Changeover prediction model for improving handover support in campus area WLAN

A handover in wireless networks is the mechanism to maintain quality of service (QoS) by transferring an ongoing call or data connection from one point of access to another. IEEE802.11-based wireless local area networks (WLANs) do not explicitly support handovers, but permit a changeover if two access points (APs) have identical identifiers. This changeover is triggered by the mobility of a mobile host or an AP's inability to support hosts in its vicinity, and such changeovers are not seamless. The changeover is initiated by the mobile host which results in disruptions to ongoing connections. In this paper, we develop a changeover prediction model as an evidence-based tool to help minimise such disruptions. We determined the best set of predictors (i.e. a model) for changeover prediction using variable selection and multiple regression. The results show that our approach yields a consistent, reproducible changeover prediction model across different locations around a campus area WLAN. This finding can provide significant insights to the design and development of future handover algorithms built on top of IEEE 802.11 WLAN.