Machine learning-based traffic offloading in fog networks
No Thumbnail Available
Wi-Fi offloading in fog networks is believed to be one of the best ways to solve the significant data increase in cellular networks, since nodes located close by are used as relays for offloading traffic and computations. This alarming growth has affected these networks and has put its mark on their performance. Some operators might try upgrading the wide area networks, but in most scenarios, this is not the most cost-effective and optimal solution. These operators would benefit more from intelligent offloading solutions. Therefore, we can use Wi-Fi networks to send some of the packets, thus releasing cellular networks and decongesting traffic. This paper deals with offering a complete offloading solution and presents various profiles aimed at different purposes: saving the battery, getting the maximum data rate, or balancing the two, as well as offering a simulator that reproduces the behavior of the devices in an environment as close to reality as possible. Through extensive analysis, we show that the proposed solutions are able to improve certain metrics based on user requirements.
Fog, MobileWi-Fi, Mobile broadband, Multi-path TCP, Offloading, Machine learning, Genetic algorithm