了一种集中式Core-WSDN离散鲁棒性拥塞控制算法,经理论分析,得到了鲁棒性拥塞控制算法的稳定充分条件。数值结果表明该算法满足稳定充分条件,能将网络参数长时间稳定工作在最大化吞吐量的优化值处,解决了在传播时延和外部干扰下集中式Core-WSDN离散拥塞控制的全局鲁棒性控制问题。与已有算法相比,简化了网络拥塞控制算法过程,使鲁棒性拥塞控制算法易于实现。
[Abstract]:The wireless network data transmission will cause network congestion, resulting in the decline of the network performance, wireless software defined network (Wireless Software Defined Network, WSDN) the stability of congestion control technology is to solve the technical means of network congestion. The technology by optimizing the WSDN network parameters in the propagation delay and disturbance, maintain a stable network performance analysis in this paper. Respectively from the distributed and centralized congestion control technology to the robustness of the algorithm of propagation delay stability algorithm and propagation delay and external disturbance is studied, and to carry out the stability of WSDN congestion control technology research model, control algorithm and performance analysis, the main contributions of this paper are summarized as follows: (1) for a large number of data the transmission leads to the instability problem of congestion control in WSDN, which is how to make the stability of congestion control in transmission delay under WS Stability of the DN network optimization value in maximizing the global throughput, keep long time and network parameters in the propagation delay and external disturbance under the stable state in robust congestion control, we propose a layered distributed WSDN network architecture. The network architecture by WSDN core network (Core-WSDN) and clustering algorithm in wireless sensor networks subnet (Subnetwork Cluster Wireless Sensor Network, Sub-CWSN) the two part composition. By mixing the mixed layered layered way, WSDN Sub-CWSN and Core-WSDN in distributed centralized congestion control stability in the propagation delay and external disturbance were studied. The stability of congestion control algorithms, through theoretical analysis, the stability of the wireless network transmission parameters under the optimized value of delay stability sufficient condition at maximum global throughput; and puts forward the corresponding robust congestion control algorithm, By theoretical analysis, the parameters of a wireless network propagation delay and external disturbance can maintain stability sufficient condition in the optimization condition. To solve the global stability problem of WSDN congestion control. (2) according to the global throughput performance optimization problem of the stability of WSDN congestion control, scheduling algorithms are proposed to maximize the overall throughput of the corresponding. The network parameters based on WSDN node resource allocation analysis of data stream processing in the network. The algorithm is a fair distribution of cyber source, provides an ideal target for the stability control of congestion control algorithm, WSDN algorithm for the stability and robustness of the congestion control provides support for distributed Sub-CWSN (3). The global stability of the congestion control, through the analysis of the network parameters and the required adjustment of detection technology, this paper first presents an additive increase multiplicative decrease based on Less (Additive-Increase Multiple-Decrease, AIMD) strategy, random early detection (Random Early Detection RED) algorithm to adjust competition and neighbor channel. On the basis of this algorithm, gives the concept of Sub-CWSN cluster synchronization, namely all nodes in the same cluster sending rate increase, and then proposes a congestion control algorithm of Sub-CWSN cluster synchronization stability overall, using the method of stability analysis in the control theory, through the analysis of the Lyapunov-Krasovskii function, the stability of wireless network congestion control strategy and control theory analysis combined with the theory of the Sub-CWSN global congestion control algorithm into congestion control global cluster synchronization algorithm, finally the stable sufficient conditions of the stability of congestion control is obtained. Numerical results show that the the algorithm meet the stability sufficient conditions can be optimized to maximize the throughput in the stability of the network parameters, to achieve a steady Qualitative congestion control has advantages over the traditional RED algorithm. (4) according to the propagation delay and external disturbance led to increase the cost of network problems and unstable Sub-CWSN and Core-WSDN congestion control system, established the model of congestion control in propagation delay and external disturbance, is proposed to realize the model robustness of the corresponding congestion control algorithm in the premise of the wireless network. The algorithm is stable, can overcome the effects of propagation delay and external disturbance, to achieve robust congestion control, through theoretical analysis, obtained the sufficient conditions for robust stability of congestion control. The numerical results show that the algorithm meets the stability sufficient conditions can be optimized to maximize the throughput in the network parameters at the time of stable value, solve the robust control problem in the global wireless network congestion control propagation delay and external disturbance, is better than the traditional congestion control algorithm Potential. (5) in order to simplify the Core-WSDN congestion control algorithm, through the central controller based on global congestion information, are waiting to be filled with the concept of time, established a model of congestion control in propagation delay and external disturbance, proposes a centralized Core-WSDN discrete robust congestion control algorithm, through theoretical analysis, we obtain sufficient conditions the stability of the congestion control algorithm is robust. The numerical results show that this algorithm can satisfy the stability sufficient conditions can be optimized to maximize the throughput in the network parameter value is stable for a long time, to solve the robust control problem of discrete centralized Core-WSDN congestion control in propagation delay and external disturbance. Compared with the existing algorithms, the algorithm is simplified the process of network congestion control, the robust congestion control algorithm is easy to implement.

【学位授予单位】：电子科技大学
【学位级别】：博士
【学位授予年份】：2017
【分类号】：TN915.0

【参考文献】