Dimensionamento de rede sem fio de um campus universitário baseado na análise de logs de conexão

Abstract

Computer networks are increasingly heterogeneous and their users access different types of applications using different technologies. Access to computer networks using mobile devices has increased, making massive use of wireless connections. It becomes necessary to know the connection locations of users within the network to be able to design and scale networks with quality of service and improve the user experience. With the connection information, it is possible to improve resource provisioning, direct traffic more effectively, depending on the type of service and user location, save device energy or even predict security and network performance issues. In this work, it was proposed the generation of a database (dataset) based on the files of logs connection to theWi-Fi network of a university campus, with approximately 15 thousand users. This dataset contains information for a period of 4 months, with time intervals divided into 15 minutes, connection locations, types of users, number of connections, among other data. This dataset was anonymized and serves as a basis to assist in making decisions regarding the dimensioning of the number of access points needed. Thus, a methodology based on the minimum bandwidth per user was defined for calculating resource sizing. Frequency allocation, coverage overlap between APs operating on the same radio channel and transmission power control were not evaluated. All these issues are automatically resolved by the controller of the wireless network equipment. A limitation found is due to the challenge of the so-called Ping Pong effect, which corresponds to the recording of a succession of connections/disconnections between APs while the user is stopped. To work around this problem, the APs of the same building/sector were grouped together. This work differs from previous works in the area by creating and making available a categorized database with connection locations and by proposing equipment sizing based on bandwidth for the user based on the number of connections per location. To define the minimum bandwidth reserved for each user, a guide of typical online services was used with the minimum download speed necessary for the adequate performance of each application described in the text. The result indicates which locations need the addition of equipment. Interactive georeferenced maps were also created, where it was possible to locate the places with the highest concentration of users over time.