Event prediction in wireless sensor networks

OData support
Dr. Vida Rolland
Department of Telecommunications and Media Informatics

Recent development in micro-electro-mechanical systems (MEMS) technology, wireless communications, and digital electronics make it possible to establish low-cost, low-power, multifunctional sensor nodes, which are small in size and communicate untethered in short distances. These sensor nodes contain processing, sensing, communicating components as well. With the implementation of these sensor nodes the idea of wireless sensor networks could come into reality. These are based on the collaboration of large number of independent sensor nodes. The network itself consists of many autonomous intelligent sensors, and they solve a specific problem with divided mode. The reason of the divided working mode is mainly that the expansion of the physical system, that we would like to measure or make effect on it, make it impossible to solve the problem with one central device.

An interesting and useful feature of this kind of network is called event forecasting. In this feature the sensor nodes try to forecast specific events from the changes of the environmental parameters and from formerly registered measurements. There are many possible usages of this feature in the application area. For instance it can be used in the field of healthcare in patient monitoring systems, in intelligent transport systems or in the sleep scheduling protocols of WNSs as well. In my thesis I introduce such a kind of forecast system that is based on the fuzzy set theory.

In the dissertation I introduce the basic working method of wireless sensor network, the elements of its architect and its software environment. The radio interface and the ATMEL processor of the CrossBow Micaz sensor node are also described in general. After that I present an implemented sensor network, the purpose of which is to forecast from the formerly registered discrete events. The system posesses a user interface that can run on Windows operating system. It can set some of the adjustable parameters of the sensor nodes. It can also visualize the current state of the sensor network according to the state messages of the nodes.


Please sign in to download the files of this thesis.