Project
Names
Department of Electrical and Computer Engineering
Motivation
Cognitive Radio Systems (CRS)
Methodology
Future Work
Acknowledgements
Spectrum Sensing in Cognitive Radio Systems Shaun Kotikalapudi, Neeraj Venkatesan
Advisor: Dr. David Daut Slade Scholars Program
Rutgers School of Engineering
Spectrum Sensing Techniques
Figure 1: Edge detection of a
simulated signal structure with
sharp edges using continuous
wavelet transforms (CWT). The
signal has an SNR of 20 dB. The
wavelet transforms were taken for
dyadic scales 2^j, with
j=1,2,3,4,5,6,7. A Haar mother
wavelet was used to take the CWT
of the signal. As we can see in the
second graph, the edge detection
improves when the wavelet scales
increase. The algorithm is able to
detect all four edges at 200, 400,
600 and 800 MHz.
R𝑥𝛼 τ = lim
𝑇→ ∞ Rx t +
τ
2, t −
τ
2e−j2παt𝑑𝑡(1)
𝑇2
−𝑇2
𝑆𝑥𝛼 𝑓 = 𝑅𝑥
𝛼 𝜏 e−j2πfτ𝑑𝜏(2)
∞
−∞
Equation 1: Cyclic
Autocorrelation
Figure 2: BPSK Spectral Correlation Estimate, 1024 Hz Symbol Rate, 4096
Hz carrier, 16.384 KHz sampling rate, 128 signal frames, 128-point FFT
• Survey existing spectrum sensing algorithms in
Cognitive Radio Systems and examine signal
structure-based sensing algorithms as opposed to
energy detection`
Equation 2: Spectral Correlation Function
• Cognitive Radios use knowledge of their RF
environment to alter certain to attain a predefined
goal
• In recent years Cognitive Radios have been used as a
solution for an overcrowded spectrum
• One of the key aspects of Cognitive Radio Systems
is spectrum sensing, which was the focus of this
research effort
• Spectrum sensing traditionally measures the RF
energy within a certain spectral band in order to detect
the presence of a primary user, due to low complexity
• In a CRS paradigm spectrum sensing includes
obtaining spectrum characteristics across multiple
dimensions including frequency, time and space
• Some of the key spectrum sensing techniques
developed for CRS are as follows
• Energy detection: Measures the energy in the band of
interest to detect presence of a user
• Cyclostationary method: Uses cyclostationary
properties of certain signal structures to detect
presence of a user
• Matched Filter detection: Using matched filters to
correlate incoming signal with a known signal
structure to detect the presence of user
• Wavelet Detection: Converting incoming signals into
the wavelet domain and performing energy detection
in order to detect the presence of a user
• Conducted a literature review on existing spectrum
sensing techniques
• Decided to focus on implementing cyclostationary
and wavelet based spectrum sensing algorithms
• Cyclostationary Detection
• Initially tried to implement FFT
Accumulation method (FAM) algorithm to
study cyclostationary properties of certain
signal structures
• Settled on a slightly different spectral
correlation approach
• Traced cyclic domain profiles for different
modulation schemes (BPSK, BFSK and
QPSK)
• Wavelet Detection:
• Settled on a edge detection based Wavelet
approach for spectrum sensing
• Uses knowledge of frequency
discontinuities or “edges” in a signal PSD
to detect presence of primary user
• Implemented edge detection algorithm for
signals with sharp edges(using Haar
mother wavelet)
• Continue working on implementing spectrum
sensing decision algorithms for cyclostationary and
wavelet detection
• Work on alternative decision algorithms as opposed
to energy detection for wavelet and maximum
likelihood for cyclostationary
• Dr. David Daut
• IEEE