Comparing Space Versus Surface Gateways for Internetworking pLEO Satellite Networks

Multiple governments and companies are deploying massive scale "proliferated Low Earth Orbit" (pLEO) satellite networks consisting of hundreds to thousands of satellites and promising low-latency global connectivity. However, no architecture currently exists for internetworking independently managed pLEO networks. We construct a model of two representative pLEO networks and compare performance metrics under two architectures: interconnecting via surface gateways versus space-based gateway satellites. We also formulate and evaluate two algorithms for establishing links between gateways and pLEO networks: (1) a latency-optimal algorithm, and (2) an algorithm that reduces link switching (reducing handovers and outages). By evaluating scenarios of 1 to 50 gateways, we find that surface gateways enable up to 8.4 ms lower mean inter-network latency than space gateways, but space gateways may provide up to 7× more inter-network capacity and up to 6.4 × reduction in link switch rate. Furthermore, our link selection algorithm for reducing link switching in space gateways is shown to achieve a 3× reduction in link switch rate at the cost of a 6.9 ms increase in mean user-to-user latency. We discuss the tradeoffs between space and surface gateways and identify future research needs to enable interconnection of independent pLEO networks.