Податочна фузија кај безжични сензорски мрежи

Abstract

Following the latest developments in computer and communication technologies, everyday objects are becoming smarter, as ubiquitous connectivity and modern sensors allow them to communicate with each other. The deployment of sensors and actuators everywhere around us adds a new dimension to the world of information and communication, which enables the creation of new and enriched services widely applied in different industrial and civilian application areas. The aim of this thesis is to develop data fusion strategies for Wireless Sensor Networks (WSN) that remove temporal or spatial redundancies between sensor measurements in order to decrease the network load. In context of spatial data fusion, nodes localization appears as the first problem to be solved in order to find spatial correlation between data from neighboring nodes. In this thesis two approaches were purposed, implemented and evaluated for solving three dimensional WSN localization problems. In context of temporal data fusion,different techniques for data reduction based of time series forecasting were analyzed, implemented and evaluated. From the evaluation of the algorithms it can be concluded that star network is the most suitable network topology by means of energy saving. If sensors are not within each other radio range, cluster-based topology could be used. Additionally, if data aggregation is applied at cluster head, this topology can achieve even greater data reduction in scenarios where loosing data precision is affordable.