Analysis of OFDM parameters using cyclostationary spectrum sensing in Cognitive
Radio
Presented by :Omer Ali
What is a Cognitive Radio ?
• Cognitive Radio is built on the basis of a Software-defined Radios SDR
• Cognitive Radio can provide the spectral awareness technology to support FCC initiatives in Spectral Use
Cognitive Radio - Continued
• Cognitive radio: A radio that can change its transmitter parameters based on interaction with the environment in which it operates.
Is Cognitive Radio SMART ?• It knows where it is• It knows what services are
available, for example, it can identify then use empty spectrum to communicate more efficiently
• It knows what services interest the user, and knows how to find them
• It knows the current degree of needs and future likelihood of needs of its user
• Learns and recognizes usage patterns from the user
• Applies “Model Based Reasoning” about user needs, local content, environmental context
Why Spectrum Sensing ?
• Spectrum awareness or spectrum sensing makes a radio environment cognitive i.e. to memorize the spectrum holes or voids that could be utilized by utilizing the available spectrum and adapting to it by changing its parameters
Spectrum Holes
Power
Time
Frequency
Spectrum in use by Primary user
Spectrum Hole
• Main aspect: One main aspect of cognitive radio is related to autonomously exploiting locally unused spectrum to provide new paths to spectrum access.
Why to sense Spectrum holes ?• As FCC agrees on utilizing the spectrum holes for DVB-T for unlicensed
users; it is vital to lease this unused spectrum to users in the vicinity.• Finding spectrum holes ? That means the spectrum should be dispersed ?• The answer is somewhat YES. Think about utilizing the primary spectrum
for DVB-T applications and the secondary spectrum for unlicensed users.
Spectrum Utilization ?
Frequency
T
I
M
E
Spectral Adaptation Waveforms
How to Sense the Spectrum?
• Spectrum sensing is currently achieved dynamically using DSS
• Are there any trade-offs in terms of different sensing techniques ?
• The Answer is YES .– One might sense a empty spectrum easily but it
might be the one with very power SNR.– So, the goal is to sense the proper spectrum for
unlicensed users
Spectrum sensing - Methods
• Energy Detector Based Sensing: The signal is detected by comparing the output of the energy detector
with a threshold which depends on the noise floor.• Inability to differentiate interference from
primary users and noise, and poor performance under low signal-to-noise ratio (SNR)values.
Spectrum Sensing Methods
• Waveform-Based Sensing: Known patterns are usually utilized in wireless systems to assist synchronization or for other purposes. Such patterns include preambles, midambles, regularly transmitted pilot patterns, spreading sequences etc.
• Waveform-based sensing requires short measurement time.
Spectrum Sensing Methods
• Cyclostationarity-Based Sensing: Cyclostationarity feature detection is a method for detecting primary user transmissions by exploiting the
cyclostationarity features of the received signals.
• The cyclostationarity based detection algorithms can differentiate noise from primary users’ signals.
Why OFDM ?• OFDM symbols are used in this research because it
supports broader bandwidth and is normally utilized in current MIMO technologies.
• The modulation scheme can be varied and the corresponding spectrum efficiency and spectrum utilization varies per modulation scheme.
• Limitations – OFDM power leakages to adjacent channels
OFDM – Advantages / Disadvantages ?
• Advantages– Simple implementation by means of FFT– High spectral efficiency considering (no. of sub-
carriers)– Anti ICI and ISI makes OFDM receiver less complex, as
almost no equalizer is needed.• Disadvantages– Requires highly linear amplifiers– Sensitive to Doppler Effect– Guard-time introduces overhead
Research Goal ?
• Using OFDM for DVB-T applications calculate the primary and secondary users
• Improve bandwidth by removing guard-band , BUT , will it have any impact on ICI?
• If ICI increases, then we should come up with something for better utilization . Cyclic prefix maybe ….
• What to do with the received signal with lots of noise ? Maybe normalize the whole received spectrum and pick-up the most healthy spectrum ….
How to generate signals that matches close to DVB-T Application ?
• DVB-T systems can be used in either 2K or 8K mode. We choose 2K mode having :– 1705 sub-carriers are used to transmit the data
out of total 2048 sub-carriers – Inverse Fourier Transform (IFFT) of the QAM of the
data is taken and guard-band intervals are added at the start of OFDM frame for DVB-T applications
How did we proceed ?
1. QAM modulation2. OFDM signal generation3. Cyclic Prefix addition at the guard-band
locations4. Incorporating AWGN channel5. Symbol Transmission through AWGN6. Signal Detection using DSS techniques7. Spectral Correlation Function of the received
function for better PSD and noise removal
OFDM Signal Generation
QAM Mapping
Pilot Insertions
S-> P IFFT P -> S Cyclic Extension
Up conversionbitstream
Analog signal
QAM mapping is a block that groups these bits together as per modulation schemes:N=1 for BPSK, N=2 for QPSK and n-QAM for higher orders
Some Maths behind OFDM signals• For a single carrier, the complex signal can be:• If we consider N samples, OFDM signal appears to be summation of these N
symbols
• During the symbol length, the amplitude and phase remains constant
• These carriers are centered around fo , the time domain representation becomes
Where T is the period of sampling frequency.• This can be represented in complex vector as
Maths behind OFDM - continued• In last equation is the representation of complex components in
frequency domain• If we follow the IFFT transform, we can see that it is the summation of
orthogonal components in frequency domain
• The simplified complex form follows , where an and bn follows the modulation scheme, hence making:
• After complex vector multiplication, real signal part can be estimated as:
Cyclic Extension
• Last serial samples are added to next OFDM frame by cyclic extension
• How its done ? Lets see some basics and maths behind cyclic extension and Spectral correlation function to see its significance
Cyclostationary Features• A very simple periodic signal
• In terms of Fourier coefficients
• After modulation with a sine-wave
• Considering a is of random wide-sense spectrum nature, we can auto-correlate and can compute the power spectral density
• Auto-correlation of a • Power spectral density of a can be found by
• Keeping that in mind the Power Spectral Density of x(t) can be found by :
• Problem with the above equation ? No sine wave components presents
Cyclostationary Feature - continued
• Lets use trigonometric identities in order to have:1. Some DC components2. Some higher order periodic components3. Simple depiction of modulated periodic symbol
A simple quadratic function
Which can be reduced to
Furthermore b(t) has a DC component that should appear at f=0
Also, the higher order components should also appear at
Cyclostationary Feature - Continued
• So, if that is True, the PSD should appear as:
-fo foSy
-2fo 2fo
f
f
f
f
Cyclostationary Feature - Continued
• Problem with previous depiction?– Not every symbol appears as a DC with some known higher order
components– In order to add random delays, we should come up with some pulse
modulation in order to have varying magnitudes.– So, we can only have a DC magnitude appearing at nth order but with no
varying magnitudes.
• Speculating that into consideration, the basic function becomes:
• Where spectral lines should appear at m.fo , where m is integer multiplier• If we equate m.fo as ἀ , we can define our approximation equation:• ἀt = m.fo for periodic
Time intervals
Cyclostationary Feature - Continued
• Now with the assumptions we can say that the function is periodic if the delay product contains spectral lines; which can roughly be modeled as:
• The cyclic auto-correlation function can then proceed with the complex vector:
• Now the basic idea of Spectral Correlation function is to find average power in frequency domain
• The last approximations were to concentrate on the received signals at the center frequency as if they were passed through a narrowband filter
Where B is modeled as the bandwidth of the function for filtering
f
x
u
v
f + ἀ/2 f + ἀ/2
+ ἀ/2
-ἀ/2
Spectral Correlation Density
• The spectral correlation density was computed by the Fourier Transform of the cyclic autocorrelation
BPF
BPF
X(t)
e-j2πἀt
j2πἀte
U(t)
v(t)
Coding behind the project
Signal Generation
Serial Conversion
Coding - Continued
Cyclic Prefix addition
Up-sampling for carrier
Coding Continued
SCF Function
The Plots
The Outcomes
The PSD of generic symbol received
Outcomes - Continued
PSD while utilizing SCF
PSD without SCF
Outcomes -Continued
Reduced noise-bed and detected primary and
secondary users around center frequency in the
presence of SCF
Detected primary and secondary users around centre frequency in the
absence of SCF