Thesis (M.Sc.) - Cairo University - Faculty of Engineering - Department of Electronics and Communications

While mobile operators invest in providing the best quality of service (QOS) to its customers, a live visibility on the actual QOS at the customer end is often needed. Mobile operators rely on drive test to measure QOS at user level thus identify the service level. When this visibility is inaccurate or not live, detecting and acting on customer problems can take lengthy timeframes.The thesis proposes machine learning models using huge historical dataset collected from actual filed readings to predict the QOS received at the customer level indifferent locations. Five ML approaches are examined, and the results were compared to identify the ML model that can offer higher prediction accuracy for QOS.Then Clustering ML model was built to divide the coverage area into small areas such that probe devices can be used to collect field readings from specific locations to improve the predictive model

Issued also as CD