Deployment Optimization Of Wireless Sensor Networks For Marine Oil Spills Monitoring

With the development of shipping and exploitation of the underwater natural resources, the potential risk of marine oil spill is increasing. Vast and complex ocean environment make it hardly to monitor oil spill event and evaluate the environment damage of the event. It is necessary to detect the event of marine oil spill and take a quick respond. The effective measures would keep the oil contamination from expansion and minimize the effects of the accident. Compared to traditional detecting methods, such as remote sensing and patrol boats, Wireless Sensor Networks (WSN) characterizes cost-effectiveness, full-time, energy-aware, flexibility and easy deployment, thus it has been the hot research interest in water monitoring. And, WSN is capable of reporting the water status and pollutant source timely. So, the information collected from the target area would support the timely and effective decision-making in the emergency situation. In a world, the application of WSN in oil spill monitoring will be promising research field.In this thesis, the target area is to monitor the marine oil spill. The deployment strategy of the network has always been one of the fundamental problems. It is defined as sensor node deployment and optimization (SNDO) problem. Combined with the coverage performance, network connectivity and energy cost, we aim to discuss an oil spill monitoring network which is based on WSN. To satisfy the coverage requirement of target area, the deployment strategy and energy-aware topology are taken as the main factors to establish a cost-effective and energy-aware network.This research is supposed to provide a guideline for WSN-based oil spill monitoring network in practice. Firstly, we formulate the SNDO problem by a mathematic model of WSN, which is under the constraints of coverage requirement and maximum hops of the communication path to sink node. So, the SNDO problem is been solved by an Integer Linear Programming (ILP). Secondly, to verify the effectiveness of the mathematic model, Gurobi optimizer is been employed to solve the simulation scenarios with 1-coverage and 2-coverage requirements. Besides, large scale deployment of WSN may degrade the efficiency of Gurobi optimizer. So, Genetic Algorithm (GA) is designed to solve the SNDO problem, and its effectiveness is also been verified. Thirdly, a GA with the variable length of chromosomes is proposed to control the mean fitness of the population. The selection function of candidate points would cut the genes with low contribution in the longer chromosomes, or fill the genes with high contribution into the shorter ones. Compared to Gurobi optimizer, simulation results show that the proposed GA could give a feasible deployment solution in large scale network.