Balancing Computational Efficiency and Detection Accuracy in Oversampled Frequency-Shift Chirp Modulation

Although first proposed a few decades ago, chirp-based modulation has recently seen a surge of popularity as a result of its application in the LoRa standard. Over the past years, this key Internet of Things (IoT) enabler has been well researched, and ever more advanced low-power wide-area networks (LPWANs) are being implemented across the globe, based on this technology. As a result of this international acclaim, multiple actors have invested efforts into implementing frequency-shift chirp modulation (FSCM), which is a more general term for the open-source physical layer modulation protocol also embedded in the LoRa standard, on software-defined radio (SDR) systems. However, while oversampling, advanced post-processing and other digital techniques have led to significant advances in the technology’s capabilities and reliability, real-world deployments of these SDR implementations have to overcome the excessive computational cost associated to these techniques. In response to this challenge, this article examines several new strategies for symbol detection methods operating on FSCM signals, such as those employed in LoRa modulation, enabling significant computational cost reductions. Examples of these are integrating frequency correction in the dechirping procedure and omitting the downsampling operation by using upsampled down-chirps when processing the received samples. In comparison to the standard detection method, computational efficiency gains between 19% and 36% are achieved. Hence, applying the methods presented in this work can yield significant reductions in power consumption for real-world SDR-based FSCM systems in state-of-the-art IoT deployments.