Advancing satellite network consensus through optimal orbital configurations

As the number of Earth-orbiting satellites continues to grow, ensuring consensus among them becomes increasingly crucial for autonomous, unbiased decision-making. With greater autonomy moving onboard satellites, the ability for them to make collective decisions on critical tasks such as space environment monitoring or disaster detection becomes essential. Consensus algorithms are key in enabling this cooperation, but fault-tolerant algorithms, like Practical Byzantine Fault Tolerance (PBFT), require significant communication between network members, leading to frequent satellite interactions. While some satellite constellations have improved communication within their own networks, inter-constellation and individual satellite communication remain limited. To support universal inter-satellite communication, close proximity between satellites is vital. This makes optimal orbital strategies essential for maximizing interactions and facilitating seamless communication across constellations. This paper explores improving consensus protocol efficiency by identifying optimal orbit strategies for PBFT. By simulating both theoretical and existing satellites, we compute the time required to reach consensus for various network sizes. This analysis helps pinpoint optimal orbital configurations that minimize consensus time. Additionally, simulated satellites are arranged into constellations - similar to space relays - further reducing consensus time. However, each simulated satellite acts as a single node in the network, avoiding centralization and maintaining the decentralized nature of the system. Real and simulated satellite orbits are propagated, allowing us to analyze potential interactions, distances, and timings between satellites. These interactions are then evaluated through multi-objective optimization, minimizing consensus time while maximizing the diversity of satellites involved. Pareto fronts from this optimization provide insights into the most efficient simulated orbits for consensus. A genetic algorithm is also employed to optimize the Keplerian elements of these orbits for further refinement. The study finds that simulated satellite orbits resembling space relays significantly reduce consensus time, although they require a large number of satellites to establish such a network. Alternatively, to optimize satellite interactions in terms of both quantity and diversity, orbits with a mean motion deviating from the norm (approximately 15 revolutions per day) are needed. Additionally, orbital inclination and eccentricity are shown to play a major role in enhancing satellite interactions. This research sets the theoretical basis for developing future decentralized orbital networks, designed for trustless decision-making in orbit, setting the stage for future autonomous on-board satellite operations.