|PIMPINELLA ANDREA||Cycle: XXXIV |
Tutor: CESANA MATTEO
Advisor: REDONDI ALESSANDRO ENRICO CESARE Major Research topic
:Machine Learning Strategies for Cellular NetworksAbstract:
Machine Learning is one of the most promising artificial intelligence tools conceived to support next generation wireless networks. On the one hand, learning strategies can be used to dinamically assist smart radio terminals in the decision-making processes, such to be able to satisfy the diverse requirements of next generation application scenarios. On the other hand, offline prediction of network performances will help operators to recognize and locate potentially under-performing network areas, such to efficiently prioritize investments for technological upgrades. Furthermore, operators can leverage learning tools to attempt to predict the Quality of Experience (QoE) of customers starting from network measurements. In fact, while network operators are interested in continuously monitoring the QoE of their customers, collecting users feedback through direct surveying campaigns is a cumbersome task and thus the possibility of predicting users satisfaction looking solely at network objective performances rapresents a chance of minimising the chunk rate (i.e. the percentage of users leaving the operator due to unsatisfactory service). This thesis will address the task of evaluating and assessing performances and impact of different machine learning algorithms applied to diverse cellularnetworking scenarios, leveraging large quantities of network data coming from direct or aggregated objective measurements.