Fuzzy Logic Model for Preference of Parameters in Network QoE Analysis

Abstract

Two important Quality of experience (QoE) parameters in computer networks service related performance metrics are throughput and delay. A low delay indicates high network efficiency thus high throughput. Similarly, the entire four parameters linked to the integrity of service are considered to be primary factors affecting any computer network. The study’s objective is to identify the vital primary parameters with preeminent results to be kept into consideration when analyzing network QoE by using fuzzy logic methodology. A local area network of 64 nodes was used for data collection of the four parameters namely: Throughput, delay, packet loss and jitter. Fuzzy logic methodology was implemented using experimental research design. Five input linguistic terms were utilized: High, very high, medium, very low and low. Five output linguistic terms were used to refer to the opinion scores: Bad, poor, fair, good and excellent. The four parameters were used for the four variables model generating 625 rules (5^4). The rules were further condensed to 240 logical rules centering on expert knowledge. Likewise delay and throughput were used for the two variables model spawning 25 rules. The rules were further condensed to 17 logical rules centering on expert knowledge. Firstly, the two models analysis results were compared in the rule viewer where the four parameters model had better QoE values for different setups compared to the two parameters model. Secondly, the analysis of results using independent sample T-test technique which compared mean performance of the two models was done. The results specified a significant consequence using the entire four parameters model compared to using a two parameters model for analysis of computer network QoE whereby the score for the p-value based on a confidence interval of 0.95 was 0.017, indicating a significance difference of the 2 models. The target population for this model is the service providers enabling them to have the best receptive measures to deal with network service provision in order to deliver the quality of service (QoS) as per service level agreement (SLAs) thus achieving finest network QoE.