A novel approach for mobile network QoS evaluation

Nowadays, with growth the number of mobile data network subscriber in very competitive environment, operators are invited constantly to assess and improve the quality of their network. They can use many tools to accomplish this task like drive test to detect coverage problem or passive probe or OSS (Operating Support System) counter to have an overview on all network at any moment. The major problem of these methods is the huge volume of collected row data that need specific treatment. In fact, during one day, we can collect many Gigabytes of data and millions of records. Traditional data base management system could not be used to extract KPI (Key Performance Indicator) which can reflect the network QoS (Quality of Service). In general, operator use specific tools and spend long time to compute desired indicators. In this work, we aim to use intelligent method as decisional support system to help to evaluate the quality of service of a mobile data network. The used method is the self-organizing maps. It helps to extract the most important parameters to compute KPI and hence reducing the huge volume of data and time processing. To accomplish this task, we have used collected data during one week from a Tunisian GPRS (General Packet Radio Service) network operator.