Scheduling vs. random access in frequency hopped airborne networks

Airborne networks often use frequency hopping to be jam resistant. Because users are highly mobile and can have large propagation delays, otherwise orthogonal frequency hopping patterns appear asynchronous, and hop-by-hop collisions will occur. We compare the achievable throughput and delay of scheduling schemes versus random access schemes in this context via modeling and simulation. Because collisions occur even when nodes are scheduled at the slot level, much less throughput is attainable, as compared to the case of orthogonal hopping patterns. Random access achieves even less throughput because it cannot control exactly how many users are active in a slot, only the average. This results in scheduling offering roughly a 10% to 20% gain in throughput over random access, depending on parameters such as code block length, the number of frequencies hopped over, and the number of users. However, a static schedule, also known as Time Division Multiple Access or TDMA, is shown to have very large delays for high traffic loads. Dynamic scheduling can achieve the throughput of TDMA and the delay of random access, but it incurs additional overhead for coordination between users which may outweigh the throughput gain. Alternatively, random access with time hopping has the potential of achieving the throughput of TDMA, but at some cost of delay.