Inter national J our nal of Electrical and Computer Engineering (IJECE) V ol. 7, No. 4, August 2017, pp. 1934 1940 ISSN: 2088-8708 1934       I ns t it u t e  o f  A d v a nce d  Eng ine e r i ng  a nd  S cie nce   w     w     w       i                       l       c       m     P erf ormance Ev aluation of Ener gy Detector Based Spectrum Sensing f or Cogniti v e Radio using NI USRP-2930 F . Z. El Bahi, H. Ghennioui, and M. Zouak Laboratoire Signaux, Syst ` emes et Composants (LSSC), F acult ´ e des Sciences et T echniques de F ` es (FSTF) Uni v ersit ´ e Sidi Mohamed Ben Abdellah (USMB A), Route Immouzzer , B.P . 2202, F ` es-Maroc Article Inf o Article history: Recei v ed: Feb 18, 2017 Re vised: May 28, 2017 Accepted: Jun 12, 2017 K eyw ord: Ener gy Detector Spectrum Sensing Cogniti v e Radio Primary User Secondary User USRP MA TLAB ABSTRA CT This paper presents the performance e v aluation of the Ener gy Detector technique, which is one of the most popular Spectrum Sensing (SS) technique for Cogniti v e Radio (CR). SS is the ability to detect the presence of a Primary User (PU) (i.e. licensed user) in order to allo w a Secondary User (SU) (i.e unlice nsed user) to access PU’ s frequenc y band using CR, so that the a v ailable frequenc y bands can be used ef ficiently . W e used for implementation an Uni v ersal Softw are Radio Peripheral (USRP), which is the most used Softw are Defined Radio (SDR) de vice for research in wireless communications. Experimental measurements sho w that the Ener gy Detector can obtain good performances in lo w Signal to Noise Ratio (SNR) v alues. Furthermore, computer simulations using MA TLAB are closer to those of USRP measurements. Copyright c 2017 Institute of Advanced Engineering and Science . All rights r eserved. Corresponding A uthor: F atima Zahra EL B AHI, Laboratoire Signaux, Syst ` emes et Composants (LSSC), F acult ´ e des Sciences et T echniques de F ` es (FSTF), Uni v ersit ´ e Sidi Mohamed Ben Abdellah (USMB A), Route Immouzzer , B.P . 2202, F ` es-Maroc. Email: f atimazahra.elbahi@usmba.ac.ma 1. INTR ODUCTION In recent years, the wireless applications and de vices ha v e de v eloped and increased rapidly . Since the access to electromagnetic spectrum is fix ed and limited, thus the a v ailable frequenc y bands are inef ficiently utilized which causes the spectrum scarcity problem [1]. In order to use the a v ailable spectrum ef ficiently , man y studies and re- searches ha v e propo s ed a ne w concept called Cogniti v e Radio (CR). This concept is based on the opportunistic usage of radio frequenc y bands by allo wing Secondary Users (SUs, i.e. unlicensed users) to e xploit frequenc y bands of Primary Users (PUs, i.e. licensed users). CR is a radio for wireless communications able to change its parameters, related to either transmission or reception, autonomously and dynamically based on the electromagnetic en vironment and communication requirements, in order to perform an ef ficient communication without interfering with PUs [2]. CR has the capability of sensing the spectrum, thus detect the presence PU’ s signal, using dif ferent spectrum sensing techniques. These techniques are subdi vided into tw o cate gories: cooperati v e and non-cooperati v e sensing techniques. The first cate gory is based on sharing information, in other w ords, the detection of Primary User’ s signal is performed by combining results from multiple cogniti v e radios that w orks together [3]. The most important adv antage of this technique is the capability of decreasing sensing time and impro ving the sensing accurac y . The second cate gory , non-cooperati v e sensing techniques, is also kno wn as primary transmitter detection, because the detection of PU’ s signal is based only on the recei v ed signal at a SU. The most common sensing techniques that belong to this cate gory are: ener gy detector [4], matched filter detector and c yclostationary detector [5]. Ener gy detector is the most popular spectrum sensing technique, it can also be considered as the most spectrum sensing technique used in practice because of its lo w implementation comple xity . The concept of Ener gy detector is based only on computing the total ener gy of the recei v ed signal, then comparing is to a specified threshold in order to decide the presence or absence of a PU’ s signal, thus no prior kno wledge of the PU’ s signal is required, only the v ariance of the noise is needed. In this paper , we present an implementation of the Ener gy Detector Based Spectrum Sensing, using an Uni v er - sal Softw are Radio Peripheral (USRP), a Softw are Defined Radio (SDR) de vice, in order to e v aluate the performance J ournal Homepage: http://iaesjournal.com/online/inde x.php/IJECE       I ns t it u t e  o f  A d v a nce d  Eng ine e r i ng  a nd  S cie nce   w     w     w       i                       l       c       m     DOI:  10.11591/ijece.v7i4.pp1934-1940 Evaluation Warning : The document was created with Spire.PDF for Python.
IJECE ISSN: 2088-8708 1935 of the spectrum sensing technique. W e used for the implementation an NI USRP-2930, which belongs to the USRP se- ries of the National Instruments’ brand. F or the programming, we used LabVIEW and MA TLAB for both transmission and reception. W e implemented for transmission an OFDM (Orthogonal Frequenc y Di vision Multiple xing) Signal, which is used as a Primary User’ s signal. W e chooses the OFDM modulation because it is the most used in wireless communication, due to its high bandwidth ef ficienc y [6], such as W iF i [7], 3GPP/L TE [8] (3rd Generation P artnership Project/Long T erm Ev olution) f o r the do wnlink, D VBT [9] (Digital V ideo Broadcast T errestrial) and W iMAX [10] (W orldwide Interoperability for Micro w a v e Access). The performance e v aluation of the Ener gy Detector is based on e xperimental measurements obtained using USRP’ s recei v er , which included the Ener gy Detector algorithm. The rest of this paper is or g anized as follo ws. In section 2, we introduce the Ener gy Detector based sensing after defining the problem formulation. Section 3 defines the implementation details using NI USRP-2930. Finally , in section 4, the e xperim ental results obtained from the implementation are pro vided in order to illustrate the performance of Ener gy Detector . 2. ENERGY DETECT OR B ASED SENSING 2.1. Pr oblem F ormulation In our system model, a SU senses the presence of a PU. In order to perform a good detection of a spectrum opportunity , tw o h ypotheses, H 0 and H 1 are defined respecti v ely for the absence and the presence of a PU signal. Hence, our h ypothesis model for transmitter detection can be e xpressed as follo ws: ( H 0 : s ( l ) = n ( l ) H 1 : s ( l ) = x ( l ) + n ( l ) ; (1) where, s ( l ) represents the recei v ed data, x ( l ) is the transmitted signal by the primary user and n ( l ) denotes the White Gaussian noise independent from the transmitted signal, with zero mean and v ariance 2 n . From the tw o h ypotheses, tw o probabilities describe the performance of the spectrum sensing technique: the f alse-alarm probability P f a , which is the probability of declaring wrongly H 1 and the detection probability P d , which is the probability of declaring correctly H 1 . The main purpose of all spectrum sensing techniques i s to maximize the detection probability for a lo w f alse-alarm probability . 2.2. Ener gy Detector Ener gy Det ector , also kno wn as radiometry , is the most popular and widely used spectrum sensing technique because of its lo w computational and implementation comple xities. It is a simple sensing technique that does not need prior kno wledge of the PU s signal, only the v alue of the White Gaussian Noise is needed. Urk o witz [4] w as the first to in v estig ate the detection of an unkno wn signal in a White noise channel using the ener gy detector based sensing. The PU signal is detected by comparing the total ener gy of the recei v ed signal, o v er a specified time duration, with a threshold. Thus, the test statistics of the ener gy detector is written as: T E D = 1 L L X l =0 j s ( l ) j 2 ; (2) where, L denotes the size of the observ ation sequence. The presence of a PU’ s signal is thus detected if the ener gy is greater than the threshold. The decision is then e xpressed as follo ws: T E D H 1 ? H 0 ; (3) where, denotes the threshold. The Probability Density Function (PDF) of the test statistics T E D can be modeled as a Gaussian distrib ution [11] according to the tw o h ypothesis as follo ws: 8 > < > : H 0 : T E D N 2 n ; 4 n L H 1 : T E D N 2 x + 2 n ; ( 2 x + 2 n ) 2 L ; (4) P erformance Evaluation of Ener gy Detector using NI USRP-2930 (F . Z. El Bahi) Evaluation Warning : The document was created with Spire.PDF for Python.
1936 ISSN: 2088-8708 where, 2 n represents the v ariance of the White Gaussian noise and 2 x denotes the v ariance of the transmitted PU’ s signal. Based on the PDF of the test statistics, t he detection probability P d and the f alse-alarm probability P f a can be e xpressed as: P f a = P r ( T E D > j H 0 ) = Q  2 n 1 p L (5) P d = P r ( T E D > j H 1 ) = Q   2 n 1 s L 2 + 1 ! ; (6) where, = 2 x 2 n denotes the Signal to Noise Ratio (SNR) and Q ( : ) is the Marcum Q-function defined as: Q ( y ) = 1 p 2 R 1 y e u 2 2 du . F or a tar get f alse-alarm probability , the v alue of the threshold can be calculate d by in v erting the relation described in Eq. 6 as follo ws: = 2 n Q 1 ( P f a ) p L + 1 ; (7) where, Q 1 ( : ) is the in v erse Marcum Q-function. 3. IMPLEMENT A TION DET AILS In this section, we pro vide the practical implementation details of the Ener gy Detector bas ed sensing. W e used for both transmission and reception an NI USRP-2930 (Uni v ersal Softw are Radio Peripheral) and a Desktop Computer with LabVIEW 2014 and MA TLAB R2013a to control the USRP using a Gig abit Ethernet Cable as sho wn in Figure 1. Figure 1. Implementation’ s structure 3.1. NI USRP-2930 NI USRP-2930 is a Softw are Defined Radio (SDR) transcei v er , able to transmit and recei v e RF (Radio Fre- quenc y) signals, from the USRP series of the National Instruments’ brand, it is widely used for both teaching and research in wireless communications. Furthermore, it enables a wide range of RF applications co v ering common standards such as GSM Cellular , broadcast radio, W iFi, GPS and digital TV . The USRP hardw are is a straightforw ard RF platform for rapid prototyping applicat ions such as spectrum monitoring and ph ysical layer communication. It has the ability to transmit and recei v e RF signals across a frequenc y range from 50 MHz to 2.2 GHz. Moreo v er , the NI USRP-2930 has an inte grated GPS-disciplined clock that pro vides GPS position information, impro v ed frequenc y accurac y and synchronization capabilities [12]. 3.2. T ransmitter The programming and design parts of the transmitter are implemented in LabVIEW in order to control the NI USRP-2930. F or our e xperiment , we consider an OFDM (Orthogonal Frequenc y Di vision Multiple xing) transmitted signal with a carrier frequenc y set to 200 Mhz. As seen in Figure 2, the transmitter’ s front pane l is di vided into tw o parts. The left part contains three blocks; USRP P arameters, OFDM signal parameters and Deb ug (Figure 3). In the OFDM signal parameters block, we can choose the OFDM standard to transmit among the follo wing ones: 3GPP/L TE, W iMax 802.16, D VBT -2K and 802.22-1K. The right part of the transmitter represents the PSD (Po wer Spectrum Density) of the OFDM transmitted signal. IJECE V ol. 7, No. 4, August 2017: 1934 1940 Evaluation Warning : The document was created with Spire.PDF for Python.
IJECE ISSN: 2088-8708 1937 Figure 2. T ransmitter’ s front panel. Figure 3. T ransmitter’ s tab control. 3.3. Recei v er In the same w ay as the transmitter , we ha v e implemented the recei v er in LabVIEW as sho wn in Figure 4. The recei v er’ s front panel is also di vided into tw o parts, the left one contains three blocks; the USRP P arameters, Ener gy Detector and Deb ug. The recei v er’ s deb ug part is the same as the one of the transmitter . In the Ener gy Detector part, as sho wn in Figure 5, we calculate the P d (Detection Probability) for a specific v alue of P f a (F alse-alarm Probability) and realizations. The right part of the recei v er contains the PSD of the recei v ed OFDM signal, which is almost the same as the one of the OFDM transmitted signal. Figure 4. Recei v er’ s front panel. P erformance Evaluation of Ener gy Detector using NI USRP-2930 (F . Z. El Bahi) Evaluation Warning : The document was created with Spire.PDF for Python.
1938 ISSN: 2088-8708 Figure 5. Recei v er’ s tab control. 4. RESUL TS AND DISCUSSION In this section, we present e xperimental results of the implementation of the Ener gy Detector based sensing using NI USRP-2930 and a v eraged o v er 1000 realizations. 4.1. Effect of SNR In this e xperiment, we test the impact of dif ferent SNR (Signal to Noise Ratio) v alues on the Detection Probability ( P d ). W e fix the F alse-alarm Probability ( P f a ) to 0.01 and v ary the SNR v alue from -24 dB to 0 dB with a step of 2 dB. Each measurement result is the a v erage v alue of 1000 measurement results for the same SNR v alue. W e generated an OFDM signal with 64 subcarriers, 10 symbols and c yclic prefix equals to 8. Figure 6a and 6b sho w the MA TLAB simulation (Computer simulations) and measurement results (USRP implementation) of the P d v ersus SNR of a 3GPP/L TE signal and D VBT -2K signal respecti v el y . W e can notice from both figures that the P d increases with the SNR v alues. Thus, the lar ger SNR, the better the detecti on of the OFDM PU. Furthermore, USRP implementations are closer to MA TLAB simulations. (a) (b) Figure 6. Detection probability ( P d ) vs. Signal to Noise Ratio (SNR) (with P f a =0.01) of (a) 3GPP/L TE signal and (b) D VBT -2K signal. IJECE V ol. 7, No. 4, August 2017: 1934 1940 Evaluation Warning : The document was created with Spire.PDF for Python.
IJECE ISSN: 2088-8708 1939 4.2. Recei v er Operation Characteristic In order to e xamine the ef fect of P f a on the detection performance of the Ener gy Detect or , we fix the v alue of the SNR to -10.4 dB and v ary the P f a from 0.1 to 1 with a step of 0.1. The achie v ed P d as a function of P f a is called the R OC (Recei v er Operating Characteristic) curv e. W ith the same w ay as the abo v e e xperiment, each result is a v eraged o v er 1000 measurements. Figure 7a and 7b represent the MA TLAB simulation and USRP implementation of the P d v ersus P f a of a 3GPP/L TE signal and D VBT -2K signal respecti v ely . As sho wn in those figures, USRP implementations are closer to MA TLAB simulations and the Ener gy Detector achie v es good performance for all P f a v alues. (a) (b) Figure 7. Detection probability ( P d ) vs. F alse-alarm probability ( P f a ) (with SNR=-10.4 dB) of (a) 3GPP/L TE signal and (b) D VBT -2K signal. 5. CONCLUSION In this paper , we ha v e pro vided an e xperimental performance e v aluation of the Ener gy Detector based sensing using NI USRP-2930, which is a Softw are Defined Radio (SDR) transcei v er . W e ha v e tested the impact of SNR on the detection probability . Furthermore, a R OC curv e w as obtained for a lo w SNR v alue. Experimental results sho ws that the Ener gy Detector achie v es good performances for lo w SNR v alues and for all P f a v alues and are closer to those of computer simulations using MA TLAB. REFERENCES [1] S. S. Ali, C. Liu, and M. Jin, ”Minimum Eigen v alue Detection for Spectrum Sensing in Cogniti v e Radio, Int. J . Electr . Comput. Eng . , v ol. 4, no. 4, p. 623, 2014. [2] Y . Saleem and M. H. Rehmani, ”Primary radio user acti vity models for cogniti v e radio netw orks: A surv e y , J . Netw . Comput. Appl. , v ol. 43, pp. 116, 2014. [3] M. S. Hossain, M. I. Abdullah, and M. A. Hossain, ”Hard combination data fusion for cooperati v e spectrum sensing in cogniti v e radio, Int. J . Electr . Comput. Eng . , v ol. 2, no. 6, p. 811, 2012. [4] H. Urk o witz, ”Ener gy detection of unkno wn deterministic signals, Pr oc. IEEE , v ol. 55, no. 4, pp. 523-531, 1967. P erformance Evaluation of Ener gy Detector using NI USRP-2930 (F . Z. El Bahi) Evaluation Warning : The document was created with Spire.PDF for Python.
1940 ISSN: 2088-8708 [5] H. Sun, A. Nallanathan, C.-X. W ang, and Y . Chen, ”W ideband spectrum sens ing for cogniti v e radio netw orks: a surv e y , IEEE W ir el. Commun. , v ol. 20, no. 2, pp. 7481, 2013. [6] M. Hu, Y . Li, X. Lu, and H. Zhang, ”T one reserv ation to minimize nonlinearity impact on OFDM signals, IEEE T r ans. V eh. T ec hnol. , v ol. 64, no. 9, pp. 43104314, 2015. [7] C. Smith and J. Me yer , ”3G W ireless with W iMAX and W iFi: 802.16 and 802.11. McGr aw-Hill Pr ofessional , 2005. [8] H. Holma and A. T oskala, ”HSDP A/HSUP A for UMTS: high speed radio access for mobile communications”. J ohn W ile y and Sons , 2007. [9] U. Ladeb usch and C. A. Liss, ”T errestrial D VB (D VB-T): A broadcast technology for st ationary portable and mobile use, Pr oc. IEEE , v ol. 94, no. 1, pp. 183193, 2006. [10] O. A. Dobre, R. V enkatesan, and D. C. Popescu, ”Second-order c yclostationarity of mobile W iMAX and L TE OFDM signals and application to spectrum a w areness in cogniti v e radio systems, IEEE J . Sel. T op. Signal Pr o- cess. , v ol. 6, no. 1, pp. 2642, 2012. [11] Y .-C. Liang, Y . Zeng, E. C. Y . Peh, and A. T . Hoang, ”Sensing-throughput tradeof f for cogniti v e radio netw orks, IEEE T r ans. W ir el. Commun. , v ol. 7, no. 4, pp. 1326-1337, 2008. [12] National Instruments Co., ”USRP-292x/293x Datasheet, [Online]. A v ailable: http://www .ni.com/datasheet/pdf/en/ds-355 , accessed 3 February 2017. BIOGRAPHIES OF A UTHORS F atima Zahra EL B AHI is an engineer in Netw ork and T elecommunication, graduated in 2014 from National School of Applied Sciences of T angier , Morocco. She is currently w orking to w ard Ph.D. de gre e in the laboratory of Signal, Systems and Components, in F aculty of Sciences and T echnologies of Fez at Sidi Mohammed Ben Abdelah Uni v ersity , Morocco. Her main research interests are signal processing and cogniti v e radio. Hicham GHENNIOUI is an assistant director of Signals, Systems and Components Laboratory at F aculty of Sciences and T echnologies, Fez, Morocco. Since 2011, he is a full-time Associate Pro- fessor at the F aculty of Sciences and T echnologies, Fez. He recei v ed the Ph.D de gree in Computer Science and T elecommunications in 2008, jointly from Mohamed V Uni v ersity and the T oulon Uni- v ersity . In 2004, recei v ed the D.E.S.A. de gree in Computer Science and T elecommunications from the Mohamed V Uni v ersity . His main research interests are signal/image processing including blind sources separation, data analytic, deblurring and cogniti v e radio. Mohcine ZOU AK is the director of the Cancer Research I nstitute of Fez, Morocco. President of the Conference of D eans of F aculties Scientific Morocco (CCMS). V ice P resident of International Conference of Heads of Uni v ersities and Scientific Institutions of French e xpression CIR UISEF . Since 2005, he is a professor of Higher Education at F aculty of Sciences and T echnologies, Fez, Morocco. He assured lessons in the areas of signal processing, electronic systems and telecommu- nications as well as in the fie lds of stochastic estimation. His research acti vities mainly concern the signal processing and telecommunications. IJECE V ol. 7, No. 4, August 2017: 1934 1940 Evaluation Warning : The document was created with Spire.PDF for Python.