Communicating with a satellite is like trying to hear a whisper in a storm: the signal drowns in noise. Engineers have created an AI algorithm that runs directly on the satellite and, with minimal power consumption, detects distortions more accurately than previous methods.
The secret to its compactness is physics: the algorithm knows how the satellite's motion (Doppler effect) shifts the signal's frequency. It’s like a listener who knows the siren’s sound changes as a car passes, and corrects for it in advance. This approach requires 29 times less computing than conventional methods and 27% fewer parameters compared to other AIs, with no loss in accuracy. Amazingly, it uses less energy than a lightbulb in your fridge, while catching a wave shift comparable to the trembling of a spider's thread in the wind.
Low-orbit satellites with such micro-AI will provide connectivity in dead zones — from mountains to the open ocean. It’s a step toward 6G, where space will weave together with ground towers into a single network.
Modern low-orbit satellites travel at about 7.8 km/s, which can shift the signal frequency by tens of kilohertz — imagine an ambulance siren changing pitch on the run, but in radio waves.
The topic echoes Arthur C. Clarke's 'The City and the Stars,' where perfect communication linked distant worlds.