[Paper Review] Decoding Superposed LoRa Signals

Nancy El Rachkidy, Alexre Guitton+1 more|arXiv (Cornell University)|Mar 22, 2018

IoT Networks and ProtocolsEngineering7 references20 citations

TL;DR

This paper proposes two novel algorithms to decode superposed LoRa signals in LPWANs: one for slightly desynchronized transmitters using timing-based symbol matching, and another for synchronized transmitters requiring one retransmission. The desynchronized algorithm achieves up to 100% throughput gain for two signals and 18–99% gain for three, while the synchronized algorithm enables up to 25% higher throughput by decoding one frame per two-colliding-frame collision with retransmission.

ABSTRACT

Long-range low-power wireless communications, such as LoRa, are used in many IoT and environmental monitoring applications. They typically increase the communication range to several kilometers, at the cost of reducing the bitrate to a few bits per seconds. Collisions further reduce the performance of these communications. In this paper, we propose two algorithms to decode colliding signals: one algorithm requires the transmitters to be slightly desynchronized, and the other requires the transmitters to be synchronized. To do so, we use the timing information to match the correct symbols to the correct transmitters. We show that our algorithms are able to significantly improve the overall throughput of LoRa.

Motivation & Objective

• To address the critical performance degradation in LoRa networks caused by signal collisions due to low duty-cycle and low bitrate.
• To enable reliable decoding of colliding LoRa signals without requiring changes to the LoRa physical layer.
• To design practical, low-complexity algorithms that exploit timing and synchronization characteristics of LoRa signals for interference cancellation.
• To evaluate the throughput gains of the proposed algorithms under realistic network conditions with multiple end-devices.
• To demonstrate that collision resolution is feasible in LoRa by leveraging its unique Chirp-Spread Spectrum modulation properties.

Proposed method

• For slightly desynchronized signals, the algorithm uses timing information to match received symbols to their respective transmitters by exploiting phase and symbol boundary alignment.
• The method performs symbol-by-symbol correlation across time windows to identify and decode individual frames from superposed signals.
• For synchronized signals, the algorithm requires one transmitter to retransmit its frame, enabling the gateway to decode the other frame using known signal structure.
• The approach relies on the fact that LoRa signals with the same SF and BW on the same channel are typically undecodable in standard LoRa, but can be resolved using timing and signal structure analysis.
• The algorithms are evaluated using simulations with varying spreading factors (SF), duty cycles, and numbers of colliding signals (n=2, n=3).
• Throughput is computed by comparing the number of successfully decoded frames using the proposed method versus standard LoRa (which drops to zero for n≥2).

Experimental results

Research questions

• RQ1 Can superposed LoRa signals be decoded when transmitters are slightly desynchronized, and if so, under what conditions?
• RQ2 What is the maximum number of signals that can be decoded using timing-based symbol matching in the desynchronized case?
• RQ3 Can synchronized LoRa signals be resolved with minimal overhead, and what is the required retransmission mechanism?
• RQ4 How do the proposed algorithms compare to standard LoRa in terms of throughput gain under realistic duty-cycle constraints?
• RQ5 What is the impact of spreading factor (SF) on the performance of the decoding algorithms?

Key findings

• The desynchronized algorithm successfully decodes both signals when n=2, achieving a 100% throughput gain over standard LoRa.
• For n=3, the algorithm achieves a throughput gain between 18% and 99%, depending on the spreading factor (SF), due to reduced frequency detection probability at higher SF.
• In the synchronized case, the algorithm enables decoding of one frame per two-colliding-frame collision, provided one node retransmits, resulting in a 25% throughput gain over LoRa.
• Throughput increases with duty cycle, and the proposed algorithms outperform LoRa by up to 60% at 1% duty cycle in a 100-node network.
• The desynchronized algorithm outperforms the synchronized one by 50% in throughput, due to the need for retransmissions in the latter case.
• LoRa fails to decode any signal when two or more signals collide on the same channel and SF, while the proposed algorithms restore significant decoding capability.