Enabling Reliable LoRa Decoding under Cross-channel Interference

This paper focuses on cross-channel interference caused by collided chirps with different bandwidths but same slope in time-frequency domain. Existing concurrency decoding methods cannot resolve it because they require the accurate time duration of all chirps. But in limited receiving bandwidth, interfering chirps with asymmetric bandwidths are incomplete. Their time duration is unpredictable due to the randomly modulated symbols. We instead propose Elora to decode target packets under cross-channel interference. Elora leverages the difference in chirps' time duration instead of absolute time duration to identify target chirp because only target chirp fills the decoding window in time and frequency. Elora adopts chirp elimination to test whether each received chirp can be completely eliminated after subtracting the reference chirp that fully occupies the decoding window because only the target chirp is completely eliminated while interfering chirps remain unaffected. However, in practice, unexpected frequency shift and phase rotation prevent the target chirp from being completely eliminated. We propose a linear-fitting method to correct frequency shift and then compensate phase rotation by an energy varying model. We also adopt window dividing to estimate energy intensity of chirp's residue. The experiments show that Elora reduces the symbol error rate by up to 86.3% compared with existing methods.