Measuring Spectrum Similarity in Distributed Radio Monitoring Systems

Abstract

The idea of distributed spectrum monitoring with RF front-ends of a few dollars is gaining attention to capture the real-time usage of the wireless spectrum at large geographical scale. Yet the limited hardware of these nodes hinders some of the applications that could be envisioned. In this work, we exploit the fact that, because of its affordable cost, massive deployments of spectrum sensors could be foreseen in the early future, where the radio signal of one wireless transmitter is received by multiple spectrum sensors in range and connected over the public Internet. We envision that nodes in this scenario may collaboratively take decisions about which portion of the spectrum to monitor or not. A key problem for collaborative decision is to identify the conditions where the nodes receive the same spectrum data. We take an initial step in this direction, presenting a collaborative system architecture, and investigating the challenges to correlate pre-processed data in the backend, with key insights in the trade-offs in the system design in terms of network bandwidth and type of over-the-air radio signals. Our results suggest that it is feasible to determinate in the backend if two sensors are reading the same analog/digital signal in the same frequency, only sampling during 200 milliseconds and sending just 1 Kbyte of data per sensor to the backend.