Artificial Intelligence Based Handoff Management for Dense WLANs: A Deep Learning Approach

Abstract

The traditional handoff management scheme in Wireless Local Area Network (WLAN) generates noticeable delays during the handoff process, resulting in discontinuity of service, which is more evident in dense WLANs. Inspired by the Software Defined Network (SDN), prior works put forward many feasible seamless handoff mechanisms to ensure the service continuity. However, when to trigger the handoff and which access point (AP) to reconnect to are still tricky problems. In this paper, we present RNN-HM, a novel handoff management scheme based on deep learning, specifically recurrent neural network (RNN). The proposed scheme enables the network to learn from the actual users' behaviors and the network status from the scratch. Centralized control over the handoff is eventually realized using SDN, setting the network free from parameter configurations. A preprocessing data representation leveraging the signal-to-interference-plus-noise ratio (SINR) is introduced to characterize the system state. Numerical results through simulation demonstrate that RNN-HM can effectively improve the data rate during the handoff process, outperforming the traditional scheme.