pyLoRaPHY

Full implementation of the physical layer of LoRa. (Python translation of jkadbear/LoRaPHY)

It also contains comments from my understanding of the code and unique implementation choices optimized for clarity, Pythonic style, and performance.

Some MATLAB values distingush column and row vectors. They are all treated as 1D arrays here. All indices are C-style.

The original code was (c) 2020-2022 jkadbear licensed under MIT.

As a project developed during Maiyun's participation in the UCSD SRIP program, The University of California may have claims to this work.

@author: Zhang Maiyun maz005@ucsd.edu Created on Thu May 23 17:28:37 2024

I try to not use self in functions that just calculate values and return them, to allow for easier testing. Also, while the original implementation passes data with class properties, this one uses parameters to pass ones that are packet-specific and only store stable configuration as instance variables.

Implementation checklist:

• Constructor tested
• init tested
• dechirp tested
• detect tested
• demodulate tested
• parse_header tested
• dynamic_compensation tested
• gray_coding tested
• diag_deinterleave tested
• hamming_decode tested
• sync
• decode tested
• dewhiten tested
• calc_crc tested

Functions for modulating side stuff:

• modulate tested
• encode tested
• gray_decoding tested
• calc_sym_num tested
• whiten: (as dewhiten)
• gen_header tested
• hamming_encode tested
• diag_interleave tested

These three look like they are not referenced anywhere:

• symbols_to_bytes
• calc_payload_len
• time_on_air tested

Other functions:

• print_bin (debugging, use logging instead)
• print_hex (debugging, use logging instead)
• log (debugging, use logging instead)
• plot_peak
• bit_reduce (as xorbits)
• word_reduce merged into hamming_encode
• topn merged into dechirp
• chirp tested
• spec: use matplotlib.pyplot.specgram instead
• read: (as demodulate_file)
• write: use ndarray.tofile instead