Localisation of wireless sensor network with mobile beacon by dynamic path

Publication Type:
Issue Date:
Full metadata record
Files in This Item:
Filename Description SizeFormat
01front.pdf137.33 kBAdobe PDF
02whole.pdf3.28 MBAdobe PDF
Small size and low-cost sensors are practicable because of evolution of the semiconductor field, which is led by increasing miniaturisation. They are still limited in processor capacity, memory size and energy resources; however, ubiquitous wireless is added to extend their communication capacity. Wireless sensor networks (WSN) are formed by large numbers of such sensors and can be used to monitor a field of interest in military and civilian areas. The resulting data are only meaningful when combined with geographical position information of the sensors. Both the Global Positioning System (GPS) and the Global System for Mobile Communication (GSM) are hungry for energy and expensive, and are not suitable to be used extensively in every sensor. But localisation is essential in WSN, which should be implemented with help of some beacons that are equipped with GPS or GSM. A mobile beacon (MB) is the replacement of many static beacons; it is movable and flexible and can be powerful so that some heavy computational mathematical methods (such as probability and graph theory) could be applied in an algorithm of localisation. The walking path of a MB will determine the rate of coverage and accuracy of localisation. The static path is planned before action and is suitable for regular terrain; whereas, the dynamic path is decided in real-time action depending on the demand of unknown sensors, and is more efficient than the static path. Concentrating on the algorithm of dynamic path to reach a better result in terms of accuracy, coverage, and trajectory of localisation in WSN, a framework of dynamic path of mobile beacon (DPMB) is proposed first, and then reinforcement learning (RL) is fit to the DPMB as the inner controller to improve the performance. Finally, direction is employed to assist the MB to find a better next position instead of distance in the DPMB. Simulations demonstrate that the performance is improved gradually.
Please use this identifier to cite or link to this item: