基于群体智能的工业无线传感器网络安全防御建模研究

In order to overcome the common challenges existing in the field of security defense in industrial wireless sensor networks, novel theory and methodology will be investigated in this project. Based on the swarm intelligence and collaboration mechanisms, the theory and method, which are of the multi-objective optimization with multi-agent modeling, are proposed. To deal with the huge amounts of data acquisition problems in high speed industrial wireless sensor networks, an adaptive and adjustable data sampling method is proposed. Based on the bio-inspired principle, the models of dynamic security defense using combination of swarm intelligence are proposed. The models include intrusion detection model using artificial immune system, risk assessment and model response using game theory, the attack source localization model using particle swarm optimisiom, the fault-tolerant secure routing model with ant colony optimisiom. Multi-objective optimization methods are studied for artificial immune system, a particle swarm optimization algorithm and ant colony algorithm. Methods of dynamic antigen anomaly evaluation standard, the chaos disturbance and partial ordered sets are introduced to improve the performance of artificial immune system algorithm, particle swarm optimization algorithm and ant colony algorithm. And the convergence of swarm intelligence algorithms for multi-objective optimization are speed up. Based on the NS-2 simulation platform,the proposed model and methods will be validated. While the dynamic character of system will be simulated, the accuracy and effectiveness of the model will be evaluated and analyzed. And also the evaluation of security defense system resource consumption performance will be done. This project belongs to the basic research of application, and the research achievements of this project have important application prospect in key infrastructure such as smart grid in our country.

针对工业无线传感器网络在安全防御方面面临的共性挑战问题，探究新的理论和方法，基于群体智能协作机制，提出多代理多目标组合优化建模理论与方法。针对高速工业无线传感器网络海量数据采集问题，提出了一种自适应间隔可调数据采样方法；基于生物启发原理，提出组合群体智能安全防御系统动态建模，包括人工免疫系统入侵检测模型、博弈论评估响应模型、粒子群攻击定位模型、蚁群容错安全路由模型等；研究人工免疫系统、粒子群算法和蚁群算法的多目标优化方法，引入动态抗原异常评判标准、混沌扰动、偏序集等技术，改进人工免疫系统算法、粒子群算法和蚁群算法，加快群体智能算法的收敛性；基于NS-2仿真平台，对提出的模型和方法进行验证，模拟工业无线传感器网络的动态性，评价和分析模型的精确度和有效性。本项目属于应用基础研究，研究成果在我国智能电网等关键基础设施中有重要的应用前景。

工业无线传感器网络（Industrial Wireless Sensor network, IWSN）多层结构使得其面临多维度安全挑战问题，如入侵攻击、内部节点攻击、路由攻击等，安全防御是IWSN亟待解决的关键问题之一。但由于IWSN存在能量、计算能力和存储空间等资源约束，难以直接沿用传统的基于密码机制的安全技术，因此，亟需研究突破资源等限制的安全防御理论与技术。.针对工业无线传感器网络在安全防御方面面临的共性挑战问题，本项目研究IWSN中群体智能多目标组合优化安全防御理论和方法。基于可信机制，研究IWSN节点的可信度以及基于D-S证据理论的节点可信模型；基于协作机制，研究群体智能IWSN安全防御动态建模，建立人工免疫系统入侵攻击检测模型、粒子群攻击源定位模型、蚁群容错安全路由模型等；基于优化策略，研究IWSN安全防御动态模型多目标组合优化求解及算法，重点研究人工免疫系统入侵检测优化算法、多目标粒子群优化定位算法、多目标粒子群攻击源定位算法及多目标蚁群路由安全算法。.基于NS-2仿真平台，模拟工业无线传感器网络在攻击下的动态特性，对提出的模型和方法进行验证。评价和分析安全防御模型的有效性，对入侵检测模型的检出率、误报率等性能指标，对多目标攻击源定位模型的定位精度等性能指标，以及对多目标蚁群路由的丢包率、平均延时、能耗等性能指标，进行了仿真试验，仿真结果证明了方案的有效性。.本项目属于工业无线传感器网络和信息安全交叉领域的应用基础研究，主要为IWSN安全防御提供理论和技术解决方案，研究成果在IWSN关键基础设施中有重要的应用前景。