Context-aware handover management for HetNets: Performance evaluation models and comparative assessment of alternative context acquisition strategies

The management of Vertical Handovers (VHO) in a heterogeneous networks (HetNet) environment has received considerable attention, with the aim of enabling effective, context-aware network selections. Various relevant architectures and frameworks have been proposed, but the available methodologies for evaluating the performance of these proposals and for comparing alternatives are still limited. In addressing this issue, the paper contributes a versatile modeling methodology that focuses on signaling in the handover preparation phase and incorporates all significant effects that are associated with the exchange, queueing and processing of the signaling messages and have an impact on delay-related performance. The resulting model is comprehensive, yet capable of admitting closed form solutions. Even more importantly, the methodology is generic and can be flexibly tailored to different VHO architectures. To demonstrate this feature, the paper applies the modeling methodology in two context-aware VHO approaches that differ in the way of acquiring dynamic resource availability context. The first approach is according to standards-based recommendations (such as the IEEE 802.21 Media Independent Handover Function (MIHF) and the 3GPP Access Network Discovery and Selection Function (ANDSF)) and obtains resource availability context reactively, each time a handover is triggered, through interaction between the sewing network (SN) and every candidate network (CN). The other approach aligns with recent research directions and introduces a local context repository that gathers proactively, in a periodical fashion, the resource availability context from the various Radio Access Networks (RANs) associated with this repository, reducing the signaling overhead and enabling a potentially prompter discovery of a proper network target. For both approaches, the model-based results are validated against simulations, confirming the effectiveness and the accuracy of the modeling methodology. Furthermore, the model-based results enable a direct performance comparison of the two approaches, demonstrating that the proactive architectural approach can provide significant performance and processing efficiency gains, in comparison to the reactive approach.