Comparing the performance of ANFIS, PSO- ANFIS and PSO-ANFIS with random input in indoor Wi-Fi Signal propagation prediction

Abstract

With the increase in the use of mobile devices fitted with wireless local area networks (WLANS) technologies there is need for accelerated studies on these systems to improve on the quality of service (QoS) provided to the users. Different methods have been used in signal modeling including deterministic and empirical models. This study is aimed at comparing the performance of predicting Wi-Fi signal propagation along a corridor using Particle Swarm Optimization (PSO) trained Adaptive Neural Fuzzy Inference System (ANFIS), ANFIS and PSO trained ANFIS with a random input. The mean square error, root mean square and standard deviation of the predicted signal were determined and compared. The study was undertaken using a Wi-Fi router as the transmitter and a mobile phone as the receiver in the process of data collection. The measured values were then used in the modeling. It was found that the predicted values based on PSO trained ANFIS with a random input were close to actual measured values as from the undertaken analysis giving the best prediction. Keywords; Wi-Fi, QoS, WLANS, ANFIS, PSO-ANFIS