Cognitive Radio Oriented Wireless Networks. 13th EAI International Conference, CROWNCOM 2018, Ghent, Belgium, September 18–20, 2018, Proceedings

Research Article

An Image Processing Approach to Wideband Spectrum Sensing of Heterogeneous Signals

Download
163 downloads
  • @INPROCEEDINGS{10.1007/978-3-030-05490-8_20,
        author={Ha Nguyen and Ha Nguyen and Binh Nguyen},
        title={An Image Processing Approach to Wideband Spectrum Sensing of Heterogeneous Signals},
        proceedings={Cognitive Radio Oriented Wireless Networks. 13th EAI International Conference, CROWNCOM 2018, Ghent, Belgium, September 18--20, 2018, Proceedings},
        proceedings_a={CROWNCOM},
        year={2019},
        month={1},
        keywords={Wideband spectrum sensing Wireless signal detection Frequency hopping Time-frequency analysis Spectrogram Waterfall image Image morphology Blob extraction},
        doi={10.1007/978-3-030-05490-8_20}
    }
    
  • Ha Nguyen
    Ha Nguyen
    Binh Nguyen
    Year: 2019
    An Image Processing Approach to Wideband Spectrum Sensing of Heterogeneous Signals
    CROWNCOM
    Springer
    DOI: 10.1007/978-3-030-05490-8_20
Ha Nguyen1,*, Ha Nguyen1,*, Binh Nguyen1,*
  • 1: Viettel Research and Development Institute, Hoa Lac High-tech Park
*Contact email: nguyenquyha@gmail.com, khanhha318@gmail.com, thaibinhnx@gmail.com

Abstract

We introduce a simple yet efficient framework for the localization and tracking of fixed-frequency and frequency-hopping (FH) wireless signals that coexist in a wide radio-frequency band. In this spectrum sensing scheme, an energy detector is applied to each Short-time Fourier Transform of the wideband signal to produce a binary spectrogram. Bounding boxes for narrowband signals are then identified by using image processing techniques on a block of the spectrogram at a time. These boxes are also tracked along the time axis and fused with the newly detected boxes to provide an on-line system for spectrum sensing. Fast and highly accurate detection is achieved in simulations for various FF signals and FH signals with different hopping patterns and speeds. In particular, for the SNR of 4 dB over a bandwidth of 50 MHz, 97.98% of narrowband signals were detected with average deviations of about in time and in frequency.