The new era of satellite communications is approaching as the next generation of constellations is set to launch during the next few years. Digital payloads, steerable beams, and larger constellations will allow operators to serve better and more demand than ever before. However, better service comes at the cost of operational complexity. Manual distribution of satellite resources becomes unfeasible and automatic tools have to be developed to deal with the additional variables.From all the different sub-problems that constitute the resource allocation problem in satellite communications, this paper focuses on analyzing the beam placement problem in high-dimensional scenarios using graph theory, and, driven by the shown NP-hardness, solving it using a genetic algorithm (GA). To optimize the problem from a system perspective, we develop novel problem-specific metrics (number of beams and number of frequency slots per beam) that intend to reflect a trade-off in the general resource allocation problem.The results show a fast convergence towards the Pareto-Front that allows us to solve a new-era constellation with thousands of users in around 50 generations. The full system-level analysis proves that the problem-specific metrics developed in this work represent a system-level trade-off independently from allocation algorithms chosen for the other satellite resources. On top of that, when we compare our approach with a previously published heuristic, we show that we reduce both power usage, by between 20% and 40%, and unmet demand, by between 50% and 100%.
A Genetic Algorithm for Beam Placement in High-Throughput Satellite Constellations
2021-06-21
1777566 byte
Conference paper
Electronic Resource
English
Optimal Earth Orbiting Satellite Constellations Via a Pareto Genetic Algorithm
British Library Conference Proceedings | 1998
|