无线传感网的智能安全路由策略的研究与实现

【作者】 张超；

【导师】 刘智斌；

【作者基本信息】 曲阜师范大学 ， 工程硕士（专业学位）， 2021， 硕士

【摘要】 现代无线传感器网络的发展速度越来越快,无线传感器网络中的传感器也越来越多。随着社会的发展,人们对于信息安全和传输速度要求越来越高,整个无线传感器网络中需要传输的数据呈指数性增长。面对以上现状研究人员提出多种传输协议。无线传感器网络的常用协议有Gossiping协议、SPIN协议、GPRS协议、TBF协议等。这些协议将节点接收到的数据转发到一个或若干个相邻节点时会遇到各种各样的问题。例如转发过程中的数据传输重叠问题、节点资源占用问题和内存爆炸问题,并且对于恶意节点的攻击行为,这些路由协议并没有好的方法来进行应对。找到一个方案来解决无线传感器网络中恶意节点的攻击问题和路径转发过程中出现的数据重叠、内存爆炸问题成为当前急需解决的。现阶段已经众多研究人员提出方案,比如采用强化学习的自适应性来实现路由的选择,解决传输和转发过程中的资源占用和浪费问题。整个传感器网络将节点路由信息整合在一起,通过一个已知的强化学习算法进行学习,获得到达某个传感器节点的最优策略。但是这个方案只考虑如何找到最优路由,但是对于环境中的攻击行为没有好的抵御手段。为了解决节点在传输过程中遇到的安全问题,本文提出建立一个信任模型,该信任模型采用能量值E和工作量W来计算信任值T.在这个信任模型中信任值与工作量之间是正相关,可以用线性相关表示,将和(2计算后求得。结合Q-learning提出WSN-Q算法,信任值与WSN-Q算法结合来抵御现阶段下无线传感器网络所面临的来自恶意节点的攻击。针对来自链路上的攻击,提出采用非对称加密技术,本文使用RSA加密技术,该加密技术十分可靠,能够满足无线传感器网络的使用需求。经论证,本文所提出的WSN-Q算法可以证明通过节点端对全局的数据进行学习能够实现无线传感器网络传输效率提升。并且设计了一定的可扩展性,后期能够对路由算法进行改进。为了验证该算法比传统协议传输效率更高,本文搭建一个仿真环境进行实验,对比在各种条件下传统RIP路由协议和WSN-Q算法之间的性能,面对恶意节点攻击时的平均时延。通过研究其实验结果发现相比传统的算法而言WSN-Q算法性能更为优秀,并且能够有效降低平均时延,但稳定性有待提高。实验证明该算法仍然有很高的提升空间,适合在传感器节点能量充足且整体相对封闭的区域使用。更多还原

【Abstract】 The development speed of modern wireless sensor networks is getting faster and faster,and there are more and more sensors in wireless sensor networks.With the development of society,people have higher and higher requirements for information security and transmission speed,and the data that needs to be transmitted in the entire wireless sensor network is increasing exponentially.Facing the above situation,researchers have proposed a variety of transmission protocols.Commonly used protocols for wireless sensor networks include Gossiping protocol,SPIN protocol,GPRS protocol,TBF protocol and so on.These protocols will encounter various problems when forwarding the data received by a node to one or several neighboring nodes.For example,the data transmission overlap problem,node resource occupation problem and memory explosion problem in the forwarding process,and the attack behavior of malicious nodes,these routing protocols do not have a good way to deal with it.How to find a solution to solve the attack problem of malicious nodes in the wireless sensor network and the problems of data overlap and memory explosion in the path forwarding process have become an urgent need to solve at present.At this stage,many solutions have been proposed,such as using the adaptiveness of reinforcement learning to implement route selection,and to solve the problem of resource occupation and waste in the transmission and forwarding process.The entire sensor network integrates node routing information,learns through a known reinforcement learning algorithm,and obtains the optimal strategy to reach a certain sensor node.But this scheme only considers how to find the optimal route,but it has no good defense against attacks in the environment.In order to solve the security problems encountered by nodes in the transmission process,this paper proposes to establish a trust model,which uses energy value E and workload W to calculate trust value T.In this trust model,the relationship between trust value and workload is positive.Correlation can be expressed by linear correlation,and T is calculated by calculating E and W.The WSN-Q algorithm is proposed in combination with Q-learning,and the trust value T is combined with the WSN-Q algorithm to resist attacks from malicious nodes that wireless sensor networks face at this stage.In response to attacks from the link,asymmetric encryption technology is proposed.This article uses RSA encryption technology,which is very reliable and can meet the needs of wireless sensor networks.After theoretical argumentation,the WSN-Q algorithm proposed in this paper can prove that learning the global data through the node end can achieve the improvement of wireless sensor network transmission efficiency.And designed with a certain degree of scalability,the routing algorithm can be improved later.In order to verify that the algorithm is more efficient than the traditional protocol,this article builds a simulation environment for experiments to compare the performance between the traditional RIP routing protocol and the WSN-Q algorithm under various conditions,and the average delay in the face of malicious node attacks..Through studying its experimental results,it is found that the performance of WSN-Q algorithm is better than traditional algorithms,and it can effectively reduce the average delay,but the stability needs to be improved.Experiments show that the algorithm still has a high room for improvement,and it is suitable for use in areas where the sensor node energy is sufficient and the overall relatively closed area.更多还原