RAT selection for heterogeneous wireless networks using support vector machine

The next-generation cellular networks are envisioned to operate with heterogeneous wireless networks. Under such scenario, user device is required to periodically perform sensing for available RATs and switch between the one that best fits its current usage scenario. Such requirement may be highly energy-consuming; hence, efficient RAT selection mechanism is necessary to help user device select the best available RAT for its data transmission. In this paper, we propose a mechanism to make accurate RAT access recommendations to user devices by the Support Vector Machine tool, which can reduce unnecessary energy consumption due to periodic sensing. Simulation results demonstrate that our proposed methodology is highly accurate and capable of maintaining the average number of sensings as low as possible.