DeepFIR: Channel-Robust Physical-Layer Deep Learning Through Adaptive Waveform Filtering

Deep learning can be used to classify waveform characteristics (e.g., modulation) with accuracy levels that are hardly attainable with traditional techniques. Recent research has demonstrated that one of the most crucial challenges in wireless deep learning is to counteract the channel action, which may significantly alter the waveform features. The problem is further exacerbated ...