Low Complexity LoRa Frame Synchronization for Ultra-Low Power Software-Defined Radios

Abstract

Low power wide area (LPWA) wireless networks based on the LoRa physical layer have attracted huge attention in recent years, both from industry and from academic researchers. While this rising popularity is due to this technology's demonstrated effectiveness and low cost, unfortunately, due to their complexity, the timing and frequency synchronization algorithms required to detect LoRa-modulated frames, in the context of minimum sampling rate optimum receivers, have received little attention. The aim of this paper is to fill this gap and describe how robust frame detection can be performed while focusing on minimal complexity implementations of the proposed algorithms. The ultimate goal is to propose frame detection techniques applicable to recently proposed ultra-low power software-defined receivers.

LoRa 2.4 GHz Communication Link and Range

“…The forwarding of an uplink message by multiple gateways ensures that a handover can take place without any loss of data. In another study, a robust frame detection algorithm was proposed in order to detect LoRa-modulated frames with minimal complexity implementations [ 9 ]. Furthermore, a frame relay strategy was found to be a feasible way to improve the link quality of poorly connected nodes and successfully extend the range of LoRaWAN [ 10 ].…”

“…Vangelista [ 22 ], for instance, has provided a mathematical model, called Frequency Shift Chirp Modulation (FSCM), that describes the LoRa modulation process. The same model has been adopted by Bernier et al [ 9 ]. On the other hand, Knight [ 21 ], Robyns et al [ 23 ] and Ghanaatian et al [ 24 ] have provided a model called Chirp Spread Spectrum (CSS) modulation based on the reverse engineering of LoRa’s physical layer.…”

On Time-Frequency Synchronization in LoRa System: From Analysis to Near-Optimal Algorithm

“…It must be emphasized that, in the range of indexes p close to p 1 and p 2 , the overlap of one function (say A 1 asinc γ1 ) over the other one (A 2 asinc γ2 ) can be neglected. Then we can assume the approximation in (14), which is much simpler than the expression in (8) as the phase terms can be omitted. It remains to analyze the behavior of the last term involving the sum in (14), that we denote by Q for a matter of clarity.…”

“…Then we can assume the approximation in (14), which is much simpler than the expression in (8) as the phase terms can be omitted. It remains to analyze the behavior of the last term involving the sum in (14), that we denote by Q for a matter of clarity. For a given value of (τ, θ f ), the function Q only depends on k, since p also depends on k from (12) and (13).…”

“…For this reason, it has been reused in most of the methods in the literature dealing with synchronization in LoRa [12]- [15]. In [12], [14], the authors suggest to sequentially estimate the integer and the fractional contributions of the time and frequency offsets, based on the whole LoRa preamble, whereas it is proposed in [15] to track the oscillator fluctuations by means of the demodulated symbols within a LoRa frame. To achieve a lowcomplexity algorithm, the authors in [14] suggest to process the received signal at Nyquist rate while [12] investigates the effect of the oversampling rate.…”