Académique Documents
Professionnel Documents
Culture Documents
Partha Pratim Bhattacharya et al, International Journal of Computer Science & Communication Networks,Vol 1(2), 196-206
1. Introduction
Demand of radio spectrum is increasing day by day
due to increase in wireless devices and applications.
Current radio spectrum allocation is not efficient due to
regulations that have limited spectrum access, spatial
restriction on frequency usage. Peak traffic planning
causes temporal under utilization since spectrum
demand vary with time. A cognitive radio must detect
the presence of primary user to avoid interference. The
use of allocated spectrum varies at different times and
over different geographical regions. In accordance to a
report by Spectrum Policy Task Force of FCC, the Figure 1. Spectrum concentration
spectrum is under or scarcely utilized and this 2. Spectrum sensing
situation is due to the static allocation of the The important requirement of cognitive radio
spectrum rather than the physical shortage of the network is to sense the spectrum hole. Cognitive radio
spectrum as shown in Figure 1 [1]. Thus, to overcome has an important property that it detects the unused
the spectrum deficiencies and the inefficient utiliz- spectrum and shares it without harmful interference to
ation of the allocated frequencies, it is necessary to other users. It determines which portion of the
introduce new communication models through which spectrum is available and detects the presence of
frequency spectrum can be utilized whenever the licensed users when a user operates in licensed band.
The spectrum sensing enables the cognitive radio to Figure 2 shows the detailed classification of
detect the spectrum holes. Spectrum sensing spectrum sensing techniques. They are broadly
techniques can be classified as frequency domain classified into three main types, transmitter detection
approach and time domain approach. In frequency or non cooperative sensing, cooperative sensing and
domain method estimation is carried out directly interference based sensing. Transmitter detection
from signal so this is also known as direct method. In technique is further classified into energy detection,
time domain approach, estimation is performed using matched filter detection and cyclostationary feature
autocorrelation of the signal. Another way of detection [5].
categorizing the spectrum sensing and estimation
methods is by making group into model based 2.3.1. Non-cooperative detection This technique is
parametric method and periodgram based non- based on the detection of the weak signal from a
parametric method [2]. Another way of classification primary transmitter. In primary transmitter based
depends on the need of spectrum sensing as stated detection techniques, a cognitive user determines
below [3]: signal strength generated from the primary user. In
this method, the location of the primary receivers are
2.1. Spectrum sensing for spectrum not known to the cognitive users because there is no
opportunities signaling between the primary users and the
a. Primary transmitter detection: In this approach, cognitive users. Basic hypothesis model for
detection of a signal from a primary transmitter is transmitter detection can be defined as follows [6]
based on the received signal at CR users whether it is 𝑛 𝑡 𝐻0 ,
x(t) = (1)
present or not. It is also known as non-cooperative ℎ𝑠 𝑡 + 𝑛 𝑡 𝐻1 ,
detection. This method includes matched filter based where x(t) is the signal received by the cognitive
detection, energy based detection, cyclostationary user, s(t) is the transmitted signal of the primary
based detection, and radio identification based user, n(t)is the AWGN(Additive White Gaussian
detection. Noise) and h is the amplitude gain of the channel. H0
b. Cooperative or collaborative detection: It refers is a null hypothesis, H1 is an alternative hypothesis.
to spectrum sensing methods where information
from multiple Cognitive radio users is incorporated 2.3.1.1. Energy detection This technique is
for primary user detection. This approach includes suboptimal and can be applied to any signal.
either centralized access to the spectrum coordinated Conventional energy detector consists of a low pass
by a spectrum server or distributed approach. filter to reject out of band noise and adjacent signals.
Implement-ation with nyquist sampling A/D
2.2. Spectrum sensing for interference converter, square-law device and integrator as shown
detection in Figure 3(a) [7] –[15]. An energy detector can be
a. Interference temperature detection: In this implemented similar to a spectrum analyzer by
method the secondary users are allowed to transmit averaging frequ-ency bins of a FFT.
with lower power then the primary users and
restricted by interference temperature level so that Without loss of generality, we can consider a
there is no interference. Cognitive radio works in the complex baseband equivalent of the energy detector.
ultra wide band (UWB) technology. The detection is the test of the following two hypo-
b. Primary receiver detection: In this method, the theses: [15]
interference and/or spectrum opportunities are H0: Y [n] = W[n] signal absent
detected based on primary receiver's local oscillator H1: Y [n] = X[n] +W[n] signal present
leakage power.
n=1,…,N; where N is observation interval
2.3. Classification of spectrum sensing (2)
techniques
The noise samples W[n] are assumed to be additive
white Gaussian (AWGN) with zero mean and
variance σw. In the absence of coherent detection,
the signal samples X[n] can also be modeled as
Gaussian random process with variance σx. Note
that over-sampling would correlate noise samples
and in principle, the model could be always reduced
Figure 2. Classification of spectrum sensing into equation (2).
techniques [4]
A decision statistic for energy detector is: is significantly inferior to the optimum matched filter
2 detector whose sensing time scales as O(1/SNR).
T= 𝑁 𝑌 𝑛 (3)
interference from other secondary users sharing the large [19]. Another disadvantage of match filtering is
same channel and with the primary user. (3) It has large power consumption as various receiver algorit-
poor performance under low SNR conditions.This is hms need to be executed for detection. Further this
because the noise variance is not accurately known at technique is feasible only when licensed users are
the low SNR, and the noise uncertainty may render cooperating. Even in the best possible conditions,
the energy detection useless.(4) It doesnot work for the results of matched filter technique are bound by
the spread spectrum techniques like direct sequence the theoretical bound .
and frequency hopping.
2.3.1.3. Cyclostationary feature detection It have
been introduced as a complex two dimensional
2.3.1.2. Matched filter detection Matched-filtering
signal processing technique for recognition of
is known as the optimum method for detection of
modulated signals in the presence of noise and
primary users when the transmitted signal is known
interference [19]. To identify the received primary
[16]. The main advantage of matched filtering is the
signal in the presence of primary users it exploits
short time to achieve a certain probability of false
periodicity of modulated signals couple with sine
alarm or probability of misdetection [17]. Block
wave carriers, hopping sequences, cyclic prefixes
diagram of matched filter is shown in Figure 5.
etc. Due to the periodicity, these cyclostationary
signals exhibit the features of periodic statistics and
spectral correlation, which is not found in stationary
noise and interference [20] – [28]. Block diagram is
Figure 5. Block diagram of matched filter [18] shown in Figure 6.
complexity [20], [30]. In addition, feature detection Figure 8 (a) and (b) shows collaborative gain as a
needs the prior knowledge of the primary users . function of the number of colleborative radios for
sine wave and QPSK signals.
2.3.2. Cooperative or collaborative detection In
this tecchnique cognitive radio users are incoop-
erated, for this two methods Gains and Group
Intelligence are discussed .
Figure 9(a) shows that it is benifical to space the each SU broadcasts its hard decision (PU present or
antennas at distance larger the λ/2 to maximize the absent). The training signal for each SU is obtained
gain. Figure 9 (b) shows that performance improves by fusing these decisions using majority logic
with number of antennas , 2 antenna improves 𝑄𝐷 by decision. In the second case, knowledge of (P d, Pf)
4 % and 4 antennas maximize it by 25% gain. for each SU is assumed and training signal is
obtained using Log Likelihood Ratio Test (LLRT).
2.3.2.2. Group intelligence A spectrum sensing
techniq-ue using group intelligence is proposed
where multiple users, each with incomplete
information, can learn from the group‘s wisdom to
reach a supposedly correct conclusion. He proposed
adaptive threshold based on group intelligence
accuracy of the spectral occupancy has improved by
training the Secondary User (SU).This training is
done with the help of the global decision derived
from other SUs in the network. This approach is
suitable in two ways. (1)To indicate the presence or
absence of the PU separate training signal are not
required. Training can be done continually, not Figure 11. 𝑃𝑚 vs SNR with SNR Estimate
limited to the training signal. (2)Training signal Error=1dB, mean SNR of group= 5dB [35]
derived from spatially diverse SUs should be robust
and less sensitive to local channel fading. Together Figure 11 shows the simulation results under same
the SUs are expected to reach a unanimous decision, conditions as before, except here the error in SNR
which should be correct in all situations [35]. estimation is smaller 1 dB. It can be see that the
method is relatively insensitive to SNR estimation
Group Intelligence emerges from the collaboration error and continues to perform better than the
and competition among many individuals, enhancing distributed sensing methods. Thus the decision
the social pool of existing knowledge. The concept margin is an indicator of the confidence we have in
of Group Intelligence obtains prominence in the our final decision and can also be interpreted as
context of Multi Agent scenarios where in the global reliability of the result. Figure 12 represents this
decision made is an essence of the all the individual scenario.
local decisions. In the Cognitive Adhoc Network
(CAN) scenario, the group intelligence in terms of
the accuracy of the spectral information learned by
the group of SUs. In other words lesser the number
of decision errors, wiser is the network [36].
requirement decreases. Figure 13 shows the sensiti- overhead and system requirements. An alternate
vity benefits obtained from a partially cooperative solution to this is cognitive relaying proposed by
coordinated centralized scheme consid-ered by [37]. [39] shown in Fihure 15. In cognitive relaying the
This also shows a -25 dBm reduction in sensitivity secondary user selflessly relays the primary users
threshold obtained by this scheme [38]. transmission thereby decreasing the primary users
transmission time. Thus cognitive relaying in effect
creates ―spectrum holes‖.
2.3.4.4. Large Sensory Data Cognitive radio have this architecture RF signal to be converted down to a
the ability to use any ununsed spectrum hole so it fixed lower intermediate frequency (IF), replacing a
will have to scan wide range of spectrum, resulting low Q tunable RF filter with a low-cost high-Q IF
in large amount of data. filter. A local oscillator (LO) is used to down-
convert an RF band to IF . This local oscillator is
2.4. Interference based sensing For interference tuned to a frequency such that when mixed with the
based spectrum sensing techniques there are two incoming RF signal, the desired RF band is down-
methods proposed. (a) Interference Temperature converted to the fixed IF band. In all of these
Management (b) Primary Receiver Detection. receivers, there is inevitable reverse leakage, and
therefore some of the local oscillator power actually
2.4.1. Interference temperature management couples back through the input port and radiates out
Interference is actually happen at receivers, it is of the antenna [42].
regulated in trans-centric way and is controlled at the
transmitter through the radiated power and location
of individual transmitters. Interference temperature
model is shown in Figure 16. The interference temp-
erature is a measure of the RF power available at a
receiving antenna to be delivered to a receiver, refle-
cting the power generated by other emitters and
noise sources [39]. specifically, it is defined as the
temperature equivalent to the RF power available at
a receiving antenna per unit bandwidth [41], i.e., Figure 17. Superheterodyne receiver [42]
𝑃𝐼 (fc ,B)
𝑇𝐼 (𝑓𝑐 , B) = (9)
𝑘𝐵
discussed that basically involves non cooperative, Networks,‖ in Proc. IEEE Global Telecomm. Conf.
cooperative and interference based detection. For (Globecom), 2006, San Francisco, California, USA.
cooperative detection group intelligence technique is [15] D. Cabric, A. Tkachenko, and R. Brodersen,
also discussed. ―Spectrum Sensing Measurements of Pilot, Energy and
Collaborative Detection,‖ in Proc. IEEE Military
Commun. Conf., 2006, Washington, D.C., USA, pp 1–7.
4. References [16] J. G. Proakis, ―Digital Communications‖, 2001,4th ed.
McGraw-Hill.
[1] Federal Communications Commission (FCC),― Spec- [17]R. Tandra and A. Sahai, ―Fundamental Limits on
trum Policy Task Force,‖ Report, 2002, pp 2-135. Detection in Low SNR Under Noise Uncertainty,” in Proc.
[2] Mansi subhedar and Gajanan Birajdar, ―Spectrum IEEE Int. Conf. Wireless Networks, Communication and
Sensing Techniques in Cognitive Radio: a Survey‖, Mobile Computing, 2005, vol. 1, Maui, HI, pp 464–469.
International Journal of Next Generation Networks,2011, [18] Shahzad A.,―Comparative Analysis of Primary
vol. 3, No.2. Transmitter Detection Based Spectrum Sensing
[3] Bruce A. Fette, ―Cognitive Radio Technology‖,Newnes Techniques in Cognitive Radio Systems,‘‘ Australian
Publisher,2006. Journal of Basic and Applied Sciences, 2005, pp 4522-
[4] Ekram Hossain, Dusit Niyato, Zhu Han, ―Dynamic 4531, INSInet Publication.
Spectrum Access and Management in Cognitive Radio [19]W.A.Gardner, ―Signal Interception: A Unifying
Networks‖, Cambridge University Press,2009. Theoretical Framework for Feature Detection‖, IEEE
[5] Tevfik Yucek and Huseyin Arslan (2009), ―A Survey Trans. on Communications,1988, vol. 36, No. 8.
of Spectrum Sensing Algorithms for Cognitive Radio [20] D. Cabric, S. Mishra, and R. Brodersen, ―Implem-
Applications‖, IEEE Communication Surveys & Tutorials, entation Issues in Spectrum Sensing for Cognitive Radios,‖
Vol.11, No.1, pp 116-130. in Proc. Asilomar Conference on Signals,Systems and
[6] A. Ghasemi, E.S.Sousa, ―Collaborative Spectrum Sen- Computers,2006,vol.1,PacificGrove,California, USA,, pp
sing for Opportunistic Access in Fading Environment‖, in: 772–776.
Proc. IEEE DySPAN 2005, pp 131–136. [21] N. Khambekar, L. Dong, and V. Chaudhary,
[7] S. Shankar, C. Cordeiro, and K. Challapali, ―Spectrum ―Utilizing OFDM Guard Interval for Spectrum Sensing,‖
Agile Radios: Utilization and Sensing Architectures,‖ in in Proc. IEEE Wireless Commun. And Networking
Proc. IEEE Int. Symposium on New Frontiers in Dynamic Conf.,2007, Hong Kong, pp 38–42.
Spectrum Access Networks,2006, Baltimore, Maryland, [22] P. Qihang, Z. Kun, W. Jun, and L. Shaoqian, ―A
USA, pp.160–169. distributed Spectrum Sensing Scheme Based on Credibility
[8] Y. Yuan, P. Bahl, R. Chandra, P. A. Chou, J. I. Ferrell, and Evidence Theory in Cognitive Radio Context,‖ in
T. Moscibroda, S. Narlanka, and Y. Wu, ―Knows: Proc. IEEE Int. Symposium on Personal, Indoor and
Cognitive Radio Networks Over White Spaces,‖ in Proc. Mobile Radio Commun., 2006,Helsinki, Finland, pp 1–5.
IEEE Int. Symposium on New Frontiers in Dynamic [23] M. Oner and F. Jondral, ―Cyclostationarity Based Air
Spectrum Access Networks, 2007,Dublin, Ireland, pp 416– Interface Recognition for Software Radio Systems,‖ in
427. Proc. IEEE Radio and Wireless Conf., 2004,Atlanta,
[9] G. Ganesan and Y. Li, ―Agility improvement through Georgia, USA, pp 263–266.
cooperative diversity in cognitive radio,” in Proc. IEEE [24] A. Fehske, J. Gaeddert, and J. Reed, ―A New
Global Telecomm. Conf. (Globecom), 2005, vol. 5, St. Approach to Signal Classification using Spectral Correl-
Louis, Missouri, USA, pp 2505–2509. ation and Neural Networks,‖ in Proc.IEEE Int. Symposium
[10]——, ―Cooperative Spectrum Sensing in Cognitive on New Frontiers in Dynamic Spectrum Access
Radio Networks,‖ in Proc. IEEE Int. Symposium on New Networks,2005, Baltimore, Maryland, USA, pp 144–150.
Frontiers in Dynamic Spectrum Access Networks,2005, [25] M. Ghozzi, F. Marx, M. Dohler, and J. Palicot,
Baltimore, Maryland, USA, pp 137–143. ―Cyclostationarity Based Test for Detection of Vacant
[11] D. Datla, R. Rajbanshi, A. M. Wyglinski, and G. J. Frequency Bands,‖ in Proc. IEEE Int. Conf. Cognitive
Minden, ―Parametric Adaptive Spectrum Sensing Radio Oriented Wireless Networks and Commun.
Framework For Dynamic Spectrum Access Networks (Crowncom),2006,Mykonos Island, Greece.
(2007),‖ in Proc. IEEE Int. Symposium on New Frontiers [26] N. Han, S. H. Shon, J. H. Chung, and J. M. Kim,
in Dynamic Spectrum Access Networks, 2007, Dublin, ―Spectral Correlation Based Signal Detection Method for
Ireland, pp 482–485. Spectrum Sensing in IEEE 802.22 WRAN systems,‖ in
[12] T. Weiss, J. Hillenbrand, and F. Jondral, ―A diversity Proc. IEEE Int. Conf. Advanced Communication
approach for the detection of idle spectral resources in Technology,2006, vol. 3.
spectrum pooling systems,‖ in Proc. of the 48th Int. [27] J. Lund´ en, V. Koivunen, A. Huttunen, and H. V.
Scientific Colloquium, 2003, Ilmenau, Germany, pp 37–38. Poor, ―Spectrum Sensing in Cognitive Radios based on
[13] F. Digham, M. Alouini, and M. Simon, ―On the Multiple Cyclic Frequencies,” in Proc. IEEE Int. Conf.
energy detection of unknown signals over fading Cognitive Radio Oriented Wireless Networks and
channels,‖ in Proc. IEEE Int. Conf. Commun., vol. 5, Commun. (Crowncom),2007, Orlando, Florida, USA.
2003,Seattle, Washington, USA, pp 3575–3579. [28] K. Kim, I. A. Akbar, K. K. Bae, J.-S. Um, C. M.
[14] P. Pawełczak, G. J. Janssen, and R. V. Prasad, Spooner, and J. H. Reed, ―Cyclostationary Approaches to
―Performance Measures of Dynamic Spectrum Access Signal Detection and Classification in Cognitive Radio,‖ in
Proc. IEEE Int. Symposium on New Frontiers in Dynamic