Onboard AI-Driven Consensus for Distributed Spectrum Sensing in CubeSat Networks
Type
Supervisor
Date
2026-03-12
Collections
Research Projects
Organizational Units
Journal Issue
Abstract
Inter-satellite communication in CubeSat networks
is inherently dynamic and delay-prone, posing serious challenges
for reliable spectrum sensing in cognitive radio (CR) systems.
Conventional distributed consensus methods, which typically rely
on fixed weight assignments, struggle to cope with the fluctuating
link quality and node reliability found in space-based environments.
This paper introduces an AI-driven adaptive consensus
approach designed specifically for cooperative spectrum sensing
in CubeSat-enabled CRNs. Each CubeSat runs a neural network
model trained offline and deployed onboard, allowing it to
autonomously adjust consensus weights based on real-time local
metrics such as signal-to-noise ratio (SNR), packet loss rate, and
sensing confidence.
The proposed method is evaluated through simulations
and hardware-in-the-loop testing using software-defined radios
(SDRs). Compared to static-weight schemes, it demonstrates up
to a 15% improvement in detection accuracy, a 30% reduction in
convergence time, a 20% reduction in communication overhead,
and robust performance under up to 20% packet loss. By
embedding adaptive intelligence directly into the consensus process,
this work addresses a critical gap in CubeSat-based CRNs
and provides a scalable framework for autonomous spectrum
management in future space communication systems.
Department
Publisher
Sponsor
IEEE