Optimising the Use of Robust Header Compression Profiles in NEMO Networks

Abstract

Mobile Internet access in moving entities such as, cars, buses, trains, and airplanes is in great demand. The mobility protocols Mobile IPv6 and NEMO, use a bi-directional tunneling mechanism and all communications pass through this tunnel. If this tunnel provides network mobility transparency to the nodes within the network and to their correspondents, it also introduces high protocol overheads since multiple Internet header are carried on each message. This becomes critical since the tunnel is established over different access technologies such as Wireless LAN and 3G where bandwidth is a scarce resource. Moreover, for each mobile terminal (MN) or mobile router (MR), the mobility protocols create a tunnel with the home agent (HA) which can become a bottleneck when it has to serve thousands of mobiles. ROHC header compression mechanism is known to be able to reduce header size and improve performance in low bandwidth links. Thus, ROHC can be used to reduce the tunnel header overhead and improve the performance of the HA. However, ROHC is a complex mechanism and its use needs to be optimised for this specific usage. In this paper, we present two approaches to optimise the use of ROHC profiles when ROHC compression is used in NEMO networks. In the first approach, we suggest to use a reduced number of ROHC profiles. It is proposed to use only the IP profile of ROHC as IP header is the largest header. Also, header compression is applied per user in order to reduce the number of contexts and thus to reduce the processing load and the memory consumption at HA. The second approach proposes to classify the ROHC profiles into different levels depending on the resources that are used at the MR and HA. Optimising the use of ROHC profiles, although reduces the header compression performance, in return it saves resources. Moreover, the complexity of the ROHC mechanism is reduced. We present a performance evaluation of the header overhead reduction for different types of IP flows under the two approaches when they are applied to mobile networks.