Dual optimization with convolutional coding and AI-based noise filtering to enhance LoRa resilience for IIoT

Abstract

In the context of IIoT infrastructures specific to Industry 4.0, LoRa technology stands out for enabling long-range, low-power, and cost-effective wireless communication. However, industrial environments remain challenging due to noise, interference, and fading, which degrade transmission reliability. This work investigates two complementary techniques to enhance the robustness of LoRa communications for two transmission modes: one to many and many to one. First, the integration of convolutional coding into the LoRa frame is examined: A rate 1/2 code yields an SNR gain of 3–10 dB at the cost of a reduced net data rate, whereas a rate 3/4 code maintains that data rate but provides only 1–5 dB of gain. Second, to overcome this trade-off, an adaptive filter based on an artificial neural network (ANN) is implemented at the receiver and coupled with the rate 3/4 code; the filter predicts and subtracts channel noise, raising the overall SNR gain to 11–17 dB (up to + 10 dB under LOS and + 17 dB under NLOS) while preserving the higher throughput of the rate 3/4 scheme. A comparison with conventional methods—Hamming coding combined with FIR or IIR filters—confirms the superiority of our approach: The corresponding gains do not exceed 6 dB in LOS and 9 dB in NLOS for the FIR filter or 5–7 dB for the IIR filter. The hybrid scheme combining rate 3/4 convolutional coding with ANN-based filtering offers the best trade-off between throughput and resilience, paving the way for reliable LoRa deployments in the most demanding industrial environments.