Maximizing available spectrum for cognitive radios

Shridhar Mubaraq Mishra

EECS Department
University of California, Berkeley
Technical Report No. UCB/EECS-2010-1
January 7, 2010

http://www2.eecs.berkeley.edu/Pubs/TechRpts/2010/EECS-2010-1.pdf

Cognitive radios have been proposed to address the problem of spectrum under-utilization. Previous research in the area of cognitive radios has concentrated on signal processing (SP) innovations for improved detection with a limited number of samples (a.k.a detection sensitivity). However the SP perspective alone is unable to recover much unused spectrum. In this thesis we take a spatial perspective on spectrum usage and examine advanced algorithms which use other radios and frequencies to aid the detection process.

The spatial perspective is necessitated by the FCC’s decision to open up TV bands for opportunistic spectrum use. Examination of the FCC’s choice of parameters reveals the political and engineering tradeoffs made by the FCC. The FCC’s rules applied to the database of TV transmitters and population of the United States as per the census of 2000 reveal that the geo-location rules enable an average of 9 white space channels per person. This number does not change significantly even if secondaries operate only where the pollution from TV transmitters is low. The corresponding political tradeoff sacrifices 1 people-channel for broadcast use to gain 8 people-channels for white space use. The FCC’s sensing rule of -114dBm is very conservative and yields only a average of 1 white space channel per person.

This spatial analysis points to the inability of traditional detection metrics to predict the spatial performance of sensing. To overcome this problem, we propose the twin metrics of Fear of Harmful Interference (FHI) and Weighted Probability of Area Recovered (WPAR) to quantify the performance of sensing relative to geo-location. Fear of Harmful interference, captures the safety to the primary users and is largely the fading-aware probability of missed detection with modifications to allow easier incorporation of system-level uncertainty. Weighted Probability of Area Recovered, captures the performance of spectrum sensing by appropriately weighting the probability of false alarm (PFA) across different spatial locations.

Cooperative Sensing, in which secondaries use sensing results from multiple nearby radios has been proposed as a mechanism to improve the performance over a single radio. A median rule is proposed as a robust cooperation rule for spatial holes. This rule can be implemented as a hard combining rule and is robust to uncertainties in the fading models and untrusted radio behavior.

Unfortunately all cooperation suffers when there is a loss in spatial diversity. To combat shadowing correlation across space, assisted/calibrated detection in the form of multiband sensing is proposed. In multiband sensing, detection results from nearby frequencies are used to aid detection in the channel of interest. Cooperation across multiband radios is robust against channel correlation and provides gains by weeding out radios afflicted by adverse propagation environments. A mobile, wideband testbed was designed and used to capture TV signals in the 500-700MHz band at various locations in Berkeley, CA. Analysis of these measurements reveals high shadowing and multipath spread correlation across frequencies. These new insights into the spectral environment are used to design detectors that perform better than a single band detector.

The ability to coexist with primaries of different scales (high power TV transmitters and low power wireless microphones) is an important requirement for a cognitive radio system. When sensing a large scale primary, a small scale secondary user can make its own decision about transmission based on the sensing results from its neighborhood. This assumption fails when the scale of the primary is comparable to the scale of the secondary user. In this scenario, we need to decouple sensing from admission control – a sensor network is required to perform the sensing and localize the primary.

Advisor: Robert W. Brodersen and Anant Sahai


BibTeX citation:

@phdthesis{Mishra:EECS-2010-1,
    Author = {Mishra, Shridhar Mubaraq},
    Title = {Maximizing available spectrum for cognitive radios},
    School = {EECS Department, University of California, Berkeley},
    Year = {2010},
    Month = {Jan},
    URL = {http://www2.eecs.berkeley.edu/Pubs/TechRpts/2010/EECS-2010-1.html},
    Number = {UCB/EECS-2010-1},
    Abstract = {Cognitive radios have been proposed to address the problem of spectrum under-utilization. Previous research in the area of cognitive radios has concentrated on signal processing (SP) innovations for improved detection with a limited number of samples (a.k.a detection sensitivity). However the SP perspective alone is unable to recover much unused spectrum. In this thesis we take a spatial perspective on spectrum usage and examine advanced algorithms which use other radios and frequencies to aid the detection process.

The spatial perspective is necessitated by the FCC’s decision to open up TV bands for opportunistic spectrum use. Examination of the FCC’s choice of parameters reveals the political and engineering tradeoffs made by the FCC. The FCC’s rules applied to the database of TV transmitters and population of the United States as per the census of 2000 reveal that the geo-location rules enable an average of 9 white space channels per person. This number does not change significantly even if secondaries operate only where the pollution from TV transmitters is low. The corresponding political tradeoff sacrifices 1 people-channel for broadcast use to gain 8 people-channels for white space use. The FCC’s sensing rule of -114dBm is very conservative and yields only a average of 1 white space channel per person.

This spatial analysis points to the inability of traditional detection metrics to predict the spatial performance of sensing. To overcome this problem, we propose the twin metrics of Fear of Harmful Interference (FHI) and Weighted Probability of Area Recovered (WPAR) to quantify the performance of sensing relative to geo-location. Fear of Harmful interference, captures the safety to the primary users and is largely the fading-aware probability of missed detection with modifications to allow easier incorporation of system-level uncertainty. Weighted Probability of Area Recovered, captures the performance of spectrum sensing by appropriately weighting the probability of false alarm (PFA) across different spatial locations. 

Cooperative Sensing, in which secondaries use sensing results from multiple nearby radios has been proposed as a mechanism to improve the performance over a single radio. A median rule is proposed as a robust cooperation rule for spatial holes. This rule can be implemented as a hard combining rule and is robust to uncertainties in the fading models and untrusted radio behavior.

Unfortunately all cooperation suffers when there is a loss in spatial diversity. To combat shadowing correlation across space, assisted/calibrated detection in the form of multiband sensing is proposed. In multiband sensing, detection results from nearby frequencies are used to aid detection in the channel of interest. Cooperation across multiband radios is robust against channel correlation and provides gains by weeding out radios afflicted by adverse propagation environments. A mobile, wideband testbed was designed and used to capture TV signals in the 500-700MHz band at various locations in Berkeley, CA. Analysis of these measurements reveals high shadowing and multipath spread correlation across frequencies. These new insights into the spectral environment are used to design detectors that perform better than a single band detector.

The ability to coexist with primaries of different scales (high power TV transmitters and low power wireless microphones) is an important requirement for a cognitive radio system. When sensing a large scale primary, a small scale secondary user can make its own decision about transmission based on the sensing results from its neighborhood. This assumption fails when the scale of the primary is comparable to the scale of the secondary user. In this scenario, we need to decouple sensing from admission control – a sensor network is required to perform the sensing and localize the primary.}
}

EndNote citation:

%0 Thesis
%A Mishra, Shridhar Mubaraq
%T Maximizing available spectrum for cognitive radios
%I EECS Department, University of California, Berkeley
%D 2010
%8 January 7
%@ UCB/EECS-2010-1
%U http://www2.eecs.berkeley.edu/Pubs/TechRpts/2010/EECS-2010-1.html
%F Mishra:EECS-2010-1