Distributed Mobility Management for a Flat Architecture in 5G Mobile Networks: Solutions, Analysis and Experimental Validation

Abstract

In the last years, the commercial deployment of data services in mobile networks has been evolving quickly, providing enhanced radio access technologies and more efficient network architectures. Nowadays, mobile users enjoy broadband and ubiquitous wireless access through their portable devices, like smartphones and tablets, exploiting the connectivity offered by the modern 4G network. Nevertheless, the technological evolution keeps moving towards the development of next generation networks, or 5G, aiming at further improving the current system in order to cope with the huge data traffic growth foreseen in the future years.

One of the possible research guidelines aims at innovating the mobile networks architecture by designing a flat system. Indeed, current systems are built upon a centralized and hierarchical structure, where multiple access networks are connected to a central core hosting crucial network functions, e.g., charging, control and maintenance, as well as mobility management, which is the main topic of this thesis. In such a central mobility management system, users' traffic is aggregated at some key nodes in the core, called mobility anchors. Thus, an anchor can easily handle user's mobility by redirecting traffic flows to his/her location, but i) it poses scalability issues, ii) it represents a single point of failure, and iii) the routing path is in general suboptimal. These problems can be overcome moving to a flat architecture, adopting a Distributed Mobility Management (DMM) system, where the centralized anchor is removed.

This thesis develops within the DMM framework, presenting the design, analysis, implementation and experimental validation of several DMM protocols. In this work we describe original protocols for client-based and network-based mobility management, as well as a hybrid solution. We study analytically our solutions to evaluate their signalling cost, the packet delivery cost, and the latency introduced to handle a handover event. Finally, we assess the validity of some of our protocols with experiments run over a network prototype built in our lab implementing such solutions.