An Adaptive Positioning System for Smartphones in Zigbee Networks Using Channel Decomposition and Particle Swarm Optimization

ZigBee wireless sensor networks has the benefit of superior optimization in power consumption and extremely long battery life. In near future, the handheld devices are expected to incorporate support for wireless sensor networks technology, and these devices can become active elements in ZigBee-based applications such as automated buildings and advanced metering systems. Use of these applications is also promising for the expansion of location services into environments that do not have access to Global Navigation Satellite Systems (GNSS). Compared with the ToA (Time of Arrival) and TDoA (Time difference of Arrival) methods, RSSI (Received Signal Strength Indicator) has several advantages. It does not require additional hardware. Owing to the simplicity of RSSI, the most recent studies on localization using wireless sensor networks use RSSI-based algorithms. Previous works RSSI-based location estimation methods discussed by the literature depend on the impractical assumption that signal propagation characteristics are known and independent of both time variations and the environment. That assumption is invalid for the complex structures of indoor areas, which make the signal propagation characteristics for the same node differ significantly based on the orientation, even if the radiation pattern of the node is omnidirectional. Additionally, obtaining knowledge about signal propagation characteristics requires using manual data collection, which is impractical and time consuming. Furthermore, location estimation techniques must consider the large variations and high degree of uncertainty in RSSI observations. In this paper, we developed an adaptive localization system with low power and low cost, IEEE 802.15.4 ZigBee WSN. The under investigated system does not require manual pre-data collection and operates without performing exhaustive radio surveys. The first contribution in this paper is involving Kalman filtering for RSSI smoothing. Secondly, estimation of multi-link signal propagation characteristics; path loss exponent and Gaussian random variable in real-time by utilizing Particle Swarm Optimization (PSO). The results from the experiments are taken show that using the proposed multi-link channel decomposition estimation technique is more effective than using only one fixed value of PLE and X0. The overall localization accuracy is within two meters in root mean square.