AirPress: High-accuracy spectrum summarization using compressed scans

摘要

Spectrum summarization is the analysis of a wide-band spectrum scan to determine the number of transmitters, their time-frequency characteristics, approximate modulation and legitimacy of operation. Spectrum summarization has emerged as a critical functionality to enable next-generation dynamic spectrum access technologies and legislation. Typically, spectrum summarization is performed in a cloud-based manner, requiring full-scan transmission from the spectrum sensors to the cloud. As spectrum scans generate large volumes of data, full-scan transmission quickly incurs prohibitively-high cost in terms of bandwidth and storage requirements. To address this problem we design AirPress, a spectrum scan compression method that leverages wavelet decomposition for lossy compression of spectrum data and allows up to 64:1 compression ratio of power spectral density traces without adversely impacting the spectrum summarization accuracy. We demonstrate the utility of AirPress on real-world spectrum measurements and show that it enables high-accuracy spectrum summarization of real-world transmitters while reducing the corresponding trace by 94%.