reZig: Decompose a Collision via Reference Waveform in ZigBee
Yifeng Cao, Zhe Wang, Linghe Kong, Guihai Chen · 2019
With the proliferation of IoTs, massive deployments of ZigBee devices intensify the collision, a fundamental problem in wireless communication. Conventional solutions exploit the protocol based on CSMA to avoid collision, which incurs considerable overhead. Recently, new solutions which focus on concurrent transmission via collision decomposition are proposed instead of collision avoidance. However, existing approaches to collision decomposition in ZigBee suffer the problems of error propagation and scarce-sample error. To make up the weakness of existing solutions, we propose a novel solution, reZig, which leverages the generated reference waveform in the receiver to infer each single waveform in the collision. By inferring the chip independently, reZig is error-propagation-free. As there are multiple segments provided to decode one chip, reZig can select the segment with most raw samples to ensure the decoding reliability. We implement reZig on the USRP N210 testbed. Experiments demonstrate that reZig achieves the 1.46~ 2.8× throughput compared to mZig, the state-of-the-art collision-decomposition approach to ZigBee.