الفهرس | Only 14 pages are availabe for public view |
Abstract A Wireless sensor network (WSN) is a special type of ad-hoc networks, which consists of a large number of small lightweight sensor nodes and one of more sink nodes. WSNs are used in a wide range of applications ranging from structural, environmental and habitat monitoring to military surveillance and intrusion detection. In WSN, coverage plays an important role in the successful operation of the network. Coverage is mainly concerned with how well an area of interest is observed with the deployed sensor nodes. Although deterministic deployment is preferred as it ensures proper coverage of the deployment area, random deployment; however, is more practical in some environments, especially those with harsh conditions. In spite of its feasibility, random deployment may result in poor coverage; therefore, approaches are needed to enhance initial coverage. While, in dense static WSN, scheduling sensor nodes into sleep is usually used to extend the lifetime of the network, and ensure full coverage, mobility in mobile WSN has been recently utilized in healing coverage holes or for dynamic deployment. Dynamic deployment ensures full coverage by redistributing nodes after initial random deployment. Dynamic deployment approaches can be categorized into virtual forces, computational geometry, geometrical patterns, and evolutionary computation algorithms; such as Particle Swarm Optimization (PSO) and Articial Bee Colony (ABC). Virtual based approaches depend on o-line conguration, while computational geometry approaches are more suitable when global location information of the nodes is available, using Global Positioning System (GPS) for example. In this thesis, The use of the Harmony Search (HS) optimization algorithm to the Dynamic Deployment (DD) problem in WSNs is investigated. To this end, a family of dynamic deployment algorithms based on well known variants of the HS algorithms is proposed and their performance is analyzed. In particular,ve algorithms are implemented and evaluated; namely, Harmony Search-DD (HSDD), Improved HS-DD (IHS-DD), Global HS-DD (GHS-DD), Dierential HS-Dynamic Deployment (DHS-DD), and Self-adaptive HS-DD (SaHS-DH). The proposed algorithms aim to maximizing both network coverage and connectivity. Simulation results show that GHS-DD achieves the best coverage improvement compared to the known theoretical bound (99%-Coverage for equal sensing and communication radius, and 89%-coverage for communication radius larger than sensing radius) with the minimum moving distance. Whereas SaHS-DD provides better connectivity with a reasonable coverage improvement for dense networks. Moreover, in deployment areas with obstacles, GHS-DD shows a better performance and higher coverage degree as compared to the rest of the algorithms. Simulation scenarios are conducted to nd the minimum number of nodes that need to be moved in order to achieve the target coverage improvements above (99%-Coverage for equal sensing and communication radii, and 89%-coverage for communication radius larger than sensing radius) , and hence, prolong the overall network lifetime. Simulation results show that, target coverage improvements can be achieved by moving only 30% of the deployed sensor nodes, while keeping the rest in their original positions. In addition, a new harmony search based algorithm; namely, Harmony Search - Continuous Coverage Optimization (HS-CCO) algorithm is proposed,that attempts to continuously maintain coverage as the topology of the WSN changes due to node failure. The proposed algorithm is capable of maintaining diversity in addition to the adaptive parameter selection and the global best harmony strategies used by the proposed algorithm help in maintaining coverage. Finally, another harmony search based algorithm; namely K-Coverage Enhancement Algorithm (KCEA) is proposed, that is scalable in terms of execution time. KCEA attempts to enhance initial coverage, and to achieve the required K-coverage degree for a specic application eciently. Simulation results show that the execution time of adding a single sensor using KCEA is 2.8 seconds, while the execution time of KCRD1 is 33.1 seconds. Moreover, coverage improvements resulting from adding a single sensor node using KCEA is 5.4% compared to K-Coverage Rate Deployment (K-CRD), which achieves 3.2%. |