RESEARCH Open Access Spectrum sensing and resource ... · rithm [27,28]. The resource allocation...

Dikmese et al. EURASIP Journal on Advances in Signal Processing 2014, 2014:68http://asp.eurasipjournals.com/content/2014/1/68

RESEARCH Open Access

Spectrum sensing and resource allocation formulticarrier cognitive radio systems underinterference and power constraintsSener Dikmese1*, Sudharsan Srinivasan1, Musbah Shaat2, Faouzi Bader3 and Markku Renfors1

Abstract

Multicarrier waveforms have been commonly recognized as strong candidates for cognitive radio. In this paper, westudy the dynamics of spectrum sensing and spectrum allocation functions in cognitive radio context using verypractical signal models for the primary users (PUs), including the effects of power amplifier nonlinearities. We startby sensing the spectrum with energy detection-based wideband multichannel spectrum sensing algorithm andcontinue by investigating optimal resource allocation methods. Along the way, we examine the effects of spectralregrowth due to the inevitable power amplifier nonlinearities of the PU transmitters. The signal model includesfrequency selective block-fading channel models for both secondary and primary transmissions. Filter bank-basedwideband spectrum sensing techniques are applied for detecting spectral holes and filter bank-based multicarrier(FBMC) modulation is selected for transmission as an alternative multicarrier waveform to avoid the disadvantage oflimited spectral containment of orthogonal frequency-division multiplexing (OFDM)-based multicarrier systems. Theoptimization technique used for the resource allocation approach considered in this study utilizes the informationobtained through spectrum sensing and knowledge of spectrum leakage effects of the underlying waveforms,including a practical power amplifier model for the PU transmitter. This study utilizes a computationally efficientalgorithm to maximize the SU link capacity with power and interference constraints. It is seen that the SU transmissioncapacity depends critically on the spectral containment of the PU waveform, and these effects are quantified in a casestudy using an 802.11-g WLAN scenario.

Keywords: CR; OFDM; FBMC; Filter bank; Spectrum sensing; Energy detector; Spectrum utilization; Loading algorithms;Multicarrier

1 IntroductionOne of the major challenges in cognitive radio (CR) op-eration is to utilize the available whitespace with min-imal interference to the primary or prioritized secondarytransmission systems [1]. Several spectrum sensing tech-niques have been proposed, e.g., in [2-5] to facilitate CRoperation. Especially, energy detector-based spectrumsensing algorithms have been widely considered due tolow computational complexity. On the other hand, thefading channel capacity has already been studied froman information theoretic perspective, e.g., in [6,7] interms of resource allocation. Recently, the secondary

* Correspondence: [email protected] of Electronics and Communications Engineering, TampereUniversity of Technology, P.O. Box 692, Tampere 33101, FinlandFull list of author information is available at the end of the article

© 2014 Dikmese et al.; licensee Springer. This isAttribution License (http://creativecommons.orin any medium, provided the original work is p

user (SU) capacity has been widely studied. The SUchannel capacity for additive white Gaussian noise(AWGN) channels under different power constraint isstudied in [8]. The effect of various types of fading chan-nels on the CR capacity has been studied in [9] underoptimal power allocation strategy for the CR and sub-jected to an interference power constraint at the co-existing primary. Further, [10] discusses the effects ofpeak power and average interference power constraintson the outage capacity. In [11], the ergodic capacity, thedelay-limited capacity, and the outage capacity of the CRin block-fading channels under spectrum sharing arediscussed.In this paper, we investigate two important features of

the cognitive radio. We begin with the spectrum sensingfunction and later study the spectrum utilization

an Open Access article distributed under the terms of the Creative Commonsg/licenses/by/2.0), which permits unrestricted use, distribution, and reproductionroperly credited.

mailto:[email protected]

http://creativecommons.org/licenses/by/2.0

Dikmese et al. EURASIP Journal on Advances in Signal Processing 2014, 2014:68 of 12http://asp.eurasipjournals.com/content/2014/1/68

function implementing optimized resource allocationunder power and interference constraints. Instead ofelaborate spectrum sensing techniques, such as cyclosta-tionary and eigenvalue-based methods [2,3], energydetector-based spectrum sensing is utilized. This is moti-vated by subband-based energy detector's capability toimplement the needed spectrum analysis functions foridentifying the available spectral slots and for estimatingthe signal-to-interference-plus-noise (SINR) ratios atsubcarrier level for resource allocation purposes.For a CR system, multicarrier modulation techniques

are generally better suited as they are spectrally more ef-ficient than single carrier systems and have the flexibilityto allocate resources to the available spectral gaps andamong different users to maximize system throughput.There are various ways of improving the spectral con-tainment of multicarrier waveforms, including methodsto suppress the strong side lobes of the orthogonalfrequency-division multiplexing (OFDM) spectrum [12-14].Filter bank multicarrier (FBMC) is another multicarriermodulation scheme which has significantly reducedspectrum leakage compared to the cyclic prefix-basedOFDM systems [15]. Also, the analysis filter bank (AFB)module of an FBMC receiver can be easily used forspectrum analysis purposes [15-22].This paper includes a brief summary of our earlier

studies concerning simple energy detection-based wide-band multichannel spectrum sensing techniques foridentifying the spectrum holes, considering the 2.4-GHzISM band as a case study. We apply an AFB-basedenergy detector, which averages the subband sample en-ergies. By this way, multiple center frequencies, band-widths, and multiple spectral gaps can be identifiedrapidly, efficiently, and flexibly for potential use by theCR. A similar fast Fourier transform (FFT)-based schemeis considered as a reference.At the resource allocation stage, the transmit power of

the subcarriers must be adjusted according to the chan-nel state information (CSI) and the location of subcar-riers with respect to the primary user's (PU) spectrum.In [23], an optimal and two sub-optimal power loadingalgorithms are developed. These algorithms use La-grange formulation which maximizes the downlink cap-acity of the CR keeping the interference to the primarytransmission below a threshold, without considering thetotal power constraint. In [23,24], the spectral hole andthe signal-to-noise (SNR) are fixed to simplify the model.In [25], a low-complexity suboptimal algorithm is pro-posed. The algorithm gives maximum power to eachsubcarrier based on the results from conventional waterfilling and then modifies these values by applying powerreduction algorithm in such a way that the interferenceconstraint is satisfied. In [23,25], the used signal modelsare closer to the ideal signal model, e.g., assuming fixed

spectral hole bandwidth, instead of a realistic systemmodel. In reality, the spectral hole bandwidth varies withthe SNR. A proper system model should include also apractical power amplifier model. Our study focuses onthe missing aspects of these studies.The optimal solution which maximizes the CR link

capacity under both transmission power and interferenceconstraints requires high computational complexity, andit is unsuitable for the practical applications. Low com-plexity algorithms are proposed in [25,26]. However, inthese methods, the interfered subcarriers are deactivatedwithout considering optimized power and bit loadingbased on each subcarrier's SINR. Such optimization canbe carried out using the power interference (PI) algo-rithm [27,28]. The resource allocation method utilizesthe results of spectrum sensing in an efficient way, sothere is interdependence between the spectrum sensingand spectrum allocation functions, which has not beenaddressed in earlier work. We study this interdepend-ence, focusing on its effects on efficient utilization of thesensed spectrum.The main contributions of this study are listed as

follows:

� We have generalized the study for realistic signalmodels which can be applied to any multicarrier CRsystem.
Until now, simplistic CP-OFDM signal models havebeen used as the PU and CR signal models forspectrum allocation algorithms [23-26,29-32]. Exceptfor [27,28], CP-OFDM has also been used for the CR.The primary knowledge we assume about the PUwaveform is its transmitted power spectral density(PSD) and the receiver selectivity mask; otherwise,there are no limitations regarding the PU signalmodel. In our case study, we select the PU waveformeither as CP-OFDM following the 802.11-g standardor an FBMC waveform with similar parameterization.Furthermore, a nonlinear transmitter power amplifiermodel (the so-called Rapp model [33]) is used for thePU system in order to obtain a realistic model for thePU spectrum. To the best of our knowledge, thisaspect has not been considered in earlier work. In thisway, we are able to quantify the effects of the PUspectral characteristics on the SU capacity. It is seenthat the nonlinear power amplifier-induced spectralleakage (regrowth) effect, which is present in anyradio communication system, has a significant impacton the SU capacity. As for the SU waveform, we havechosen the FBMC scheme for the case study becauseit has the sharpest spectrum, reaching the maximumspectral containment among the alternatives.However, generic multicarrier model is included inthe overall system model, and the analysis and


optimization methods are readily applicable for anymulticarrier waveform for the CR.Furthermore, previous studies on CR resourceallocation in [23-25,27-32] consider only flat fadingchannel models. However, the performance ofspectrum sensing and resource allocation is affectedsignificantly when frequency-selective fading channelis assumed. In our study, all the links within/betweenPU and SU systems associated with spectrumestimation and spectrum utilization are modeled asfrequency selective block fading channels.

� Combined spectrum sensing and resource allocationalgorithms for cognitive radios.There has been no previous work addressing thecombined spectrum sensing and resource allocationalgorithm in the literature. Especially, different typesof spectrum sensing algorithms have been appliedwithout considering any particular spectrumutilization techniques to make efficient use of theavailable spectral holes [1-5]. Similarly, resourceallocation algorithms have only been applied withoutany spectrum sensing information so far[27-32,34,35]. Constant number of availablesubbands has been considered in the spectral hole.However, the variation of the PUs' power levelaffects the actual number of available subbands, andthis depends critically on the spectral characteristicsof the PUs. Hence, spectrum sensing plays a crucialand enabling role for spectrum utilization process.The sensing function identifies the frequency bandwhich is considered for allocation, but it is alsoneeded for detecting possible other PU's starting tooperate in the spectral gap during the SU operation.For this purpose, we assume that there are gaps inthe CR transmission. In our study, efficientspectrum utilization methods are investigated and

Figure 1 System model for spectrum sharing in CR.

applied for maximizing the cognitive radio'sthroughput based on robust spectrum sensingresults. It turns out that the PI algorithm isapplicable in our scenario, with all the mentionedgeneralizations of the system model. The maincontribution of this study is evaluating the SUperformance with the combination of energydetection-based wideband sensing algorithm and thePI algorithm for spectrum utilization in a realisticcognitive radio scenario.

The rest of this paper is organized as follows. InSection 2, the signal models for the CR and the primarytransmission system, along with the mutual interferencemodel between the CR and primary are explained. Theproblem definition for this study is given in the samesection. In Section 3, FFT- and AFB-based widebandspectrum sensing is reviewed considering the spectrumanalysis aspects related to the multicarrier techniques.Section 4 develops the algorithms for spectrum alloca-tion. Section 5 gives the numeric and graphic results ob-tained through simulations. Finally, some concludingremarks are given about the performance of thesemethods, along with discussion of possible further stud-ies in this area.

2 Signal models and problem definitionAs shown in Figure 1, the CR system works in the sameband of frequencies with PU networks. Hence, there willbe some interference between different PUs and CRs.The PU and CR systems are assumed to use the time-division multiplexing/duplexing (TDMA/TDD) princi-ples, i.e., each system is using a fixed frequency slot forcommunications between all the stations. While the CRsystem has the capability to operate in other parts of theISM band, we focus on the situation where the CR


system has identified a spectrum opportunity in thementioned frequency slot and is initiating communica-tions in it. The primary purpose of the spectrum sensingfunction is to detect possible other transmissions orreappearing PUs in the spectrum gap. It is also assumedthat the stations of the CR systems have means to ex-change control information with each other, e.g., using acognitive control channel [36].

2.1 Signal model for PUIn this study, we focus on a specific spectrum use sce-nario with two active primary radio systems which areoperating in the 2.4-GHz ISM band, using either as802.11 g-based WLAN waveforms or 802.11 g-likeFBMC signals. The WLAN and FBMC spectra consid-ered here use third and eighth channels as illustrated inFigure 2. The signals do not overlap each other, and an8-MHz spectral hole is available between the two PUspectra. Both active signals are assumed to have thesame power level, normalized to 0 dB in our scenario.This means that the spectrum leakage effects on bothedges of the white space are equally critical for the CRsystem performance.The Rapp power amplifier (PA) nonlinearity model

[33] is considered as seen in Figure 2. Using the complexI/Q baseband model, the amplitude function at the PAoutput is given as follows:

gA ¼ κA

1þ κA=A0½ �2p� �1=2p ð1Þ

where A is the input amplitude, κ is the small signal gain,A0 is the saturated amplitude, and p is the amplitude

−60

−50

−40

−30

−20

−10

0

2400 2410 2420 2430 24

−60

−50

−40

−30

−20

−10

0

Frequ

Pow

er S

pect

rum

Mag

nitu

de (

dB)

N

W5

N

5

Modest−case 15 dB Backoff

Modest−case15 dB Backoff

WLAN 1

FBMC 1

Figure 2 Effects of the Rapp power amplifier model on (a) OFDM- and

smoothness factor of the transition from linear to satu-rated amplitude range. Three cases with respect to the PAnonlinearity are considered in this study. No regrowth isthe ideal case, and the Rapp PA nonlinearity with two dif-ferent back-off values of 15 dB (modest case) and 5 dB(worst case) is illustrated in Figure 2. Parameters of theRapp model have been chosen according to the practicalmodel for PU signals based on [37]. In our study, we useκ = 1, p = 3, and A0 = 1 as Rapp model simulationparameters.The 802.11-g-based WLAN signal specifications allow

the spectral regrowth in this scenario to be at the levelof about −20 dB, i.e., close to the worst case model. Weinvestigate how the CR system performance is affectedby improved spectral containment of the PU signalthrough enhanced multicarrier waveform and/or im-proved power amplifier linearity. These effects for bothsensing and utilization functions will be addressed in thestudy.

2.2 Signal model for cognitive radioIn this work, the CR waveform is chosen as FBMC dueto its high spectral containment. Offset quadrature amp-litude modulation (OQAM) is used for FBMC-basedCRs to achieve orthogonality of subcarriers, as discussedin [18,19,38]. In Figure 1, the channels h0 and h1 are thechannels from a cognitive transmitter to a primary re-ceiver and a cognitive receiver, respectively. Channels h2and h3 are from two different primary transmitters tothe cognitive radio receiver. The channel estimate for h1is made available by usual channel estimation procedureof the CR system. The knowledge about channel h0 canbe obtained through the channel reciprocity in TDD

40 2450 2460 2470 2480ency (MHz)

o Spectrum regrowth

orst −casedB Backoff

o Spectrum regrowth

Worst−case dB Backoff

(b)

WLAN 2

FBMC 2

(a)

(b) FBMC-based primary users' spectra.


operation. Here, the channel amplitude response is suffi-cient and the phase response is irrelevant. The ampli-tude responses of channels h2 and h3 are first obtainedthrough the spectrum sensing function of the CR deviceand later on, during secondary transmission, through theSINR estimation function of the CR device. The effectsof primary spectral dynamics at the edges of the whitespace play an important role both in spectrum sensingand in spectrum allocation. This dependency is capturedby the analytical models and revealed by the simulationsto be presented later.For FBMC/OQAM, a signal model with real valued

symbol sequence at twice the QAM symbol rate is ap-plied, instead of complex symbols. The synthesis filterbank (SFB) for transmitter and AFB for receiver is de-signed with this idea in mind. The FBMC-based trans-mitted signal can be expressed mathematically as

s nð Þ ¼Xk

Xl∈Z

ak;lg n−lτ0ð Þej2π k=Nð Þnejϕk;l ð2Þ

where {k} is the set of active subbands, l is the symbolindex belonging to the set of integers, g is the pulseshape (prototype filter impulse response), and ϕk,l is aphase term. The real symbol values, obtained alternat-ingly as real and imaginary parts of complex QAM sym-bols, are denoted as ak,l and τ0, respectively, is thecorresponding half-symbol duration. Both the real andimaginary parts of the QAM sequence have zero meanand equal variances σ2r ¼ σ2

I ¼ σ2s =2 . The PSD of the

FBMC based CR waveform can be written as

ϕFBMC fð Þ ¼ σ2r

τ0

Xk

G f −kN

� ��2

ð3Þ

where G is the frequency response of the prototype fil-ter with impulse response g(n) with n = 0, 1 …, L − 1.Here, L = KN is the prototype filter length and K isthe overlapping factor (length of each polyphase com-ponent) while N is the number of subbands. Theprototype coefficients have even symmetry around the(KN/2)th coefficient, and the first coefficient is zero[13]. In our study, the prototype filter of the FBMC-based CR is designed according to the PHYDYASmodel [38], with K = 4. Then its frequency responsecan then be expressed as

G fð Þj j ¼ g L=2½ � þ 2XL=2−1r¼1

g L=2−r½ � cos 2πfrð Þ ð4Þ

Also, the nonlinear PA model can be straightforwardlyincluded in the CR signal model and the interference

models developed below. However, in the numericalstudies of this paper, we consider an ideal FBMC wave-form for the CR as the focus is on the SU capacity andits dependency on the PU waveform characteristics.Generally, good spectral containment is regarded as oneof the key requirements for the CR transmitter. Detailedevaluation of the performance-complexity tradeoffs forthe CR implementation, including the PA linearity re-quirements, is a rather complicated issue and is left as atopic for future studies.

2.3 Definition of the interference problemAccording to the above scenario, the CR system coexistswith the primary transmission system in the same geo-graphical location. The CR transmitter causes someinterference to the primary transmission system, andsimilarly, the secondary transmission between two activePU spectra is exposed to some interference due to thePUs. The secondary transmission system uses multicar-rier transmission technique. There are Ngap subcarriersin the sensed spectrum hole and the subcarrier spacingis Δf. Since the transmitter and receiver are assumed tobe static or slowly moving, the effect of inter-carrierinterference (ICI) between subcarriers can be ignored.The primary and secondary transmission systems occupycontiguous frequency slots. The interference that the CRproduces to each of the primaries is required to be isless than the maximum interference that can be toler-ated by the primary, Ith. The spectral distance dPU of aPU is defined as the frequency separation from the DCsubcarrier of the CR to the center frequency of the PU(positive for a PU above the upper edge of the gap, nega-tive for a PU below the gap). The interference to the pri-mary transmission due to the kth CR subcarrier dependson the CR subcarrier powers Pk and dPU [20]. Fixing theorigin of the frequency axis at the DC subcarrier of theCR, the interference is given by the equation

Ik Pkð Þ ¼Z kΔfþB=2

kΔf −B=2H0 fð Þj j2PkΦ f −kΔfð ÞΨ f −dPUð Þdf ¼ PkΩk

ð5Þ

Here, H0(f ) is the channel frequency response be-tween the CR transmitter and a primary receiver. Φ(f )represents the subcarrier power spectral density of theunderlying multicarrier technique employed by the CR.Ψ(f ) denotes the PU sensitivity mask characterizingthe effects of the PU receiver filtering. B denotes theCR subcarrier bandwidth which is considered signifi-cant for the interference estimation. Finally, Ωk repre-sents the combined interference factor for the kth CRsubcarrier.

FFT

AFB

Time and FrequencyFilter

DecisionDevice

DecisionDevice

| . | 2

| . | 2 Time and FrequencyFilter

WLANFBMC

Figure 3 Block diagram of energy detector with AFB- andFFT-based spectrum analysis.


The signal-to-interference-plus-noise ratio due to inter-ference introduced by primary signal to the kth subcarrierat the receiving CR is given by

SINRk ¼Pk H1;k

�� 2σ2w þ

Z kΔfþB=2

kΔf −B=2H2 fð Þj j2Φ f −kΔfð ÞψPA f −dPUð Þdf

¼ Pk

σ2w þ J k

ð6Þwhere H2(f ) is the channel frequency response betweenthe primary transmitter and CR receiver. H1,k is thechannel gain between the CR transmitter and the CR re-ceiver at the frequency of kth subcarrier. This channelcan be assumed to be flat-fading at the subcarrier level.ΨPA(f ) is the power spectral density as seen at the outputof the PU's transmitter antenna. This depends on the PUtransmission power and its subcarrier spectrum, as wellas on the spectral regrowth due to the PU power ampli-fier. Φ(f ) is the CR receiver sensitivity mask characteriz-ing the CR receiver subband filtering effects. σ2w is thevariance of the additive white Gaussian noise.The power amplifier effects of the secondary transmis-

sion are not considered in our numerical study as theyplay no role in the spectrum sensing part and the effectof PA-related spectrum leakage on the interference to thePUs is expected to be relatively small. For this reason, thesame Φ(f ) function can be used in (5) for the CR sub-carrier spectrum and in (6) for the CR receiver sensitivitymask. However, the developed generic signal model allowsto include also the CR transmitter PA effects by using dif-ferent CR-related spectral functions in (5) and (6).The interference models of (5) and (6) assume certain

knowledge about PU characteristics and the channelsbetween PUs and CRs. Regarding the interference froman active PU transmitter to a CR receiver in (6), the jointeffect of transmitter power spectrum and the transmis-sion channel can be estimated by the receiving CR sta-tion by utilizing the spectrum sensing function. Thisinformation can be communicated through the controlchannel to the transmitting CR station for optimizingthe spectrum utilization. Regarding the channel from theCR transmitter to PU receiver, the knowledge would beavailable in a TDMA/TDD-based PU system (like aWLAN) based on channel reciprocity, if the PU trans-mission power is known. Of course, for a PU stationwhich is in idle mode over long periods, such informa-tion is not available.

3 Filter bank energy detector-based spectrumsensing algorithmsEnergy detector, which is also known as radiometer, isthe most common method of spectrum sensing due to

its low computational and implementation complexity[2-5]. Furthermore, it is more generic compared to mostof the other methods as it does not need any informa-tion about the PU waveform. Subband-based energy de-tection, using FFT or AFB for spectrum analysis, is inthe focus of this study. Basically, the energy of the re-ceived signal is compared with a threshold value whichis calculated according to noise variance and desiredfalse alarm probability in detecting spectral holes.A block diagram of alternative FFT- and AFB-based

spectrum sensing algorithms is shown in Figure 3. Thesubband sampling rate is equal to the ADC samplingrate divided by the number of FFT/AFB frequency bins.With subband-wise spectrum sensing method, the sub-band signals can be expressed as [3]

Y m; kð Þ ¼ W m; kð Þ H0

S m; kð ÞHk þW m; kð Þ H1

� �ð7Þ

where S(m, k) is the transmitted WLAN or FBMC basedPU signals as seen in subband k during the mth symbolinterval with zero mean and variance σ2PU . When thereare no PU signals (hypothesis H0), the noise samples W(m, k) are modeled as AWGN with zero-mean and vari-ance σ2

w . When a PU signal is present (hypothesis H1),the WLAN- and FBMC-based PU signals can also bemodeled as zero-mean Gaussian distribution with varianceσ2PU;k þ σ2w.Time and frequency averaging techniques can be ap-

plied to obtain more reliable decision statistic [3]. Thedecision statistics at different frequencies can be ob-tained with this idea as follows in [39]:

~T m; kð Þ ¼ 1LtLf

Xkþ⌈Lf =2⌉−1

l¼k−⌊Lf =2⌋

Xm

u¼m−Ltþ1Y u; lð Þj j

2

ð8ÞIn this formula, Lf and Lt are the window lengths in

frequency and time domain averaging, respectively. Theoutput of ~T m; kð Þ is passed to a decision device to deter-mine the possible occupancy of the corresponding fre-quency band at the corresponding time interval. Thewindow length in frequency direction is selected basedon the expected minimum bandwidth of the PU signalor spectrum hole, and then the required time domainwindow length can be calculated from the target false

Spectrum

Sensing

Power Scaling (AFB)

Bit Rate Distr.

SNR & Spec.

edges (FFT) SNR & Spec.

edges (AFB)

Power Scaling (FFT)

Spec.Utili.

Spec.Utili.

Bit Rate Distr.

CR power

Figure 4 Block diagram of spectrum utilization with alternativeAFB- and FFT-based spectrum analysis schemes.


alarm and missed detection probabilities. The basic ap-proach would be to calculate (8) for a nonoverlappingset of windows. However, using a sliding window in fre-quency and/or time direction can also be done with rela-tively small addition to complexity. Time-domain slidingwindow helps to detect rapidly re-appearing PU [4,5]whereas sliding window in the frequency direction helpsto locate spectrum gaps with unknown center frequen-cies. Due to the Gaussian distribution of Y(m, k), theprobability distribution function (PDF) of ~T m; kð Þ be-comes approximately Gaussian under H0 and H1. [3].Using Gaussian approximation, it is straightforward to

model the effect of PU transmitter's spectral leakage onthe actual false alarm probability ~PFA as

~PFA kð Þ ¼ Qλ− σ2w þ Iadj kð Þ� �ffiffiffiffiffiffiffi1

LtLf

qσ2w þ Iadj kð Þ� �

0B@

1CA ð9Þ

Here,

Iadj kð Þ ¼Z f 2

f 1

H2 fð Þj j2ψPA fð Þdf ð10Þ

is the leakage power from the adjacent PU transmitter tothe sensing frequency band between frequencies f1 andf2. H2(f ) is the channel frequency response from aprimary transmitter the CR receiver. In (9), λ is the deci-sion threshold value which is calculated using a well-known analytical model from the noise variance estimateand target false alarm probability PFA.The detection probability PD and threshold value λ

can also be expressed as follows:

PD kð Þ ¼ Qλ− σ2w þ Iadj kð Þ� �þ σ2PU;k

�ffiffiffiffiffiffiffi1

LtLf

qσ2w þ Iadj kð Þ� �þ σ2

PU;k

�0B@

1CA ð11Þ

λ ¼ Q−1 PFA kð Þð Þffiffiffiffiffiffiffiffiffi1

LtLf

sσ2w þ Iadj kð Þ� �þ σ2w þ Iadj kð Þ� �

ð12ÞIn principle, if there is a reliable estimate of the PU

transmission power and reliable knowledge about itsspectrum shape, then the above analysis could be usedfor improving the spectrum sensing at the frequenciesaffected by the spectrum leakage. However, this wouldbe very challenging in practice due to the unpredictabil-ity of the PA characteristics, and the above model is usedonly for the purpose of performance analysis.For different PU SNR values, different number of empty

subbands, Ngap, will be detected due to the PU spectralleakage effects and statistical nature of the spectrum sens-ing process. The expression (9) can be used for evaluating

the false alarm probability for different sensing band-widths in different parts of the spectrum gap.The spectrum sensing function identifies groups of Lf

subbands which are deemed to be available for second-ary transmissions. In the following stage, the spectrumutilization function is employed to perform power andbit allocation to those subcarriers.

4 Spectrum utilizationAfter nonactive spectrum has been identified, spectrumutilization becomes an important consideration, whenconsidering the overall efficiency of the CR system. Thenumber of available nonactive subbands is the output ofthe sensing algorithm, along with information about thenonactive band edges.In the multicarrier case, the rate at which transmission

can take place is given by Shannon's capacity

RCR ¼XNgap

k¼1

Δf log2 1þ Pk

σ2k

� �ð13Þ

σ2k ¼ σ2w þXNPU

i¼1

Jk;i ð14Þ

where Jk,i is the effective interference power contributedby ith primary to the kth CR subcarrier as given by (6).NPU is the number of PU's contributing to the interfer-ence at the receiving CR station. In our case study, NPU =2, i.e., there is one PU adjacent to the lower and upperedges of the white space. The model could be simplifiedby assuming that these PUs affect only the lower andupper half of the subcarriers, respectively. Pk is the trans-mit power used by the CR for subcarrier k. It is assumedthat the channel changes slowly so that the channel gains,and consequently Jk,i, will be approximately the same dur-ing each transmission frame. Further, there is no ICI inthe CR reception due to low mobility. The main objectivehere is to maximize the capacity as given in (13).The block diagram of spectrum utilization is shown in

Figure 4. As seen in this block diagram, knowledgewhich comes from sensing part is passed to spectrumutilization part to obtain better capacity.


The optimization problem can be formulated as fol-lows [27]:

RCR ¼ maxPkf g

XNgap

k¼1

Δf log2 1þ Pk

σ2k

� �ð15Þ

Subject to

XNgap

k¼1

Pk ≤ PT ð16Þ

XNgap

k¼1

PkΩk ≤ Ith ð17Þ

Pk ≥ 0; ∀k ∈ 1; 2;…;Ngap� ð18Þ

This is a convex optimization problem, and the La-grangian can be written as

Gsnr ¼XNgap

k¼1

Δf log2 1þ Pk

σ2k

� �

−λ0XNgap

k¼1

Pk−PTð Þ−λ1 PkΩk−I thð Þ þ λ2XNgap

k¼1

Pk

ð19Þ

The Karush-Kuhn-Tucker (KKT) conditions [27] canbe written as

Pk ≥ 0; ∀k ∈ 1; 2; ::::;Ngap�

; ð20Þλj ≥ 0; ∀j ∈ 0; 1; 2f g ð21ÞXNgap

k¼1

Pk−PTð Þ ≥ 0 ð22Þ

λ1 PkΩk−I thð Þ ≥ 0 ð23Þ

λ2XNgap

k¼1

Pk ≥ 0 ð24Þ

The optimal solution to the problem above as given in[27] is as follows:

Pk ¼ 1λ0Ωk þ λ1

−σ2khkj j

� �þð25Þ

where [y]+ = max(0, y). The optimal solution has highcomputationally complexity; hence, a lower complexityalgorithm called the PI algorithm which divides theproblem into stages has been developed [27]. First, theinterference constraint is ignored, keeping only the totalpower constraint and this leads to the classical water fill-ing solution

Pk0 ¼ γ−

σ2k

hkj j� �þ

ð26Þ

where γ is the water filling level. When the total poweris ignored the solution [27] becomes

Pk0 ¼ 1

λ00Ωk

−σ2khkj j

� �þð27Þ

The value of λ'0 can be obtained by substituting (27)

into the constraintXNgap

k¼1

Pk0Ωk ¼ I th to get

λ00 ¼

Ngap;l

�� INgap

th þ ðXi

Ωkσ2k= hkj j2Þ

ð28Þ

The above solution is optimal only when the totalpower is greater than or equal to the power under theinterference constraint. Mostly, in practice, this condi-tion is not true and this is the motivation for the PI al-gorithm. Detailed discussion and its comparison tovarious other algorithms for spectrum utilization areavailable in [27].In this study, the PI algorithm is found to be directly

applicable in case of the developed greatly enhanced sys-tem model for the secondary usage scenario. PI algo-rithm has four stages: maximum power determination,power constraint, power budget distribution, and powerlevel re-adjustment [27].

5 Simulation resultsIn our test scenario, the CR's spectrum sensing functionhas identified a potential spectral gap between two rela-tively strong PUs, as illustrated in Figure 2. We shouldalso consider the possibility that there is another, rela-tively weak PU signal, using one of the WLAN channels4…7, and fully or partly occupying the gap betweenchannels 3 and 8. Thus, one purpose of spectrum sens-ing is to secure that there are no other PUs active in theconsidered gap. We assume that there are no additionalsignals within the spectral gap, but the spectrum sensingmakes anyway false alarms. Especially close to the edgesof the gap, the spectrum leakage from the PUs raises thefalse alarm probability. This effect depends on the powerlevel (SNR) of the PUs. In our case study, the spectrumsensing and CR transmissions use a smaller subbandspacing of 81.5 kHz, instead of the 325-kHz subcarrierspacing of WLANs, in order to reduce the effects of fre-quency selective channels. Targeting at −5 dB SNR inspectrum sensing, false alarm probability of 0.1, and de-tection probability of 90%, the required sample complex-ity is around 250 complex samples. The time andfrequency averaging lengths are chosen as 50 and 5, re-spectively. The spectral hole starts from the side lobes ofWLAN 1 signal and ends at the side lobes of WLAN 2spectrum. The available number of subbands/bandwidth


of the spectrum is obtained after subband-based energydetection, using FFT or AFB for spectrum analysis.Then, the initial SINR estimation and spectrum allocationis done based on the sensing results. Later on, the SINRestimates are updated during SU system operation to trackthe changing radio environment under frequency-selectivefading channel conditions. It is assumed that the spectrumsensing is done in regular intervals during gaps in the CRtransmission and this helps in detecting reappearing PUsignals in the spectral gap.It should be noticed that in the considered scenario,

there is no way for the CR system to determine the use-ful received power level at the PU receiver. Therefore,we choose the interference threshold to be 6 dB belowthe thermal noise level, in order not to introduce signifi-cant performance loss in case the primary receiver is op-erating close to the sensitivity level (i.e., minimumreceived power level expected to be detectable). To de-termine the threshold value, we assume a simplified sce-nario, where the path losses of channels h0 and h1 arenormalized to 1, i.e., the average power gains of channelsh0 and h1, denoted as G0 and G1, are equal to one. Fur-ther, we assume that the average SNR of the CR receiveris 10 dB. Then, the interference threshold is −16 dB inreference to the total CR transmission power PT, or Ith =PT/40. More generally, relaxing the normalization of h0and h1, this can be expressed as Ith =G1PT/(40G0).The bandwidth of the detected spectral hole is shown

in Figure 5 as a function of the average PU SNR at the

0

2

4

6

8

Ban

dwid

th o

f spe

ctra

l hol

e (M

Hz)

0

2

4

6

8

−20 −10 0 10 20

2

4

6

8

Prima

(b)

(c)

(a)

Figure 5 Average bandwidth of the detected spectral hole with targeselective channel model for (a) ideal model, (b) Rapp PA with 15 dB back-case. Different combinations of FFT/AFB-based sensing and OFDM/FBMC b

CR RX. The spectral leakage due to primary users' PAnonlinearity is affecting significantly on the width of thedetected hole. In this respect, we consider three differentcases, as explained in Subsection 2.1: ideal PA, modestPA nonlinearity with 15 dB back-off, and worst casenonlinearity with 5 dB back-off. All PU and CR channelsh0, h1, h2, and h3 use frequency-selective channel modelswith 90 ns delay spread and 16 taps [40]. We considerthe combinations of two PU waveforms, CP-OFDM- andFBMC-based WLANs, as well as two spectrum sensingtechniques, based on FFT or AFB. The CR waveform isalways FBMC.From Figure 5, it can be easily seen that AFB-based

spectrum sensing is able to detect the unoccupiedspectrum close to strong primary users much betterthan FFT-based sensing. FBMC-based transmission re-sults in much better spectral containment, which can beeffectively utilized by AFB-based sensing. However, evenwith relatively modest power amplifier nonlinearity, thisbenefit of FBMC waveform is compromised.In Figure 6, the actual false alarm probability within

the spectral hole is shown as a function of the activePU's SNR for different levels of spectral regrowth. Theresults indicate the probability of the 5 subband groupsto be detected to be occupied.The efficiency of the utilization of the 8-MHz white

space by SU in between two active PUs is shown inFigure 7 versus the PU SNR. In this figure, perfect CSI isconsidered for the CR channel, both for CR channel h1

0 30 40 50 60ry User SNR

FFT, OFDMFFT, FBMCAFB, OFDMAFB, FBMC

t PFA = 0.1. Using sample complexity of 250 samples under frequency-off as the modest case, (c) Rapp PA with 5 dB back-off as the worstased PU waveforms.

10−1

100

10−1

100

Fal

se A

larm

Pro

babi

lity

−20 −10 0 10 20 30 40 50 6010

−1

100

Primary User SNR


(a)

(b)

(c)

Figure 6 Actual false alarm probability for target PFA = 0.1. Time record length of 50 and sensing bandwidth of 5 subbands under frequencyselective channel model for (a) ideal model, (b) Rapp PA with 15 dB backoff as modest case, (c) Rapp PA with 5 dB backoff as worst case.Different combinations of FFT/AFB based sensing and OFDM/FBMC based PU waveforms.


equalization and in the PI algorithm for resource alloca-tion. Perfect knowledge of the amplitude response ofchannel h0 is also assumed, while channels h2 and h3 areknown from spectrum sensing results. The subband-wise noise + interference estimates are obtained using

0

1

2

3

4

Cap

acity

(B

its/s

/Hz)

−20 −10 0 10 20

1

2

3

4

Primary

0

1

2

3

4

(a)

(b)

(c)

Figure 7 Capacity of a CR in a spectral gap between two PUs versus P(b) Rapp PA with 15 dB backoff as the modest case, (c) Rapp PA with 5 dBsensing and OFDM/FBMC-based PU waveforms.

time filtering length of 50. According to FFT- or AFB-based spectrum sensing results, a number of subbandsare left empty in the spectrum utilization phase. Thepower of these occupied subchannels is reallocated tothe other subbands that can be used by the CR. The

0 30 40 50 60

User SNR


U SNR. PI algorithm used for power allocation (a) Ideal model,backoff as the worst case. Different combinations of FFT/AFB-based


power allocation is done utilizing the PI algorithm, andthe resulting capacity, in terms of bits/s/Hz, is shown inFigure 7. The limitations of FFT-based spectrum sensingcan again be clearly seen from the results, whereas AFB-based sensing is able to identify gaps between relativelystrong PU signals. Regarding the transmission waveform,FBMC has clear benefit due to better spectral contain-ment, if the effects of power amplifier nonlinearities canbe kept at a modest level. As expected, the capacity isproportional to the available bandwidth in the gap, i.e.,there is a clear connection between the results ofFigures 5 and 6. The resource allocation algorithm (i)optimizes the performance with frequency selectivechannels in the presence of spectral leakage from thestrong PUs and (ii) secures that the interference leakagefrom the CR transmission to the primaries is at an ac-ceptable level.

6 ConclusionsWe have studied the effects of combined spectrum sens-ing and spectrum utilization for FBMC-based cognitiveradios with realistic signal model under frequency-selective fading channel conditions. Firstly, the perform-ance of energy detection-based spectrum sensingtechnique was analyzed using both the FFT and filterbank-based spectrum analysis methods for both WLANand FBMC signal models. Then, the utilization of dy-namically identified spectral holes with spectrum alloca-tion algorithms, subject to power and interferenceconstraints, was investigated. Through this study, the ef-fect of PU waveform's spectral containment on the CRtransmission capacity was revealed. Here, we consideredthe choice between OFDM and FBMC primaries, to-gether with the effect of spectral regrowth due to poweramplifier nonlinearity.In terms of the spectrum sensing performance, AFB

has clear benefits due to much better spectral contain-ment of the subbands. One important benefit of FBMCas a transmission technique in CR systems is that it canutilize narrow spectral gaps in an effective and flexibleway, even in the presence of strong primaries at the adja-cent spectral slots. This is due to the excellent spectralcontainment properties of the FBMC system. Additionally,an FBMC receiver can use the AFB for high-performancespectrum sensing with no additional complexity.The utilization of the sensed spectrum can be opti-

mized by using proper spectrum allocation algorithms.The PI algorithm has relatively low complexity, and itimproves the capacity of the CR system as compared tothe simple water filling-based spectrum allocation. Oneof the main observations of this work was that the PI al-gorithm can be directly utilized with the developedhighly enhanced and realistic CR system model. The sys-tem model accommodates frequency-selective channel

models for all the associated transmission links betweenPUs and SUs, as well as arbitrary transmitted powerspectra and receiver frequency responses. Because of theabove features, a FBMC-based CR system achieveshigher capacity in comparison with traditional WLAN-based system. This increase in capacity can be attributedto the efficient use of the available spectrum and verysmall interference introduced to the primary transmis-sions at adjacent frequencies.One of the important aims of this study was to under-

stand the interdependence of the spectrum sensing andthe spectrum utilization parts. It can be seen that in-creased false alarm probability has a direct effect on theavailable spectrum, and hence, it heavily influences thespectrum utilization. The PU power amplifier nonlinearityinfluences the sensed secondary spectrum introducingfalse alarms, hence lowering the CR system's spectrumutilization. It was demonstrated that, with heavy poweramplifier nonlinearity, the FBMC-based primary is nobetter than the OFDM primary in what comes to theavailable capacity for secondary usage in the nearbyfrequencies.In the numerical studies of this paper, we considered

an ideal FBMC waveform for the CR, without consider-ing the PA nonlinearity effects, since the focus is on theSU capacity and it dependency on the PU signal character-istics. Generally, good spectral containment is regarded asone of the key requirements for the CR transmitter. Also,the nonlinear PA model can be straightforwardly includedin the developed interference models. Detailed evaluationof the performance-complexity tradeoffs for the CR imple-mentation, including the PA linearity requirements, is arather complicated issue and is left as a topic for futurestudies.

Competing interestsThe authors declare that they have no competing interests.

AcknowledgementsThis work was partially supported by Tekniikan Edistämissäätiö (TES), GETAGraduate School, the European Commission under Project EMPhAtiC (FP7-ICT-2012-09-318362), COST Action IC0902, Tekes (the Finnish FundingAgency for Technology and Innovation) under the project ENCOR in the TrialProgram.

Author details1Department of Electronics and Communications Engineering, TampereUniversity of Technology, P.O. Box 692, Tampere 33101, Finland. 2CentreTecnològic de Telecomunicacions de Catalunya (CTTC), Parc Mediterrani dela Tecnología (PMT), Avenida Carl Friedrich Gauss 7, 08860 Barcelona, Spain.3Signal, Communication & Embedded Electronics Research Group, EcoleSupérieur D’Électricité –SUPELEC University, F-35576 Cesson-Sévigné Cedex,Gif-sur-Yvette C.S. 47601, France.

Received: 10 December 2013 Accepted: 15 April 2014Published: 12 May 2014

References1. J Mitola, JGQ Maguire, Cognitive radio: making software radios more

personal. IEEE Pers Commun Mag 6(4), 13–18 (1999)


2. Y Zeng, YC Liang, AT Hoang, R Zhang, A review on spectrum sensing forcognitive radio: challenges and solutions. EURASIP J Adv Signal Process2010, 1–15 (2010)

3. T Yucek, H Arslan, A survey of spectrum sensing algorithms for cognitiveradio applications. IEEE Commun Surv Tutorials 11(1), 116–130 (2009)

4. S. Dikmese, M. Renfors, H. Dincer, FFT and filter bank based spectrum sensingfor WLAN signals, in Proc (European Conference on Circuit Theory and Design(ECCTD’11) (Linkoping, Sweden, 2011)

5. S Dikmese, M Renfors, Optimized FFT and filter bank based spectrum sensingfor bluetooth signal, in Proc (Wireless Communications and NetworkingConference (WCNC’12) (France, Paris, 2012)

6. AJ Goldsmith, PP Varaiya, Capacity of fading channels with channel sideinformation. IEEE Trans Inf Theory 43(6), 1986–1992 (1997)

7. E Biglieri, J Proakis, S Shamai, Fading channels: information-theoretic andcommunications aspects. IEEE Trans Inf Theory 44(6), (1998)

8. M Gastpar, On capacity under received-signal constraints, in Proc (42nd AnnualAllerton Conference Communication (Control Comput, Monticello, USA, 2004)

9. A Ghasemi, E Sousa, Fundamental limits of spectrum-sharing in fadingenvironments. IEEE Trans Wirel Commun 6(2), 649–658 (2007)

10. L Musavian, S Aissa, Ergodic and outage capacities of spectrum-sharingsystems in fading channels, in Proc (IEEE Global TelecommunicationsConference (GLOBECOM’07), Washington (DC, USA, 2007)

11. X Kang, YC Liang, A Nallanathan, HK Garg, R Zhang, Optimal powerallocation for fading channels in cognitive radio networks: ergodic capacityand outage capacity. IEEE Trans Wirel Commun 8(2), 940–950 (2009)

12. A Sahin, H Arslan, Edge windowing for OFDM based systems. Commun LettIEEE 15(11), 1208–1211 (2011)

13. S Brandes, I Cosovic, M Schnell, Sidelobe suppression in OFDM systems byinsertion of cancellation carriers’, in Proc (Vehicular Technology Conference,(VTC’05-Fall) (Dallas, USA, 2005)

14. A Loulou, M Renfors, ‘Effective schemes for OFDM sidelobe control infragmented spectrum use, in Proc (IEEE 24th International Symposium onPersonal, Indoor and Mobile Radio Communications (PIMR’13) (UnitedKingdom, London, 2013)

15. BF Boroujeny, R Kempter, Multicarrier communication techniques forspectrum sensing and communication in cognitive radios. IEEE CommunMag (Spec Issue Cogn Radios Dynamic Spectr Access) 48(4), 80–85 (2008)

16. A Amini, R Kempter, L Lin, B Farhang-Boroujeny, Filter bank multitone: acandidate for physical layer of cognitive radio, in Proc (Software Defined RadioTechnical Conference and Product Exhibition (SDR '05) (Orange County, USA,2005)

17. A Amini, R Kempter, B Farhang-Boroujeny, A comparison of alternativefilterbank multicarrier methods in cognitive radios, in Proc (Software Defined RadioTechnical Conference and Product Exhibition (SDR '06) (Orlando, USA, 2006)

18. M Bellanger, T Ihalainen, M Renfors, Filter bank based cognitive radio physicallayer, in Proc (Future Network & Mobile Summit, Santander, Spain, 2009)

19. T Ihalainen, A Viholainen, TH Stitz, M Renfors, Spectrum monitoring schemefor filter bank based cognitive radios, in Proc (Future Network & MobileSummit, Florence, Italy, 2010)

20. S Srinivasan, S Dikmese, M Renfors, Spectrum sensing and spectrum utilizationmodel for OFDM and FBMC based cognitive radios, in Proc (Signal ProcessingAdvances in Wireless Communications (SPAWC’12) (Cesme, Turkey, 2012)

21. V Ringset, H Rustad, F Schaich, J Vandermot, M Najar, Performance of afilterbank multicarrier (FBMC) physical layer in the WiMAX context, in Proc(Future Network & Mobile Summit, Florence, Italy, 2010)

22. S Dikmese, S Srinivasan, M Renfors, FFT and filter bank based spectrumsensing and spectrum utilization for cognitive radios, in Proc (InternationalSymposium on Communications, Control, and Signal Processing (ISCCSP’12)(Rome, Italy, 2012)

23. G Bansal, MJ Hossain, VK Bhargava, Adaptive power loading for OFDM-basedcognitive radio systems, in Proc (IEEE International Conference on Communications(ICC '07) (Glasgow, UK, 2007)

24. G Bansal, MJ Hossain, VK Bhargava, Optimal and suboptimal powerallocation schemes for OFDM-based cognitive radio systems. IEEE TransWirel Commun 7(11), 4710–4718 (2008)

25. T Qin, C Leung, Fair adaptive resource allocation for multiuser OFDM cognitiveradio systems, in Proc (2nd International Conference on Communications andNetworking in China (ChinaCom '07) (Shanghai, China, 2007)

26. Y Zhang, Resource allocation for OFDM-based cognitive radio systems(University of British Columbia, Vancouver, Canada, 2008). Dec, Ph.D.dissertation

27. M Shaat, F Bader, Computationally efficient power allocation algorithm inmulticarrier-based cognitive radio networks: OFDM and FBMC systems.EURASIP J Adv Signal Process 13, (2010). doi:Article ID 528378

28. M Shaat, F Bader, Power allocation with interference constraint in multicarrierbased cognitive radio systems, in Proc (7th International Workshop on Multi-Carrier Systems and Solutions (MCSS '09) (Herrsching, Germany, 2009)

29. J Jang, KB Lee, Transmit power adaptation for multiuser OFDM systems. IEEEJ Selected Areas Commun 21(2), 171–178 (2003)

30. D Kivanc, G Li, H Liu, Computationally efficient bandwidth allocation andpower control for OFDMA. IEEE Trans Wirel Commun 2(6), 1150–1158 (2003)

31. Z Shen, JG Andrews, BL Evans, Optimal power allocation in multiuser OFDMsystems, in Proc (IEEE Global Telecommunications Conference (GLOBECOM '03)(San Francisco, USA, 2003)

32. CY Wong, RS Cheng, KB Letaief, RD Murch, Multiuser OFDM with adaptivesubcarrier, bit, and power allocation. IEEE J Selected Areas Commun17(10), 1747–1758 (1999)

33. C Rapp, Effects of the HPA-nonlinearity on 4-DPSK OFDM signal for a digitalsound broadcasting system, in Proc Conf. Rec. ECSC’91 (Luettich, Germany, 1991)

34. J Cioffi, Digital Communication: Signal Processing (Standford, California, USA,2000)

35. Q Lu, T Peng, W Wang, Efficient multiuser water-filling algorithm underinterference temperature constraints in OFDMA-based cognitive radionetworks, in Proc IEEE International Symposium Microwave, Antenna,Propagation, and EMC Technologies for Wireless Communications (MAPE’07,Hangzhou, China, 2007)

36. V Stavroulaki, K Tsagkaris, P Demestichas, J Gebert, M Mueck, A Schmidt, RFerrus, O Sallent, M Filo, C Mouton, L Rakotoharison, Cognitive controlchannels: from concept to identification of implementation options. IEEECommun Mag 50(7), 96–108 (2012)

37. AA Eltholth, AR Mekhail, A Elshirbini, MI Dessouki, AI Abdelfattah, Modelingthe effect of clipping and power amplifier non-linearities on OFDM systems.Ubiquitous Comput Commun J 3(1), 54–59 (2009)

38. PHYDYAS, Physical layer for dynamic spectrum access and cognitive radio.http://www.ict-phydyas.org/. Accessed 5 June 2013

39. T Yucek, H Arslan, Spectrum characterization for opportunistic cognitive radiosystems, in Proc (IEEE Military Communication Conference, Washington, D.C.,USA, 2006)

40. R Jain, Channel Models, A Tutorial (2007). http://www.cse.wustl.edu/~jain/cse574-08/ftp/channel_model_tutorial.pdf. Accessed 5 July 2013

doi:10.1186/1687-6180-2014-68Cite this article as: Dikmese et al.: Spectrum sensing and resourceallocation for multicarrier cognitive radio systems under interferenceand power constraints. EURASIP Journal on Advances in Signal Processing2014 2014:68.

Submit your manuscript to a journal and benefi t from:

7 Convenient online submission

7 Rigorous peer review

7 Immediate publication on acceptance

7 Open access: articles freely available online

7 High visibility within the fi eld

7 Retaining the copyright to your article

Submit your next manuscript at 7 springeropen.com

http://www.ict-phydyas.org/

http://www.cse.wustl.edu/~jain/cse574-08/ftp/channel_model_tutorial.pdf

http://www.cse.wustl.edu/~jain/cse574-08/ftp/channel_model_tutorial.pdf

RESEARCH Open Access Spectrum sensing and resource ... · rithm [27,28]. The resource allocation...

Documents

Transcript of RESEARCH Open Access Spectrum sensing and resource ... · rithm [27,28]. The resource allocation...