【作者】 陈凌；

【导师】 梁加红；

【作者基本信息】 国防科学技术大学 ， 控制科学与工程， 2012， 博士

【摘要】 无人飞行器Ad Hoc网络的部署不需要预先建立基础设施，不囿于以控制基站为中心的星形网络结构，多跳转发的通信模式突破了飞行器只能工作于基站通信范围内的局限，从而扩宽了网络覆盖范围，具备改变战场通信模式的潜力。针对当前无人飞行器Ad Hoc网络研究应用中面临的任务繁杂、可控性和适应性差等难题，本文采用将通信与控制领域相结合的系统分析与设计研究方法，分别从基于容错的拓扑控制、基于延迟容忍网络的路由协议和基于仿真的分析优化等方面对无人飞行器Ad Hoc网络相关技术进行了深入研究，提出了相应的策略和方法，主要研究内容概括如下：一、论文首先对无人飞行器Ad Hoc网络相关技术及应用的国内外研究现状进行了概述，分析比较了不同方法的优缺点，提出了将通信网络性能与飞行器可控移动性两个不同领域的概念技术相结合的系统分析与设计研究思路，以期在运动控制效率和网络性能之间取得最佳折衷。二、顶点2连通是网络容错的基本属性和要求，当网络中无人飞行器节点（以下简称节点）完全可控时，研究了利用节点移动实现容错的问题。在链状构形网络中，提出了一种简单有效的分布式节点移动控制算法，节点仅需利用邻居节点的局部信息进行决策，通信开销小，资源消耗少。在平面构形网络中，将网络容错问题转化为最短移动距离的顶点2连通图强化问题，提出了一种两阶段的集中式节点移动控制算法。第一阶段中，节点不移动，通过在网络通信拓扑图中直接添加强化边来获取顶点2连通。寻找最小成本的强化边组合是NP完全问题，基于文化基因算法框架，设计了符合问题特点的全局进化策略、局部优化方法和考虑节点移动约束的强化边成本设定，计算得到了近似最优的最小成本强化边组合。第二阶段中，通过节点移动来实现这些强化边，提出了基于一致性原理的控制算法，在满足网络连通的约束条件下，移动强化边连通的节点及其关联节点，改变拓扑结构，实现网络容错。三、当网络中部分节点可控时，研究了引入辅助中继节点实现容错的问题。在平面构形网络中，将寻求最少数量中继节点及其位置的约束优化问题转化为瓶颈Steiner最小树问题，参考“Steinerized”范式，以连通分割节点所需中继节点的数量为强化边成本，提出了基于网络通信拓扑完全图中最小成本2连通子图的算法框架。研究了求解最小成本2连通子图的多项式时间近似算法和随机搜索算法，当采用近似算法时，整个算法有固定近似比，而随机搜索算法可进一步提高整个算法的平均性能。当中继节点的位置不局限于强化边上时，在算法得到的优化解中搜索可合并的中继节点组合，并用少量节点替代，可进一步减少所需中继节点数量，提高算法性能。四、当网络处于间隙性连通或分割状态，且节点完全不可控时，研究了基于延迟容忍网络的“储存—携带—转发”路由模式进行信息传输的问题。结合基于冗余机制的并行性和基于效用机制的选择性两者的优势，利用少量的消息副本和网络信息，提出了一种冗余与效用混合机制的路由协议。将最优停止理论应用于基于节点效用的消息副本分发决策过程，在消息副本的分发速度和消息副本携带节点的效用平均值之间取得折衷；将消息副本总数控制策略与消息副本分发策略相结合，可减少网络资源消耗，提高算法的可扩展性，实现了传输延迟、传输成功率和资源消耗等网络性能之间的最佳平衡。五、针对目前在网络协议算法的分析测试方面存在的难点，研究了数学分析和实验测试方法的优势与不足，采用了基于数字仿真进行网络性能分析测试的方法，结合现有软件工具，设计实现了基于分布交互仿真的无人飞行器Ad Hoc网络仿真系统，为拓扑控制算法和路由协议的分析测试提供软件实验环境。仿真系统采用高层体系结构，实现软件资源、仿真组件的有效重用和互联互操作。针对基于仿真优化的复杂系统分析设计过程中仿真模型精度与优化过程运行效率之间的矛盾，研究了仿真元模型在仿真优化中的应用，最后以网络性能优化为例，对本文的模型算法和仿真系统进行了实例验证。更多还原

【Abstract】 The deployment of the UAVs Ad Hoc networks does not need any establishedsupporting infrastructure, the network architecture isn’t limited to the star-structure withcontrol base station as the center. Moreover, the multi-hop message delivery mode breaksthe constraint on which the UAV can only work within the communication range of centralbase station. Consequently, the UAVs Ad Hoc networks, whose coverage is enlarged, hasthe potential to revolutionize the battle field communication. Currently the study of UAVsAd Hoc networks is encumbered by the difficulties such as tedious and poor controllability,bad adaptability etc. Based on the joint analysis and design of communication and controlsystems, this dissertation studies some technologies of the UAVs Ad Hoc networksincluding the fault-tolerant topology control, the routing protocol of delay-tolerant network,and the simulation-based analysis and optimization. The main contents of this dissertationcan be summarized as follows:(1) Firstly, the state of the art of the UAVs Ad Hoc networks is surveyed and theadvantages and disadvantages of different methods are compared. In order to get the besttrade-off between the effectiveness of the movement control and the network performance,the joint analysis and design strategy is proposed which combines the communicationnetwork performance with the controlling mobility.(2) Vertex-biconnectivity is basic feature of fault-tolerant networks. When the nodesin the network are controllable, the mobility-based control method is investigated to realizefault-tolerant networks. For the chain-formation network, a simple and effective movementalgorithm which requires only the local information of the neighbor nodes is proposed toreduce the spending of communication and the consume of resource. For the flat-formationnetwork, the fault-tolerant problem is converted to the mobility-based min-total-distancevertex-biconnectivity augmentation problem for graph, a two-phase centralized movementcontrol algorithm is proposed. In the first phase, the nodes are immovable, thevertex-biconnectivity of a given communication topology graph is achieved by addingedges which join the different vertexes. The search of the min-cost set of edges is aNP-hard problem. In the memetic algorithm frame, the problem specific globalevolutionary strategy, the local improvement procedure and the cost define ofaugmentation edge considering node movement constraint are developed to find thecheapest possible set of additional edges which makes the graph vertex-biconnected. In thesecond phase, with the augmentation edges as heuristic, the vertexes which the edgesconnected move towards each other to establish edges. The related nodes also move as thefollowers with the leader-follower movement control algorithm based on the consensusalgorithm. After nodes moved, the changed network becomes vertex-biconnected so thefault tolerant configuration is achieved. (3) When only part of the nodes in the network is controllable, the relay nodeplacement method is investigated to realize fault-tolerant networks. For the flat-formationnetwork, the minimum relay nodes placement problem is converted to the minimumbottleneck Steiner tree problem. Following the Steinerized research pattern, an algorithmframe based on the min-cost spanning subgraph of a complete graph is proposed. Tworelay node placement algorithms are investigated, the first one is a polynomial timeapproximation algorithm, and the second one is a random algorithm which can improve theresult on average. When the node placement locations are not limited to the straight linebetween the vertexes, an improvement method can be used to search the relay nodes setswhich can be replaced with much less nodes in the final resolution, so the number of relaynode is reduced and the performance of algorithm is improved.(4) When the network experiences frequent and long duration partition, and the nodesare uncontrollable, the information transmission based on the delay tolerant networks’“store-carry-forward” routing pattern is investigated. Combining with the parallel action ofthe redundancy-based routing and the selectivity of the utility-based routing, a hybridrouting is proposed in which the requirement of network knowledge and the duplication ofmessage are small. The optimal stop theory is used to decide whether to copy the messageto the neighbors in contacted, so the balance between the delivery speed of the messageduplication and the average utility of the message duplication carrier is achieved. Thehybrid routing combined the message duplication delivery strategy and the duplicationnumber control strategy, so the usage of network is small, and the extendibility is improved,and the best tradeoff among several performances of network is achieved, such as thedelivery delay, the delivery success rate, and the usage of network.(5) To deal with the difficulties in analysis, validation and test of the algorithms andprotocols, the mathematical analysis and the experimental test are investigated, the digitalsimulation is used to analysis and test the performance of network. The UAVs Ad Hocnetworks simulation system is developed based on distributed interactive simulation. Thesimulation system is based on the High Level Architecture, so the reusability of thesoftware resources and the interoperability of the simulation module are realized. Toaddress the dilemma between the accuracy of simulation-model and runtime efficiencywhich take place in the simulation-based optimization of the analysis and designprocedures of complex system, the metamodel-based optimization is investigated. Aninstance in optimization of the network performance is implemented to evaluate thesimulation system and the algorithm in the dissertation.更多还原