Revisiting Software Defined Radios in the IoT Era

Several years ago, software radios were seen as the future of commercial wireless infrastructure, given that they could flexibly decode any wireless technology with a simple software update. Yet, despite their numerous advantages, the industry tilted in favor of dedicated chips for wireless infrastructure sacrificing generality for performance. Today, the advent of the Internet of Things (IoT) and the resulting fragmentation of wireless technologies has led to a new billion-dollar industry: multi-technology gateways that support many radio technologies. Predictably, commercial gateways do this through multiple radio chips each decoding dedicated wireless technologies. This paper argues that software radios deserve a revisit in the IoT era. Specifically, IoT wireless technologies have significantly lax requirements in terms of throughput, latency and cost, meaning that they can be implemented on cheap software radio gateways that offer extensibility through software updates. We propose GalioT, a cloud-assisted IoT gateway that demonstrates the numerous benefits of software radios beyond programmability. We focus, in particular, on the rampant cross-technology interference in IoT networks since devices are low-power and simply "wake up and transmit". We use simple software radio base stations that detect signals including collisions of multiple radio technologies and ship them to the cloud. At the cloud, we exploit PHY-layer differences across radio technologies to concurrently decode them. A prototype implementation on an inexpensive RTL-SDR and Raspberry-Pi based gateway shows promising accuracy in concurrent decoding of XBee, Z-Wave and LoRa.