Pushing Even Lower-SNR Neural-Demodulation of LoRa Communication

• Replicate the NELoRa paper on the dechirp model with Matlab and neural enhanced decoding model with Pytorch.
• We implement parameter pruning to compress the NELoRa model by 60% and lower the SNR threshold by 0.4dB.
• ML System (Model Compression, LoRa, PyTorch)

• Report.pdf
• Modified NELoRa Code