| In recent years, with the tremendous development of wireless technology and computer networks, wireless network technology has been maturing rapidly and widely applied into military, civil and commercial areas. In the meantime, the study on the security and performance of wireless networks is becoming a focus. In this dissertation, the secure localization and the optimal routing design for wireless networks are studied based on the property of wireless networks. The main contributions of our work are listed as follows:(1) A novel secure localization scheme, called ALD (Active Location Detection), for adversary nodes in wireless networks is proposed, filling a current technological void.In current practical wireless networks, the growing interest in wireless data service and security requirements for wireless networking necessitates the development of an effective scheme which can locate an adversary in a wireless networks to make him account for his misdemeanors or expel him out of the network. On the other hand, while localization has been an active area of research recently, current research of secure localization mostly focuses on obtaining correct locations of legitimate users despite the existence that adversaries can malicious attack with deliberately hiding its true location. In fact, an adversaries equipped with advanced radio technologies can hide from almost all of the exsiting localization systems. To solve this challenging problem, a novel localization scheme called ALD is proposed, which can locate the attacker with traditional range-free localization equipments. The main idea is to use coordination of multiple access points to locate the adversary node and optimize the process with a finite horizon discrete Markov decision process (MDP). The ALD scheme can be supported by IEEE 802.11 and many other wireless network standards. According to the simulation results, the estimation error is lower than 20%.(2) A modified ALD scheme is proposed, which further improve ALD's usage in practical systems.In standard ALD, the oversize of initial estimation area may increase the localization steps and computation complexity. In this dissertation, the performance of ALD in practical wireless networks is studied and an improved algorithm for ALD is proposed, which can significantly improve the performance of the standard ALD algorithm by inserting Pre-Distance-Estimate (PDE) process in ALD to reduce the initial estimation area. As it is shown in simulation part, the improved ALD method outperforms the traditional ALD method in both performance and computation load.(3) The performance optimization of multi-hop routing protocols for wireless networks under a given performance metric.Recently, a huge number of routing protocols have been proposed for Ad hoc networks. But few research analyses the optimal performance of Ad hoc routing protocols under a certain performance metric. At the same time, the performance difference of Ad hoc routing protocols under different parameter configurations is huge. In order to assess and optimize the performance of Ad hoc routing protocols, a novel scheme is proposed which do a global optimization for routing protocols parameter setting based on genetic algorithm. The presented algorithm adopts encoding parameters and applies selection, crossover and mutation to optimize the routing protocols performance of Ad hoc networks. As shown in our simulations on ns2, the performance of optimized routing protocols can be improved more than 30%.(4) Multi-objective optimization of wireless Ad hoc networks routing algorithm is proposed.The large number of performance metrics for Ad hoc routing protocols such as lost rate, delay and throughput necessitates the performance optimization of routing protocols for wireless networks under multi-metric. In this dissertation, the performance of Ad hoc networks routing protocols is studied and a novel multi-objective heuristic algorithm is proposed to do a global optimization for routing protocols parameters setting based on Pareto Genetic Algorithm. There are five operators in Pareto GA, which are selection, crossover, mutation, Niche, Pareto set filter. The result set is provided in the form of Pareto front from which decision-makers may choose some suitable results according to their decisions. As shown in our simulations on ns2, the lost rate and average delay of optimized routing protocols can be greatly improved.(5) A novel open architecture called Orchestra (Open Architecture for Checking, Synthesis, and Training of Routing Algorithms) is proposed, which can efficiently compare the performance of different routing protocols and combine these protocols.Today, it is difficult to track all the routing protocols designs and choose the optimal protocol for an Ad hoc net work under various scenarios from huge number of existing routing protocols during wireless networks designing. In addition, it is not clear if any of the original design ideas in the huge number of existing works can be combined to form new and better routing systems according to current routing research. To address this challenge, in this dissertation, a new open architecture for routing designing is proposed, called Orchestra, which can automatically assemble and test of a great variety of routing designs with high flexibility and compatibility by the help of routing algebra. Furthermore, it can also be combined with genetic algorithm to find the most suitable protocol design for a certain network scenario as well as the parameters setting for the protocol design, as which is shown in our simulation result. Orchestra greatly reduce the implementation burden of routing protocols designer and improve their research efficiency. The research of Orchestra also have significant scientific value. It combines advanced routing algebra, formal verification theory, software engineering techniques, physical and MAC layer modeling theories. Orchestra will bring the routing research field into the new era of automatic system design. Such an evolutionary change will greatly enhance the advances of networking technology. |
PDF Full Text Request