ANFIS-based quality prediction models for AMR telephony in public 2G/3G mobile networks

The scope of this research paper is the proposal of voice quality prediction models, based on Adaptive Neuro-Fuzzy Inference Systems (ANFIS), for Adaptive Multi Rate (AMR) telephony service provided by GSM and UMTS rollout public land mobile networks. For the scope of our research we performed an experimental drive-test measurement campaign in order to evaluate objectively the End-to-End Service QoS (ESQoS) as well as radio System QoS (SQoS) parameters. Subsequently, the collected measurement data are used to train ANFIS empirical models. We then assess the prediction performance by depicting 2D/3D FIS surfaces and the impacts of SQoS to ESQoS. The prediction methodology can be successfully applied in Quality of Experience centric radio network planning, fine tuning and optimization processes by mobile operators.