Free space optical communications routing performance in highly dynamic airspace environments

Airborne networking environments utilizing high speed free space optical communications are being proposed as a method of providing immediate high speed network access and communications in a tactical environment where access is otherwise unavailable. Frequent link failures and recoveries due to wing obstruction and the dead zone produced by the wake vortex of the aircraft characterize the links. Many links additionally have large delay due both to coding/interleaving and propagation characteristics. Network availability from a user perspective is predicated on the fast response to link outages and recoveries by the routing protocol. The routing protocol must converge very quickly to ensure that the network is highly available and that traffic is not disrupted. In this work, we considered two different architectures/topologies that meet these requirements and analyzed the performance of OSPF and EIGRP on each using simulation. We fixed the scenarios such that an end-to-end path was always available through the network. We varied the input parameters for OSPF and EIGRP to maximize network availability. We determined that OSPF is limited by its minimum HELLO interval and results in gaps in path availability on the order of seconds and EIGRP may respond quickly to achieve subsecond gaps in path availability, but at a cost of increased network traffic. We have additionally introduced an analytical methodology for selection of HELLO interval lengths and router dead times which quantifies the cost and benefit of a particular selection. Scalability of the different approaches to a complete network needs further exploration.