LoRa DL: a deep learning model for enhancing the data transmission over LoRa using autoencoder

• Published in: The Journal of supercomputing Vol. 79; no. 15; pp. 17079 - 17097
• Main Authors: Shilpa, B., Kumar, Puranam Revanth, Jha, Rajesh Kumar
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.10.2023; Springer Nature B.V
• Subjects: Bit error rate; Communication; Communications systems; Compilers; Computer architecture; Computer Science; Data transmission; Deep learning; Interpreters; Performance degradation; Power consumption; Power management; Processor Architectures; Programming Languages; Signal to noise ratio; Wide area networks; Wireless communications; Deep learning; Internet of Things; Autoencoder; LoRa; Wireless Communication
• ISSN: 0920-8542; 1573-0484
• DOI: 10.1007/s11227-023-05355-4

Low Power Wide Area Networks (LPWANs) have become a popular option for modern wireless communication technologies. Long Range (LoRa) protocol was designed for LPWAN, which offers long-distance communication, low-power consumption, and simultaneous transmissions. Long-distance communication necessitates an extremely low signal-to-noise ratio at the receiver. Additionally, low power consumption necessitates less signaling, which leads to the usage of simpler protocols like ALOHA and less coordinated communication. Therefore, as the number of devices equipped with this technology grows, its performance will degrade naturally as a result of scalability and interference difficulties. Deep learning holds great promise for resolving these problems through data-driven approaches and enhances the efficiency of LoRaWAN in usage of limited spectrum resources. In this work, we present the design of an end-to-end communication system as an autoencoder architecture utilizing deep learning. This adaptable architecture is able to efficiently capture channel impairments while simultaneously optimizing the operations of the transmitter and receiver together. The autoencoder is designed with multiple system model elements, including a sender net that simulates a LoRa transmitter and modulates data, a channel net that models channel impairments, and a receiver net that acts as a LoRa receiver to demodulate and retrieve the original data. The proposed autoencoder model is trained and evaluated with LoRa samples generated by simulation, and it is shown to be a high performer by comparing with the traditional LoRaWAN in terms of Bit Error Rate (BER) and Packet Success Rate (PSR) measurements.