Investigation of Performance-Complexity Tradeoff in Filtering LoRa Signals

This paper investigates the performance-complexity tradeoff when implementing pulse shaping and matched filters in LoRa communication systems. Since LoRa gateways are expected to serve a large number of end-devices in various Internet-of-Things (IoT) applications, a high degree of spectral efficiency is an important and desirable feature. Filtering with squareroot raised cosine filters can reduce the occupied bandwidth and adjacent channel interference of transmitted signals, while minimizing the amount of inter-symbol interference (ISI) inherently introduced by practical finite-length filters. The LoRa communication system with filtering was simulated and evaluated in terms of the modulation error ratio, occupied bandwidth, and bit-error rate for signals with various LoRa and filter parameters. The obtained spectra of the filtered signals are compared to those of the unfiltered signals generated by a real LoRa device. The results showed that the occupied bandwidth and the amount of ISI experienced by LoRa signals can be reduced with no degradation in the bit-error rate. Provided that the additional filter is designed and implemented efficiently, filtering is an excellent method to enhance the performance of LoRa devices.