Node Localization in Wireless Sensors Network using Application of Evolutionary Algorithms

Abstract

Wireless Sensor Networks (WSNs) comprise a large number of sensing units/nodes which ingest data and communicate wirelessly. Localization of these sensing units/nodes plays a vital role in WSNs applications. WSNs appear promising for a number of military and commercial applications. WSN deployments should never be considered completely static; instead, the mobility component should be considered as well. Establishment of the coordinates of an unknown node/target node (whose position is not known) in a network is a challenging process, and such techniques are referred to as localization techniques Localization methods are categorised as static vs dynamic, distributed vs centralised, outdoor vs indoor, two-dimensional vs three-dimensional, and so on. In static localization, the target node coordinates can be computed only once. However, in a mobile context, coordinates estimation of the target node is a continuous process. In such mobile circumstances, more time, energy, and the availability of a quick localization service are necessary. When mobility is considered in the network, several hurdles arise during establishing the coordinates. Furthermore, there are three types of dynamic localization processes: dynamic targets and static anchors (whose position are known), static targets and dynamic anchors, and both dynamic. Dynamic behaviour of either anchor or target nodes is a main feature of such mobility models. Because node position change often in mobile environments, target node position must be re-evaluated on a regular basis. Even in harsh environments, the precise location of target nodes in wireless sensor networks becomes more important. Anisotropy, on the other hand, has an impact on determining the precise location. The accuracy of the target node's position is determined by connectivity, power variation, and signal attenuation. To reduce the effect of anisotropy conditions and improve location accuracy, distance and angle information of the signal are fused together, which are extracted separately from Received Signal Strength (RSS) and Angle of Arrival (AoA). A novel concept of virtual anchor node is considered here for target node position estimation. Furthermore, emerging artificial intelligence applications such as particle swarm optimization (PSO), Hybrid PSO (HPSO), and Firefly Algorithm (FA) are used separately to achieve optimal target node positions. When compared to existing techniques, the proposed methods outperform them in terms of accuracy (average error of 0.274 metres), energy, scalability, and convergence time. The main advantage of this work is that it only requires an anchor node to estimate target node 2D coordinates. According to the thesis' second contribution, the functioning of WSN is bound by the limited battery power of sensor nodes, and thus energy conservation has become a crucial design challenge in such networks. Excessive power transfer not only shortens the lifespan of sensor nodes, but it also causes interference in the shared radio channel. WSN localization is investigated from the perspective of transmit power regulation. Closed form formulas for the minimal transmitted power required for sensor node localisation in anisotropic WSNs. To achieve the lowest transmitted power for localization, log normal shadowing, deterministic path loss, and Rayleigh fading channel models are used. The effects of anchor nodes spatial density and the propagation environment on the minimum transmitted power are investigated both analytically and through simulations in an anisotropic environment for node localization. The network's energy efficiency and life-time can be increased by statistically characterising the minimum transmitted power required for a sensor node's localizability. The last contribution of the thesis asserts that distance information in WSN is non-linear, characterised by unpredictability and uncertainty, resulting in nonconformity, which has a significant impact on target node position. The non-linearity influence between received signal strength (RSS) and distance of target nodes has been minimised using a knowledge base system to describe edge weight in a fuzzy logic system. In addition, to reduce node location error, ideal membership functions based on RSS and edge weights are generated using the butterfly optimization procedure. The 3D coordinates of target nodes in an anisotropic environment are determined utilising range-free localization techniques employing a single anchor node. The anchor node has been placed at the top layer, and target nodes have been distributed uniformly across the underneath layers. In terms of target node 3D location accuracy, the proposed strategy greatly exceeds previously published range-free methods. The proposed method can be used in a range of scenarios, such as coal mining, habitat monitoring, and plant processing.