Automate Minimization of Drive Tests for QoE Provisioning: The Case of Coverage Mapping

Abstract

The concept of Minimization of Drive Tests (MDT) has been developed in Third Generation Partnership Project (3GPP) specifications standard. With MDT, Mobile Network Operators (MNOs) were enabled to remotely collect measurements indicating the network Quality of Service (QoS), as experienced by their users, correlated with actual location information. This results in wider application of use cases that allows network monitoring and optimization, without the need for conventional drive tests. To facilitate access to distributed geospatial data through a set of policies, common rules, and standards that would improve interoperability, this paper, proposes an open architecture, in compliance with the Open Geospatial Consortium (OGC) web services and the MDT architectures, enabling the automation of the MDT use cases and real-time service management. General TermsService management, location-based services, geospatial web services Keywords Cellular radio coverage, quality of experience, quality of service

Semi-supervised learning-based coverage hole detection in cellular networks

“…The same authors in [13] proposed a REM cognitive toolbased approach that provides REM where data was coming from location-aware devices or basically MDT. The approach seemed to function very well but the input data was from a planning tool and the obtained results were based only on models [14]. The authors of [14,15] focused on the QoS evaluation using different Key Performance Indicators (KPIs) and correlated with location data to investigate how satis ied end users were.…”

“…The approach seemed to function very well but the input data was from a planning tool and the obtained results were based only on models [14]. The authors of [14,15] focused on the QoS evaluation using different Key Performance Indicators (KPIs) and correlated with location data to investigate how satis ied end users were. The authors in [16] also focused on the QoS and used ML algorithms such as k Nearest Neighbour (kNN) to characterise the satis ied and unsatis ied users.…”