A software-defined sensor architecture for large-scale wideband spectrum monitoring

Abstract

Today's spectrum measurements are mainly performed by governmental agencies which drive around using expensive specialized hardware. The idea of crowdsourcing spectrum monitoring has recently gained attention as an alternative way to capture the usage of wide portions of the wireless spectrum at larger geographical and time scales. To support this vision, we develop a flexible software-defined sensor architecture that enables distributed data collection in real-time over the Internet. Our sensor design builds upon low-cost commercial off-the-shelf (COTS) hardware components with a total cost per sensor device below $100. The low-cost nature of our sensor platform makes the sensing approach particularly suitable for large-scale deployments but imposes technical challenges regarding performance and quality. To circumvent the limits of our solution, we have implemented and evaluated different sensing strategies and noise reduction techniques. Our results suggest that our sensor architecture may be useful in application areas such as dynamic spectrum access in cognitive radios, detecting regions with elevated electro-smog, or simply to gain an understanding of the spectrum usage for advanced signal intelligence such as anomaly detection or policy enforcement.

Electrosense: Open and Big Spectrum Data

“…The PSD mode aids spectrum data retrieval using a restricted network bandwidth which enables applications where phase information is not crucial to analyze data in terms of signal power (Section IV-B). In addition, wideband scanning techniques are applied that prioritize frequency bands of bursty activities over bands which are stationary [3] to overcome the hardware-limitations of low-cost radios, for instance the limited sampling bandwidth of the low-end ADCs (e.g 2.4 MHz for RTL-SDR). For applications which require phase and non-averaged information, the sensor implements an IQ pipeline.…”

A low-cost sensor platform for large-scale wideband spectrum monitoring

“…This technique allows us to significantly improve the overall quality of the monitored spectrum since many bands are either not utilized or contain signals whose spectra do not change significantly over time. Our software architecture as well as the techniques to correct frequency errors and improve scanning times are describe in more details in [6].…”

Recursive Estimation of Dynamic RSS Fields Based on Crowdsourcing and Gaussian Processes

“…P. M. Djurić was supported by the National Science Foundation under Award CNS 1642965. measurements obtained by expensive equipment do not scale well and cannot be extended to large areas. An alternative and appealing solution is to exploit crowdsourcing, where many sensors with low-quality acquire much less accurate measurements [3], [4]. For example, one can use not very accurate measurements obtained by a large number of smart phones, and yet can create a more accurate spectrum map than that obtained by a few sparsely located expensive sensors.…”