The data-driven radio frequency signal separation challenge
Published in 2024 IEEE International Conference on Acoustics, Speech, and Signal Processing Workshops (ICASSPW), 2024
[arXiv]
The radio-frequency (RF) signal separation challenge involves recovering a signal-of-interest (SOI) from a super-imposed co-channel interference signal. The SOI is a digital communication waveform of known modulation, pulse-shape, timing, etc. The interferer is unknown and must be learned from data. Submissions featured a blend of signal processing strategies, leveraging RF-specific domain knowledge and novel neural network architectures with careful hyperparameter selection/optimization. The resulting solutions establish new benchmarks for data-driven RF modeling and interference cancellation.
Recommended citation: T. Jayashankar, B. Kurien, A. Lancho, G. C. Lee, Y. Polyanskiy, A. Weiss, and G. W. Wornell, “The data-driven radio frequency signal separation challenge,” in 2024 IEEE International Conference on Acoustics, Speech, and Signal Processing Workshops (ICASSPW). IEEE, 2024, pp. 53–54.