Adaptive Neural Fuzzy Inference System for Accurate Localization of Wireless Sensor Network in Outdoor and Indoor Cycling Applications

When localizing wireless sensor networks, estimating the distances of sensor nodes according to the known locations of the anchor nodes remains a challenge. As nodes may transfer from one place to another, a localization technique that can measure or determine the location of a mobile node is necessary. In this paper, the distance between a bicycle when moves on the cycling track and a coordinator node (i.e., coach), which positioned on the middle of the cycling field was estimated for the indoor and outdoor velodromes. The distance was determined based on two methods. First, the raw estimate is done by using the log-normal shadowing model (LNSM) and later, the intelligence technique, based on adaptive neural fuzzy inference system (ANFIS) is applied to improve the distance estimation accuracy, especially in an indoor environment, which the signal is severely dominated by the effect of wireless multipath impairments. The received signal strength indicator from anchor nodes based on ZigBee wireless protocol are employed as inputs to the ANFIS and LNSM. In addition, the parameters of the propagation channel, such as standard deviation and path loss exponent were measured. The results shown that the distance estimation accuracy was improved by 84% and 99% for indoor and outdoor velodromes, respectively, after applying the ANFIS optimization, relative to the rough estimate by the LNSM method. Moreover, the proposed ANFIS technique outperforms the previous studies in terms of errors of estimated distance with minimal mean absolute error of 0.023 m (outdoor velodrome) and 0.283 m (indoor velodrome).