Context-Sensitive and Adaptive Routing in Wireless Mobile Ad-Hoc Networks Using Cross-Layer Design

Keywords: routing protocols, wireless ad hoc networks, mobile computing, cross-layer design, MANETs, network emulation

Wireless ad hoc networks consist of an ensemble of mobile nodes operating in a particular area without using any existing infrastructure. Each mobile node acts as an intermediate router forwarding messages received by other nodes. A wireless ad hoc network is only operational if nodes offer their forwarding capabilities to other nodes. The special conditions of wireless communication, such as limited transmission range, limited bandwidth and possible interference, require dedicated routing algorithms that take these conditions into account.

Over the last few years, many routing protocols were developed to deal with the specific challenges of mobile ad hoc networks. However, most of these approaches interpreted and implemented the network as a hierarchy of layers that are independent and non-cooperating. This results in a less than optimal performance of ad hoc wireless networks in general, especially when energy is a constraint or the application has high bandwidth needs and/or stringent delay constraints. To achieve substantial gains in throughput, efficiency and quality of service, a cross-layer protocol design that supports adaptivity and optimization across layers of the protocol is required. In an adaptive cross-layer protocol stack, the routing algorithm can adapt links and routes to meet the requirements of the application given current channel and network conditions. The offered routes can be adapted based on underlying link and interference conditions as well as delay constraints and application requirements. Adaptive routing protocols can be developed based on current link, network, and traffic conditions.

The aim of this project as part of the DFG priority program (SPP) 1140 is the development and evaluation of a scalable, context-specific routing framework for self-organizing mobile networks using an cross-layer approach. Currently, cross-layer interactions between application and routing layer and between link and routing layer are being investigated and integrated into the current architecture of the routing framework shown in figure 1, which intends to provide the desired adaptive, context-aware routing functions.

Figure 1: Current architecture of the routing framework

To analyze user applications and routing modules for MANETs it is vital to have a suitable emulation testbed. Therefore, we are developing MarNET, an emulation system consisting of a distributed topology enforcement software and the paravirtualization solution Xen. The overall architecture is shown in Figure 2.

Figure 2: MarNET Architecture

Using Linux software bridges it is feasible to expand the control- and emulation network using multiple physical hosts. Paravirtualization offers the possibility to operate multiple virtual hosts on a physical machine leading to a high number of nodes and efficient resource usage. To introduce communication parameters like packet loss and message delay, NetShaper is used, a linux device driver developed by the University of Stuttgart. MarNET Mobility And Topology Manager is responsible for the calculation and distribution of these parameters.

There are several open questions that will be tackled in the future. For example, the necessary requirements of how and where topology information should be provided have to be defined in conjunction with other application-oriented projects of the DFG priority program (SPP) 1140. The functional modules for additional services like a distributed name service or the availability of services have to be developed, validated with our network emulator and have to be implemented in the routing framework. Furthermore, the aim of the proposed routing framework is to offer information and application interaction to minimize resource consumption. However, the necessary modules themselves require additional resources (bandwidth and energy) in order to gather and distribute this information. Therefore, the possible resource gains of the routing framework have to outweigh the additional consumption in a significant manner, which still requires a detailed analysis of the gain and cost of routing services.

Funding: Deutsche Forschungsgemeinschaft (DFG, SPP 1140), since 2002