Title:Nephalai: Towards LPWAN C-RAN with Physical Layer Compression

Abstract:We propose Nephelai, a Compressive Sensing-based Cloud Radio Access Network (C-RAN), to reduce the uplink bit rate of the physical layer (PHY) between the gateways and the cloud server for multi-channel LPWANs. Recent research shows that single-channel LPWANs suffer from scalability issues. While multiple channels improve these issues, data transmission is expensive. Furthermore, recent research has shown that jointly decoding raw physical layers that are offloaded by LPWAN gateways in the cloud can improve the signal-to-noise ratio (SNR) of week radio signals. However, when it comes to multiple channels, this approach requires high bandwidth of network infrastructure to transport a large amount of PHY samples from gateways to the cloud server, which results in network congestion and high cost due to Internet data usage. In order to reduce the operation's bandwidth, we propose a novel LPWAN packet acquisition mechanism based on Compressive Sensing with a custom design dictionary that exploits the structure of LPWAN packets, reduces the bit rate of samples on each gateway, and demodulates PHY in the cloud with (joint) sparse approximation. Moreover, we propose an adaptive compression method that takes the Spreading Factor (SF) and SNR into account. Our empirical evaluation shows that up to 93.7% PHY samples can be reduced by Nephelai when SF = 9 and SNR is high without degradation in the packet reception rate (PRR). With four gateways, 1.7x PRR can be achieved with 87.5% PHY samples compressed, which can extend the battery lifetime of embedded IoT devices to 1.7.