【作者】 陈征；

【导师】 周文利；

【作者基本信息】 华中科技大学 ， 微电子学与固体电子学， 2021， 博士

【摘要】 随着5G移动网络的日益成熟,物联网(Internet of Things,IoT)也得到了快速发展。移动无线传感器网络(Mobile Wireless Sensor Network,MWSN)因其节点具有移动感知能力的优势成为IoT的一种重要感知网络,但是自组织的MWSN存在的感知数据类型的多样性、节点的移动性、网络资源的有限性等特点,也使其自身以及所处IoT的QoS(Quality of Service)保障能力面临着挑战。特别是,集成了多种类型传感器的节点所采集的数据对QoS还存在差异性的需求。因此,针对因MWSN节点的感知任务多样性、资源受限和移动性而容易出现的数据业务QoS难以保障的问题,改进其网络运行状态、提升网络效率一直是MWSN的QoS保障研究的重要方向。这不仅要解决MWSN中存在的QoS问题,还要考虑其上下行链路数据业务的QoS需求及其负载的差异性。目前,以QoS需求为条件的相关路由策略中,尚未有同时针对MWSN数据业务差异QoS保障和负载均衡的解决方案。本文针对IoT中MWSN的上下行数据业务的特点,改进了传统无线传感器的节点模型,以提出的动态移动网关(Dynamic Mobile Gate Way,DMGW)为连接架构,设计了一种能同时满足差异服务需求和负载均衡的路由策略。对上行链路的数据业务采用了差异服务的路由策略,对下行链路的数据业务采用了自适应网络拓扑变化的路由策略。利用NS2平台,在节点速度为10～50m/s的移动场景中对相应协议进行了仿真研究。本文的主要研究工作和成果如下:(1)针对移动感知节点接入IoT网关(IoT Gate Way,IGW)存在频繁切换的问题,提出了动态移动网关(DMGW)的接入方案,选择与IGW连接最稳定的MWSN节点作为DMGW,并辅以链路维护和故障处理措施提高DMGW与IGW的连接稳定性,并采用AOMDV(Ad-hoc On-demand Multipath Distance Vector)路由协议实现。仿真结果显示,在节点密度高、移动速度大的场景中,DMGW比IGW连接稳定性高,当节点数为50、速度为50m/s时,DMGW的分组投递率(Packet Delivery Ratio,PDR)比IGW高14.95%,切换延迟时间低8.01%。(2)针对IoT下行数据业务的完整性需求,提出了一种自适应网络拓扑变化的路由策略(Topological change Adaptive Ad hoc On-demand Multipath Distance Vector,TA-AOMDV),通过对链路质量的评估选出最优路由路径,链路监测机制根据丢包率和时延调整路由策略,并通过链路故障自修复机制减小路由开销。仿真结果显示,随着节点数增多,丢包率减小,归一化端对端时延(End to End Delay,E2ED)和路由开销增大;随着节点速度提高,这三项指标都升高。在节点速度45～50m/s的高速场景中,TA-AOMDV性能均优于QMR(QoS aware Mulitpath Routing protocol)、LRMR(Link Reliable Multipath Routing)、QoS-AOMDV;其中当节点数为30、速度为50m/s时,PDR分别提高了3.2%、5.34%、4.6%,端到端延时(End to End Delay,E2ED)分别降低了19.1%、24.62%、21.3%,吞吐量分别提高了19.6%、32.7%、26.3%,路由开销增大了43.1%、34.7%、29.2%。(3)针对IoT上行的多样性数据业务存在的差异QoS需求,提出了一种预留路径的多径路由策略(Path Reservation Multipath Routing,PRMR),满足不同数据业务的QoS需求,并通过分组调度实现预留路径间的负载均衡。仿真结果显示,随着节点速度的提高,完整性需求数据的丢包率、时延敏感数据的归一化E2ED和路由路径节点能量平均偏差均呈现升高趋势。在节点数为60～120的场景中,PRMR性能优于MQRTS(Multipath QoS Routing protocol for Traffic Splitting)和Improved-AOMDV协议,节点密度越高优势越明显;当速度为10m/s、节点数为120时,对完整性敏感的数据,PRMR的PDR比MQRTS和Improved-AOMDV协议分别高6.42%和32.4%,对时延敏感数据,PRMR的E2ED与MQRTS的相当,分别为5.37ms和7.55ms,优于Improved-AOMDV的56.51ms。PRMR的负载均衡优于MQRTS但略逊于Improved-AOMDV,路由路径节点能量平均偏差比MQRTS低18.35%,比Improved-AOMDV高3.68%。(4)综合本文提出的PRMR上行路由和TA-AOMDV下行路由策略,本文实现了一种兼顾上行感知数据差异QoS和下行指令数据完整性的MWSN多径路由协议(Uplink&Downlink heterogeneous AOMDV,UD-AOMDV)。仿真结果显示,相比仅考虑上行或下行服务的PRMR和TA-AOMDV,UD-AOMDV所选路径提供的差异服务和数据完整性的质量下降<8%,路由开销最多增大12%。在高速(30～50m/s)场景中,UD-AOMDV比MQRTS+QMR组合协议的QoS有明显的提升,当节点数为100、速度为50m/s时,以路由开销增加16.4%的代价降低了9.1%的丢包率和6.74%的归一化E2ED。UD-AOMDV协议保障了上行数据传输的差异服务质量且兼顾了负载均衡,提高了下行数据的完整性。本文研究的路由策略可为异构移动网络融合的QoS性能优化和无线传输的冲突处理提供借鉴及参考,本文的研究结果也可为进一步对相关策略进行QoS可知的协议优化提供依据。更多还原

【Abstract】 With the increasing maturity of 5G communication networks,the Internet of Things(IoT)has achieved its rapid developments.Mobile Wireless Sensor Network(MWSN)has become an important sensing Network in IoT because of the advantages of mobile sensing capability of its nodes.However,the diversity of sensor data types,node mobility,and the limitation of network resources in self-organized MWSN also face great challenges in their own and IoT QoS guarantee capabilities.In particular,the data collected by nodes with multiple types of sensors have different requirements on QoS.Therefore,in view of the difficulty in guaranteeing QoS of data service due to the diversity of sensing tasks,limited resources and mobility of MWSN nodes,improving its network running state and improving network efficiency has always been the focus of its research.This not only needs to solve the QoS problem existing in MWSN,but also needs to consider the difference of QoS demand and load of data service of upstream and downstream links between IoT users and sensing nodes.At present,in the routing strategy with QoS demand as the constraint condition,there is no solution that considers both the data service difference QoS guarantee and load balancing simultaneously.Considering the characteristics of uplink and downlink data service of MWSN in IoT and based on the proposed dynamic mobile gateway(DMGW),with the modified mobile wireless sensor node model,this dissertation designs a routing strategy that can meet the different service requirements and load balancing simultaneously.The routing strategy of differential service and adaptive network topology change are adopted for the data service of uplink and downlink,respectively.Using NS2 platform,the corresponding protocols are simulated and researched in mobile scenario with 10～100 nodes at speed in the range of10～50 m/s.The main work and research results of this dissertation are as below.(1)To solve the problem of frequent switching between MWSN nodes and the IoT Gate Way(IGW),a dynamic mobile gateway(DMGW)access scheme is proposed.The MWSN node with the most stable connection with IGW is used as DMGW,and the connection stability between DMGW and IGW is improved by link maintenance and fault handling measures.AOMDV(Ad-Hoc On-Demand Multipath Distance Vector)protocol is employed to implement the routing of DMGW.The simulation results show that the packet delivery ratio(PDR)of the DMGW scheme is 14.9% higher than that of the IGW scheme and the average switching time is 8.01% lower for the scenario of 50 nodes at 50m/s.(2)Aiming at the integrity requirements of IoT downlink data service,a routing strategy based on adaptive network topology change named as Topological change Adaptive Ad hoc On-demand Multipath Distance Vector(TA-AOMDV)is proposed.The optimal routing path is selected through the evaluation of link quality,the link monitoring mechanism adjusts routing policies based on packet loss rate and delay,and the routing overhead is reduced through the link fault self-repair mechanism management.The simulation results show that the packet loss rate decreases with the increase of the node number,but the normalized End to End Delay(E2ED)and routing overhead increase at the same time;These three indexes all increase with the increase of node speed.The performance of TA-AOMDV is better than that of QoS aware Mulitpath Routing protocol(QMR),Link Reliable Multipath Routing(LRMR)and QoS-AOMDV in high-speed scenarios with node speeds at 45～ 50m/s.In comparison,in the case of 30 nodes at 50m/s,the PDR of TA-AOMDV is increased by3.2%,5.34% and 4.6% respectively,and End to End Delay(E2ED)is decreased by 19.1%,24.62% and 21.3% respectively.The throughput is increased by 19.6%,32.7%,26.3%,and routing overhead is increased by 43.1%,34.7%,29.2%,respectively.(3)Targeting at the different QoS requirements of IoT upstream data services,a multi-path routingstrategy with reserved paths named as Path Reservation Multipath Routing(PRMR)is proposed,while the load among reserved paths is balanced through packet scheduling.The simulation results show that with the increase of node speed,the packet loss rate of the integrity requirement data,the normalized E2 ED of time-sensitive data and the average deviation of the energy of the routing path nodes all present their increasing trends.In scenarios with 60～120 nodes,PRMR outperforms both MQRTS(Multipath QoS Routing Protocol for Traffic)and Improved AOMDV,with the advantages more obvious in higher node density.In the case of 120 nodes at 10m/s,for integrity-sensitive data,the PDR of PRMR is 6.42%,which is 32.4% higher than that of MQRTS and Improved AOMDV,respectively;for time-sensitive data,the E2 ED of PRMR is similar to that of MQRTS,which is 5.37 ms and 7.55 ms,respectively;they are both much better than that of Imporved-AOMDV,i.e.,56.51 ms.The average energy deviation of routing nodes of PRMR was 3.68% higher than that of Improved-AOMDV and 18.35%lower than that of MQRTS,i.e.,the load balancing of PRMR was better than MQRTS but a little bit worse than Improved-AOMDV.(4)Combine the proposed PRMR uplink routing strategy and TA-AOMDV downlink routing strategy on the DMGW architecture,an Uplink & Downlink heterogeneous AOMDV(UD-AOMDV)is also implemented in this dissertation for MWSN,which takes into accounts of the differential QoS of the uplink sensing data and the integrity of the downlink instruction data.To reconcile the transmission requirements of upstream and downstream data,the quality of differential service and data integrity provided by the paths chosen by UD-AOMDV are decreased by <8% and the routing overhead increased by 12%at most,compared with the paths chosen by PRMR and TA-AOMDV that only considered uplink or downlink services.At high-speed(30～50m/s),UD-AOMDV has obvious improvement compared to the combined protocol of MQRTS+QMR.For example,in a scenario of 100 node at 50m/s,the packet loss rate and normalized E2 ED of UD-AOMDV are decreased by 9.1% and 6.74%,respectively,at the cost of 16.4% increase in routing overhead.The routing strategy studied in this dissertation could provide references for QoS performance optimization of heterogeneous mobile network merging and conflict processing of wireless transmission.The results in this work could also render some evidences to implement the proposed strategies in the view of QoS awared promotion.更多还原