Intelligent Test Bed Tuning for Improved Wireless Local Area Network Service Quality in Academic Environments

Abstract

Developing real time wireless local area network (WLAN) solutions require in-depth understanding of the WLAN system, performance evaluation in realistic setting, and visualisation of the service quality (SQ) in a very efficient manner. This paper presents the design and construction of WLAN test bed infrastructure to support intelligent tuning and visualisation of the SQ. To achieve this, received signal strength indication (RSSI) information and SQ field trials were performed on an academic environment, and, the requirements as well as challenges for developing suitable test bed infrastructure, appraised. An intelligent system model was then developed using the Interval Type-2 Fuzzy Logic (IT2FL), to simulate the SQ using RSSI information captured across three major campuses of the study environment. The IT2FL enabled the efficient modelling of uncertainties inherent in the field data for accurate estimation of the SQ. The processed test bed infrastructure provided direct visualisation as an initial assessment, before deploying personnel for corrective measures. Such measures are indeed necessary to assist in solving the poor quality of experience in academic environments. To ensure intelligent test bed tuning for effective coverage optimisation of the study environments, a particle swarm optimisation (PSO)- and genetic algorithm (GA)- adaptive neuro-fuzzy inference system (ANFIS) (or evolutionary ANFIS: PSO-ANFIS and GA-ANFIS) were independently trained. Results obtained showed that both systems performed well – as their root mean square error (RMSE) and mean absolute error (MAE) for both test and train data were very close, but PSO-ANFIS yielded the lowest RMSE and MAE for test data – indicating a more quality and accurate algorithm.