Routing and Internet Gateway Selection in Aeronautical Ad Hoc Networks

Abstract

In-flight Internet access for passengers in commercial aircraft is currently typically provided by geostationary satellites. Recently, connectivity in some areas is also provided by means of a cellular network of base stations on the ground. However, satellite based Internet access is relatively expensive, and suffers from large propagation delay due to the extremely large distance of the geostationary satellites. Cellular networks can only be deployed over land areas, but not in oceanic regions, and their deployment on land in remote regions may not be economically feasible. Due to these drawbacks of the existing solutions for in-flight Internet access, ad hoc networks formed by air to air links between the aircraft have recently been proposed as an alternative possibility, especially for oceanic regions with a su�ciently high amount of air tra�ffic, such as the North Atlantic corridor. Aircraft that have a direct connection to the Internet, either through a ground station or a satellite link, may act as Internet Gateways for other aircraft in the network. All data packets that are generated at or destined for one of the aircraft in the network must pass through one of these Internet gateways. At the same time, the gateways' connections to the Internet may exhibit signifficantly different characteristics regarding their average packet delay or their available capacity. Therefore, the allocation of tra�ffic flows to Internet gateways is crucial for the overall network performance. This problem of Internet gateway selection, with the objective of minimizing average packet delay, is the core problem that is addressed in this thesis. The allocation of flows to gateways is closely related to the quality of the path through the network between the aircraft and the gateway. Obviously, this path quality is also closely linked to the availability of wireless channel resources along the path. Therefore, we consider the joint problem of gateway selection, routing, and channel access in aeronautical ad hoc networks. We �first analyze the typical air traffi�c characteristics in the North Atlantic region and show that such a network would indeed be feasible. Then, we formulate the joint gateway selection, routing, and scheduling problem as a mathematical optimization problem with the goal of minimizing the average packet delay in the network. As a less computationally complex alternative, we define a Genetic Algorithm to solve the optimization problem in a heuristic manner. By means of simulations, it is shown that the performance of the Genetic Algorithm approach comes close to the mathematical programming approach in small networks and the algorithm scales well to larger networks. After these centralized approaches to solving the gateway selection and routing problem, the Minimum Downstream Delay algorithm is defined as a distributed approach to the problem and its integration in a protocol within the IPv6 protocol stack is addressed. The performance of this protocol is then simulated in a realistic environment that models the air tra�ffic situation in the North Atlantic and generates realistic data traffi�c mimicking the Internet usage of passengers on board the aircraft.