Investigation and analysis of topology in automated HF radio networks

A major issue with providing acceptable network performance has been the fact that most strategies and designs focus on the performance of an individual link. Typically, network designers pay more attention to the needs of the individual radio links than to the effect of those links on the performance of the network. Certainly, the performance of a network is affected (indeed, limited) by the performance of the links which make up the network. However, the relationship between link throughput and network throughput is a complex one, in which each can have a great effect on the other. Traditionally, radio network design has relied on heuristics applied by experienced radio engineers. However, heuristics are often over-generalizations of more complex truths. In this paper, I describe a number of experiments with a large laboratory network. These experiments compare the relative effectiveness of a number of candidate topologies under a number of different load levels. The aim of these experiments is to validate the most common radio network topologies, and determine under what conditions each candidate topology is valid.