Mobility management in the future mobile network
Morteza Karimzadeh is a PhD Student in the research group Design and Analysis of Communication Systems. His supervisors are professor Hans van den Berg and professor Aiko Pras from the faculty of Electrical Engineering, Mathematics and Computer Science.
In the last years, with the worldwide deployment of mobile networks and the commercial success of smart portable devices e.g., smart-phones and tablets, the use of mobile data services and data traffic has exploded. This growth is foreseen to remain in future years, due to the continuous increase in popularity of mobile services and novel mobile applications, e.g., Internet of Things (IoT) and Machine to Machine (M2M). Mobile network operators are therefore already planning their strategies with the objective of enhancing the scalability of the overall system to cope with the future traffic growth. The current mobile network architectures are heavily hierarchical and centralized, which implies that all traffic must be traversed through a centralized core entity. This makes the network prone to several limitations, e.g., suboptimal communication paths, low scalability, signaling overhead, and single point of failure. The hierarchy of the data forwarding procedure in current network architectures can be eliminated by letting IP flows be routed in a more flexible way. One of the key enablers for this is to shift towards fatter (decentralized) network architectures. This approach allows to place distributed anchor points in the proximity of the radio access network to locally handle Mobile Nodes' (MNs') connections and data traffic, and accordingly reduce the load of traffic in the core network.
Long-Term Evolution (LTE) is expected to be the leading mobile networking technology in the coming decade. It is estimated that LTE network will handle about 79% of the worldwide mobile data traffic by 2021. Therefore, we first review the current (centralized) LTE network system, and next discuss an approach to realize deployment of an LTE network with decentralized architecture. Although, decentralization of the LTE network architecture would be, on the long term, a sustainable and cost-effective approach to deal with the future data traffic demands, its gain on the network performance and efficiency is an open question. Hence, we carry out a hybrid study comprising simulation and analytical models to compare the efficiency of the current and decentralized LTE network architectures. Particularly, our analysis for both network architectures presents numerical results, quantifying the impact of the number of attached MNs on the load at network routers and links, on the latency, and on the processing cost of the user's data and control planes.
When dealing with the transition from the centralized and hierarchical LTE network architecture to a decentralized one, mobility management is one of the key functions that needs to be adapted to the new architecture. In the current LTE system, the mobility management function follows a centralized scheme, in which the Packet Data Network Gateway (PGW) aggregates the traffic flows to and from the access networks, and redirects traffic flows to MNs' new locations, for those moving from one access network to another (i.e., from one Serving Gateway (SGW) to another). However, in an LTE network with decentralized architecture, the PGW and SGW functions are co-located into the single entities (i.e., S/PGWs) closer to the radio access network, that handle the MNs' connection functions, data traffic and mobility locally. This, demands additional mechanisms to maintain the MN's ongoing data session active, when it performs a handover with S/PGW relocation.
To address this, we develop three novel mechanisms. Our first solution exploits the Network Address Translation (NAT) technique, which is widely used in the current Internet. Therefore, its implementation is feasible with limited overhead and complexity. In the second solution, we take the advantage of the Software Defined Network (SDN) paradigm on detaching the control plane from data plane, to manage MNs' mobility in a decentralized LTE network. SDN is one of the core technologies in the transformation of current mobile networks to next generation networks (i.e., 5G). It is an essential enabler for many anticipated features and functionalities of 5G networks, such as elasticity, scalability, adaptability, slice-ability and program-ability. Knowing this, makes the proposed solution feasible and ideally suited for the next generation of mobile networks. The NAT-based and SDN-based solutions conduct the MNs' mobility management in the network layer. This accordingly, hides any changes from the upper layers and keeps the end points unaware from the related functions. However, these approaches demand to introduce a few new components in the network and some infrastructural modifications. In the third solution, we benefit from the Multipath-TCP (MPTCP) protocol feature on supporting multiple IP addresses for a single data-stream and its break-before-make option to handle MNs' mobility. This solution keeps the network infrastructure intact and implements the whole mobility management functionality in the transport layer of only the end host entities. MPTCP is already deployed on smart-devices working with iOS (since version 7) or Android (since version 4.4) mobile operating systems. This makes the proposed approach readily feasible to implement at large scale with minimal overhead and complexity. Finally, we make a comparison of the proposed mobility management solutions and discuss their pros and cons.