Spectral Occupancy at VHF: Implications for Cognitive Radios

Transcription

1 Spectral Occupancy at VHF: Implications for Cognitive Radios Steve Ellingson Mobile & Portable Radio Research Group (MPRG) Dept. of Electrical & Computer Engineering Virginia Polytechnic Institute & State University Fall 2005 IEEE Vehicular Technology Conference (Dallas)

2

3 Frequency/Bandwidth-Agile Cognitive Radio An emerging paradigm for the operation of radio networks in which individual radios will be able to monitor the available spectrum and select combinations of frequencies and bandwidths which are mutually optimum in some sense. Possible solution to present-day rigid, inefficient use of spectrum The VHF ( MHz) band is particularly bad off in this respect: Badly fragmented, with much spectrum left unused or used with only very low duty cycle Preliminary questions: How much spectrum might actually be available for this? How difficult is it to access (esp. receiver dynamic range)

4 Receiver Sensitivity / Dynamic Range Trade-Off Tradeoff can be made less painful by reducing bandwidth, but 10 s of MHz will probably be required in any event.

5 Urban Measurement Setup Columbus, OH (Business/Residential Area) Antenna 14 m above ground Coax cable Agilent E4407B Spectrum Analyzer RS232 PC AOR Model DA MHz Measured Sensitivity -87 db(mw/[30 khz]), i.e db(mw/hz) Linear Power Detection 30 khz Channels 83 µs dwell ~ 400 sweeps

6 VHF Noise Backgrounds

7 Urban Measurement Setup Columbus, OH (Business/Residential Area) Antenna 14 m above ground Coax cable Agilent E4407B Spectrum Analyzer RS232 PC AOR Model DA MHz Linear Power Detection 30 khz Channels 83 µs dwell ~ 400 sweeps Measured Sensitivity -87 db(mw/[30 khz]), i.e db(mw/hz) Sensitivity Relative to ITU Business Model 30 MHz: +7 dbf am 300 MHz: +34 dbf am

8 Urban: HF to 90 MHz HF Comm TV4 TV6

9 Urban: HF to 90 MHz HF Comm TV4 TV MHz

10 FM Urban: MHz 2-Way Radio Aeronautical/ Satellite

11 FM Urban: MHz 2-Way Radio Aeronautical/ Satellite MHz

12 TV10 Urban: MHz

13 Spectral Occupancy in MHz

14 Spectral Occupancy in MHz

15 Statistics of Spectral Occupancy Band # openings Mean (MHz) >= 30 khz Opening khz 67% BW Implied dynamic range < 30 db khz 25% BW Implied dynamic range > 60 db -87 db(mw/[30 khz]) Detection Threshold

16 Rural Measurement Setup Rosman, NC Antenna 2 m above ground Coax cable R&S FSH3 Spectrum Analyzer RS232 PC Custom-Built Fat Dipole Linear power detection 300/30 khz channels

17 Rural FSH3 300 khz Spectrum analyzer ( ν=300 khz) at end of feedline Galaxy, VSWR=1 Galaxy, VSWR=12 Galaxy, VSWR=100 A4

18 Rural: khz RBW FSH3 300 khz FSH3 30 khz Measurements PARI (Rosman, NC). Spectrum analyzer ( ν=300 khz) at end of feedline dbm in [30,85] MHz A4

19 Rural: khz RBW 30 khz 300 khz TV Ch 4 Spectrum analyzer ( ν=300 khz) at end of feedline A4

20 Rural Measurement Setup (Spectrometer) Rosman, NC Antenna 2 m above ground Coax cable Custom Spectrometer PCI PC Custom-Built Fat Dipole Linear Power Detection 4 MHz Swept BW 610 Hz Channels Measured Sensitivity -100 db(mw/[30 khz]), i.e db(mw/hz) Sensitivity ~F am relative to ITU Rural model ~13 db improvement over urban measurement

21 Rural: MHz

22 Rural: Spectragram

23 Rural: Zoom Around 38 MHz Most of the spectrum starts to look pretty good at resolutions below 1 khz! FSH3 30 khz PLFM 610 Hz Max Hold Integration Measurements PARI (Rosman, NC). Spectrum analyzer ( ν=300 khz) at end of feedline dbm in [30,85] MHz A4

24 Conclusions Most of the VHF spectrum away from persistent broadcast signals appears to be useful for frequency-agile cognitive radio E.g., 80% of the MHz band was clear to within 7 db of ITUdefined noise floor in an urban area Tens of empty gaps found in the MHz and MHz bands, with mean bandwidths of 100 s of khz (WRT -87 db(mw/[30 khz]) threshold) Very little activity above -93 db(mw/[30 khz]) in the rural setting Much more thorough measurement campaign needed to completely understand this 1 ms x 1 khz time-frequency resolution Continuous observation over many day/week cycles of human activity