Predictive routing of contexts in an overlay network

While mobile nodes (MNs) undergo handovers across inter-wireless access networks, their contexts must be propagated for seamless re-establishment of on-going application sessions, including IP header compression, secure mobile IP, authentication, authorization, and accounting services, to name a few. Routing contexts via an overlay network either on-demand or based on prediction of an MNs' mobility, introduces a new challenging requirement of context management. This paper proposes a context router (CXR) that manages contexts in an overlay network. A CXR is responsible for (1) monitoring of MNs' cross-handover, (2) analysis of MNs' movement patterns, and (3) context routing ahead of each MN's arrival at an AP or a network. The predictive routing of contexts is performed based on statistical learning of (dis)similarities between the patterns obtained from vector distance measurements. The proposed CXR has been evaluated on a prototypical implementation based on an MN mobility model in an emulated access network. Our evaluation results show that the prediction mechanisms applied on the CXR outperform a Kalman-filter-based method with respect to both prediction accuracy and computation performance.