Research On Optimization Models And Algorithms For Some Key Issues In Wireless Sensor Networks

A wireless sensor network is a network of wireless connected devices that can sense or monitor physical or environmental conditions cooperatively. It has a wide range important applications such as in military target tracking and surveillance, natural disaster prediction, biomedical health monitoring, and hazardous environment exploration and seismic sensing. WSNs has become a new hot research area in the field of information science and technology. Therefore there are many key technologies to be studied. Our research in WSNs aims to meet the actual constraints by introducing new design concepts, creating or improving existing models, building new applications, and developing new algorithms to improve the robustness and reliability of WSNs. The main contributions of this thesis are as follows:1. It is crucial that wireless sensor nodes should be properly located to facilitate the operation of the whole network. When WSNs is exposed in malicious and dangerous environment, attackers may attack the locating process of the nodes and cause incorrect location results which may lead to the complete breakdown of the entire network. The sensor location probability model based on maximum likelihood estimation is one kind of commonly-used location models. However, it has two drawbacks: First, it usually treats the standard deviation of the received signal strength(RSSI) as a constant in order to reduce the calculation complexity, which results in the inaccuracy of location results. Second, it is not secure enough. Under malicious node attack, this model usually cannot fulfill its location function. This research induces the function relations between RSSI standard deviation and distance through fitting test data and thus fixes the first problem. To tackle the second problem, this research analyzes the reasons behind the failure of location under attack and improves the probability formula of node location, then designs a new sensor node location probability model based on the characteristics of variant variance. As this new model is a highly nonlinear characteristic global optimization problem that is difficult to work out, this research has designed a new and effective evolutionary algorithm for it and proven the global convergence of the proposed algorithm.2. The exposure on the minimum exposure path(MEP) is one of the important indicators to measure the quality of coverage, showing how well a mobile target through the sensing field is monitored or not and how well the sensing and monitoring abilities that a given WSN can offer. To avoid the defects of the classical methods for the MEP problem, like the grid-based and the Voronoi-diagram-based methods, that are not accurate enough, too complex, and not applicable with large-scale sensor nodes, this paper builds up the numerical functional extreme(NFE) model for the MEP problem based on functional analysis theory. This model is mathematically a high-dimension non-linear optimization problem which classical mathematical optimization methods fail to solve. This paper proposed a hybrid genetic algorithm by incorporating with the background characteristics of the sensor nodes coverage problem, and designed unique and efficient crossover operator, mutation operator, local search operator and upside-down operator. The global convergence of this evolutionary algorithm is proved and the experiment results indicate the proposed model is effective and the proposed algorithm is efficient.3. The original MEP problem in wireless sensor network does not consider the constraints for paths in practice and thus can not reflect the real situation. This thesis puts forward a new MEP Problem with requiring path go along a part of the boundary of the special protection area(BPA-MEP). For this problem, the classic methods(such as grid-based method and Voronoi-based method) is unable to set up the corresponding graph model and thus cannot work on the BPA-MEP problem. To solve the BPA-MEP problem, a highly nonliear and high dimensional optimization model is tailored and established, and then by taking the characteristics of the distribution of the sensor nodes into consideration, a hybrid artificial bee algorithm is proposed to solve this complex optimization model. The experiments are conducted and the results show that the proposed model is resonable and the proposed algorithm can effectively solve BPA-MEP problem.4. The survival time of a WSN is usually very limited due to the limited energy of common sensor nodes and the fact that the energy will be used up eventually. This will result in the so called coverage holes. It is an important and very hard issue to design the optimal path connecting the unworking nodes such that these nodes can be repaired before their energy is used up.This thesis designs a repairing strategy for the coverage of the sensor network based on multi-mobile nodes and path planning. Such mechanism can ultimately repair common nodes in a timely fashion via energy consumption analysis of the nodes by using genetic algorithm to calculate the reasonable moving path of the mobile nodes, which can avoid coverage holes caused by the early death of sensor nodes. The global convergence the designed algorithm is proved. Stimulation experiments are conducted and the results indicate the effectiveness of the designed mechanism and the relevant solution algorithm.