Performing Gateway Load Balancing in MANETs

Abstract

During the last decades, the advances in Information Technology have formed the basis for increased interest and research activity in the field of ad hoc wireless multihop networks or simply ad hoc networks. This emerging technology enables internetworking between wireless nodes that are deployed in an ad hoc and temporary manner. All nodes in an ad hoc network take the role as both hosts in an end-to-end communication session, or as routers to collaboratively relay data traffic in a multihop fashion on behalf of other nodes. Furthermore, ad hoc networks are highly dynamic in nature, i.e. nodes can join or leave the network at any time, and additionally, the nodes have also the flexibility to move around while being in the network. The fact that ad hoc networks can be rapidly deployed with minimal prior planning, cost, and without the need of any pre-existing infrastructure makes this technology very attractive and suitable in a number of applications, including emergency and rescue operations, and military operations.

Although ad hoc networks represent a promising technology that offers a broad range of potential useful applications, this technology is still in an immature phase. There are yet many issues and challenges that need to be resolved, which mainly arise from the inherent unreliability of wireless communication, the dynamic nature of these networks, the limited availability in resources with respect to bandwidth, processing capacity, battery power, and from the possibly large scale of these networks. These challenges require that the networking protocols at all layers in the network stack, that in many cases were originally designed for wired networks, must be modified or optimized, in order to adapt to the characteristic of the wireless environment.

The focus of this thesis has been devoted to the investigation of two specific issues within the field of ad hoc networking, i.e. node mobility and load balancing. The aim is to provide solutions in order to improve the overall performance in ad hoc networks.

Node mobility is one of the most important features in ad hoc networks, however, it is also the reason for frequent link breaks and the constant change in the topology. An ongoing data transmission that is interrupted by a link break, must be rerouted to alternative paths in order to circumvent the broken link. However, this process of rerouting traffic takes a certain amount of time, which is referred to as the rerouting time. Minimizing the rerouting time is essential in order to reduce packet loss and improve network performance. In this thesis we investigate the factors that affect the rerouting time in proactive routing protocols and propose solutions for minimizing it.

Load balancing refers to the process of distributing traffic load more evenly in the network in order to minimize congestion and to optimize the usage of network resources. Performing load balancing in ad hoc networks is generally very challenging due to the inherently interfering nature of the wireless medium. In this thesis we therefore investigate the feasibility and the potential benefits of performing load balancing in ad hoc networks. We consider two scenarios, i.e. load balancing for intradomain and interdomain traffic.

Intradomain traffic is traffic between nodes inside an ad hoc network. Performing load balancing on intradomain traffic can be done in two ways. The first is referred to as multipath load balancing where a traffic flow between a source and destination pair is distributed over multiple alternative disjoint/semi-disjoint paths. The aim is to maximize throughput and reduce the risk for packet loss. However, a number of previous work has investigated and reported that this type of load balancing can only provide a rather limited improvement in performance due to the interference between the paths [1] [2]. Due to this reason, multipath load balancing is therefore not considered in this thesis. Instead we focus on the second way which is referred to as transit routing. Transit routing is about routing part of the local traffic over a backbone network in order to relief the traffic load in the ad hoc network. The assumption behind this concept is a network architecture similar to a Wireless Mesh Network (WMN), where a high capacity backbone network is an integrated part of the Mobile Ad hoc Network (MANET). This backbone network is commonly used to provide Internet-connectivity services, but can also be exploited to alleviate the traffic load in the MANET. In addition, for certain source and destination pairs, performing transit routing can considerably increase the throughput compared to if the traffic is routed within the ad hoc network.

Interdomain traffic refers to traffic between a node inside the ad hoc network and a remote node outside of the ad hoc network. Load balancing for interdomain traffic considers the potential of distributing interdomain traffic among multiple gateways in order to avoid congestion at the gateways and maximize the capacity for interdomain traffic. This type of load balancing is commonly referred to as gateway load balancing in the literature. Furthermore, interdomain can either be inbound or outbound traffic. The work in this thesis mainly focuses on performing load balancing for outbound traffic. However, we believe that the results in our work are also applicable to inbound traffic as well.

The main contributions in this thesis are the investigation and the proposals of different solutions for intradomain and interdomain load balancing.