Distributed IP mobility management for hosts and networks
Doctoral Thesis
2014
Permanent link to this Item
Authors
Journal Title
Link to Journal
Journal ISSN
Volume Title
Publisher
Publisher
University of Cape Town
Department
License
Series
Abstract
The Internet was originally designed for stationary nodes. With the advancement of mobile nodes (such as smartphones and tablets) that have wireless Internet access capability, the original design of the Internet is no longer sufficient. These mobile nodes are capable of communicating while moving and changing their point of attachment in the Internet. To maintain communication session(s) continuity for these mobile nodes, the Internet needs mobility management mechanisms. The main mobility management protocols standardised by the Internet Engineering Task Force (IETF) are mobile IP (MIPv6 and MIPv4) and their numerous extensions and variants, including proxy MIP (PMIPv6 and PMIPv4). The architectural structures of these protocols employ a centralized mobility anchor to manage the mobility of the mobile nodes in the control and data planes. The mobility anchor manages the mobility binding information and the forwarding of data packets for all mobile nodes registered in the network. However, in the context of the rapid growth in the number of mobile users and the data traffic volume, as well as the trend towards a flat architecture in mobile networks, the centralized mobility management approach provides insufficient mobility support to the mobile nodes. For example, to manage the demand for increased mobile users, a huge amount of data traffic will be pushed to the centralized mobility anchor. Yet, routing huge volumes of traffic via the centralized mobility anchor can be non-optimal in terms of routing efficiency. Thus, the centralised mobility anchor can be a potential bottleneck, and a single point of failure. Consequently, failure of the mobility anchor may lead to a service outage for a large number of mobile nodes. Ultimately, the centralized mobility management approach does not scale well with the increase in number of mobile users and the data traffic volume. These problems are also costly to resolve within the centralized mobility management approach and its related centralized network architecture. Distributed mobility management (DMM) is one recent approach that can efficiently address the shortcomings of centralized mobility management. It provides an alternative paradigm for developing IP mobility management – without employing centralized mobility anchors. In this paradigm, either the mobility anchors, or their mobility management functions, are distributed to different networks/elements. The mobility anchors, or the mobility management functions, are brought to the edge of the networks, which is closer to the mobile nodes. Distributed mobility management also offers dynamic mobility features that allow a mobile node to anchor traffic at different mobility anchors. However, to date, mobility management schemes that have been developed based on the DMM approach are still in the preliminary stages, and there is no current standard in place. These developed DMM schemes are still experiencing problems, such as long routing paths, especially for long-lasting data traffic, a lack of route optimization for ongoing communication, and a lack of synchronization of the mobile nodes‟ location in different networks. Moreover, the majority of these proposed schemes still need to be analysed, in order to quantify their feasibility. The thesis proposes three novel network-based distributed mobility management schemes, which are based on the DMM approach. The schemes enhance PMIPv6 to work in a distributed manner, in order to address the problems of centralized mobility management. Furthermore, the schemes address the following issues: (1) the lack of route optimization for ongoing communication; (2) the lack of synchronization of the mobile nodes‟ location in different networks; and (3) the long end-to-end packet delivery delay problems in recently proposed DMM schemes. The first scheme, called the network-based distributed mobility management scheme with routing management function at the gateways (DM-RMG), decomposes the logical mobility management functions of the Local Mobility Anchor (LMA) in PMIPv6 into internetwork location management (LM), routing management (RM), and home network prefix allocation (HNP) functions. After the decomposition, the RM function is collocated at the gateways of different networks. In this way, the data-plane routing function of the respective mobile nodes is served by the corresponding local RM function at the network gateway. The DM-RMG scheme offers distributed mobility management for individual mobile nodes (i.e., mobile hosts) during mobility events. DM-RMG also implements a mechanism to optimize the handover delay. The results obtained from analytical modelling and simulation show that the DM-RMG scheme outperforms the centralized mobility management schemes, as well as currently proposed distributed mobility management schemes in terms of the end-to-end packet delivery delay under different network load conditions. The optimized handover performance of the DM-RMG scheme, investigated under different traffic patterns and mobile node speeds, shows that the scheme also mitigates the internetwork handover delay and packet loss. The second proposed scheme, called network-based distributed mobility management for the network mobility (NDM-RMG), uses a similar approach to DM-RMG. However, it proposes a network-based DMM scheme for Network Mobility (NEMO). The main goal of the NDMRMG scheme is to address the problems of centralized mobility management protocols for NEMO, including the pinball routing problem in nested NEMO. NDM-RMG is compared with centralized mobility management schemes for NEMO, and recently proposed distributed IP mobility management schemes for NEMO by means of analytical modelling and simulation evaluations. NDM-RMG shows better performance in terms of reducing the packet delivery latency, the size of the packet header, and the packet overhead experienced over the wireless link. The third proposed scheme, called network-based distributed mobility management scheme with RM and HNP allocation functions distributed to the access routers (DM-RMA), distributes the RM and the HNP allocation functions at the access routers with the mobility client function. This brings the mobility-related functions closer to the mobile nodes, that is, to the edge of the network. An analytical model is developed to investigate the mobility cost performance of the scheme, due to signalling, packet delivery, and tunnelling. The analytical results indicate that DM-RMA performs better than the previous DMM schemes in terms of packet delivery, tunnelling and total costs. Network simulator-2 (ns-2) is used to model the DM-RMA scheme. The simulated scenarios confirm that DM-RMA performs better than other proposed DMM schemes in terms of reducing the location update latency at the location managers, end-to-end packet delivery delay, handover delay, and packet loss. In addition to the three proposed DMM schemes, this thesis proposes a routing optimization scheme for PMIPv6. The main goal of this scheme is to enable PMIPv6 to offer route optimization to mobile nodes in a PMIPv6 domain. The scheme reduces the route optimization-establishment latency, the packet delivery latency, and the packet loss. Using ns-2 simulations and considering different simulated scenarios, the results show that the scheme reduces route optimization-establishment latency and delayed packets during the route optimization operation, as compared to previously proposed PMIPv6 route optimization schemes. The results also show that the scheme reduces packet loss when a mobile node undergoes handover in the PMIPv6 domain.
Description
Includes bibliographical references.
Reference:
Ernest, P. 2014. Distributed IP mobility management for hosts and networks. University of Cape Town.