Plugariu Ovidiu*
Abstract:” Cognitive Radio have become a key research area in the communications field over the past few
years as a powerful application addapted to the latest technological demands. Cognitive radios can work collaboratively
to make observations, share knowledge of the environment, and relate lessons from experience. In this project I have
studied how you can determine the modulation format using the computational method called recurrence plot analysis.”
I Cognitive Radio have become a key research area in the communications field over the past few years as a
powerful application addapted to the latest technological demands. Radio cognitive represents an evolved concept
based on the earlier software defined radios (SDR), software being the tool which gives them their enormous advantage
– “continous adaptability” [3].
The radio cognitive systems enable the use of artificial intelligence (AI) on flexible communications platforms,
such as software defined radios (SDR) to enable on-board, real-time optimization of frequency, time, power, and other
parameters. They can readapt to existing signal conditions through spectrum sensing and assignment of adequate
resources. Here is what experts say about cognitive radios : “ Making radios so smart that they can autonomously
discover how, when and where to use radio spectrum to obtain information services without having previously
programmed to do so”.
Their key characteristics are:
1. Sense the radio environment;
2. Make decisions;
3. Learn from experience to improve future decision making.
To obtain advantages given by the reconfigurability there has to be an efficient management of the entire
decision-making process. Cognitive radios achieve more efficient spectrum utilization by opportunistically finding
empty frequency bands. This new paradigm relies on the fact that a significant portion of the spectrum allocated to
lincesed services show little usage over time. This is a great advantage because the size of a frequency band is
physically determining the capacity of a radio network [3].
Cognitive radios can work collaboratively to make observations, share knowledge of the environment, and
relate lessons from experience. They can readapt their communications strategy with “friendly users” according to the
received feedback, they can switch radio bands or modulation/ demodulation protocols. This gives them great
adaptability for military applications because in this specific area friendly signals should be securely transmitted and
received whereas hostile signals must be located, identified and jammed.
II Automatic modulation classification (AMC) is an important component that improves the overall
performance of the cognitive radio.
AMC is often a difficult task to perform when there is no apriori information about the signal, such as signal
position, power carrier frequency and timing parameters.
There are many efforts oriented to improving AMC methods because communications protocols are frequenly
performed in non-cooperative environments. Moreover, in addition to multiple- propagation there are other problems
to be solved, like the frequency-selectivity and time-varying nature of a channel, especially when no prior knowledge
of the incoming signal is available.
2 Proper selection of the classification algorithm. There are two ways to do this:
IEEE has developped a special standard for cognitive radios (IEEE 802.11k) which includes the following
characteristics:
1. The utilization of a certain frequency band;
2. Noise histogram (estimated for a noise-only situation).
3. The hidden node ratio (existence of hidden receivers)
4. Time histogram in which the environment is loaded or free.
- The non-802.11k energy levels are measured
- The non 802-11k interferences are measured
The usage of Cognitive Radio technology has an overall emprovement of the spectrum management of 15% [4].
A signal is a time series which can be real or complex, with noise and modulations more or less difficult to
recognize and analyse. One possible analysis can be performed using a time-frequency approach.
x(t X(f
t f
Recurrence plot analysis : helps visualizing the recurrences in a two dimensional plot, hence it provides
information about the dynamics of the system highlighting the presence of certain patterns.
This method is appropriate for computation purposes because it can analyse high-dimensional dynamical-systems.
(1
Recurrence diagram
)
RM MM,M(R) RMi,j=H( )
H[n]=the Heaviside function (1)
H[n]= H[n]=
This recurrence matrix has the capacity to resume in a compact form the essential characteristics of a analysed
signal; being also known as “the signal’s fingerprint”.
When we reconstruct an RP we must choose an specific threshold . In an TRP (threshold RP) the pixel laying at
(i,j) is black if the distance falls within a specific corridor or white otherwise
Structures in RP’s
The principal arguments for the utilization of RP for the characterization of signals:
- It’s not based on signal’s energy based characterization/ energetic measurements decision;
- It doesen’t need a huge number of samples;
- Visualising the signal’s trajectory in state space allow us to reveal some auto-similarities which are difficult to
observe with a second order correlation;
- The recurrence plot representation is more flexible than a frequencial approach;
- Utility for recognition and detection of deterministic signal with noise
- RPA transforms a signal (one dimension) into a picture 2D:
a. we cand analyse the signal as a picture
b. it’s a redundant picture
c. the picture depends on the characteristics of the signal
d. moreover identifying could be useful to detection, characterization or
classification
Modulation must allow data transmission between transmitter and receiver without ambiguity. It’s a way of
protecting the signal from noise and adapting it to the transmission line. I have analysed two signals using the RPA;
first one is a sine wave (with and without noise) and the other is an FSK4 signal(with and without noise).
SINE ASK4
FSK4 WITH RECURRENCE PROCESSING NOISE FSK4 WITH RECURRENCE NOISE AND SYSTEM NOISE
FSK8 FSK8 WITH NOISE
Bibliography:1. www.recurrence–plot.tk
2. “ A Survey of Automatic Modulation Classification Techniques:
Classical Approaches and New Trends”
Octavia A. Dobre, Ali Abdi, Yeheskel Bar-Ness and Wei Su
3. “Cognitive radio architecture” Joseph Mitola
4. http://www.networkworld.com/news/tech/2004/0329techupdate.html
802.11k reference
Plots have been constructed with “CROSS RECURRENCE PLOT TOOLBOX 5.15” for Matlab developped by
Marwan&Kurts. Filtering algorithms are developed by the author.