Study of WiMAX Based Communication Channel Effects on the Ciphered Image Using MAES Algorithm

Transcription

1 Stuy of WiMAX Bse Communition Chnnel Effets on the Ciphere Imge Using MAES Algorithm Ahme G. Wy, Slim M. Wi, Hyer J. Mohmme Ali A. Aullh Communitions Tehs. Engineering Deprtment, Al-Furt Alwst Tehnil University, Irq. Tehnil Institute of Njf, Al-Furt Alwst Tehnil University Astrt Protetion of multimei informtion from ifferent types of ttkers n sen it over ommunition hnnels hs eome importnt for people n governments. The form of reonstrute imge from n tul trnsmission hnnel is n importnt hllenge fes the iphering lgorithms. MAES lgorithm is n importnt moifie version of AES lgorithm use to ipher n imge with high seurity level n low omputtion ost. This pper presents n stuy of Chnnel effets on MAES iphering lgorithm. An enrypte imge y MAES is trnsmitte n reeive over WiMAX stnr then hek its fetures. Comprison etween originl n reonstrute versions of seret imge ws one. Mny prmeters suh s visul sene, histogrm nlysis, orreltion oeffiients n entropy were use to hieve this omprison. The results of stuy hieve in this work prove tht the iphere lgorithm is strong ginst WiMAX hnnel prolems espeilly when the SNR of the hnnel inrese. Keywors: WiMAX, imge protetion, MAES, hnnel ommunitions INTRODUCTION The seure trnsmissions for igitl imges hve its importne in toy's vieo onferening n imge ommunitions [1]. Owing to the long-term the inresing growth of igitl multimei n teleommunition servies, resoures of igitl re otinle for puli ommon esses. However, the importnt issue is to keep intelletul property of worthy igitl msterwork while Gurntee the puli essiility [2]. Vieo n Imge enryption hve enforements in iverse fiels inluing meil imging, Tele-meiine, internet, multimei n militry ommunition [1]. Mny of the ryptogrphi lgorithms hve een propose n wiely use suh s DES, IDEA, Triple-DES, AES, n RSA et. In spite of the AES lgorithm onsiers quite resistne versus ttks, hnnel ttk hs eome the new mene ginst AES [3]. Wie pplitions for AES lgorithm in our ily life were otine, suh s ell phones, smrt rs, WWW servers n utomte teller mhines. AES is reognize lok ipher lgorithm n hve vrious fetures, suh s implementtion ility n high seurity level [4]. On the other sie, very fst spre of smrt phones hs proue n growing emn for trnsfer multimei over wireless networks. Leing two stnrs, i.e., IEEE m WiMAX n LTE- Avne, fit out high trnsmission t rtes for multimei of the rel-time pplitions ( e.g., vieo onferening) tht require to trnsmit vst of t[5, 6]. Worlwie Interoperility for Mirowve Aess (WiMAX, IEEE ), is moern wireless hnnel for ron pplitions with trnsmission in muh pplitions n high nwith[7]. WiMAX rising network overge n improving performne of ommunition provies extensive overge re of ron servies. The ron ess oul fitting the wnting link for the onnetion in wireless metropolitn re networks [8]. In present stuy, performne of our previous propose lgorithm MAES lgorithm [9] is evlute uner hnnel onition. A WiMAX network is use s ommunition meium in this evlution opertion. This pper is orgnize s follows: In setion II rief esription of AES lgorithm is given. Setion III esries the etils of WIMAX Stnrs physil lyer. Setion IV presents simultion n isussion of results. Pper is ene with onlusion. AES ALGORITHM Cryptogrphy is the importnt proess, whih plys role in t seurity. Where ustomers n store importnt n sensitive informtion n then trnsfer it over inseure networks so tht unuthorize people nnot re it. 1. Algorithm: AES is lok ipher well--kown roun-se symmetri, where sizes of the input n output re equl re to 128 its n fixe. The length of AES seret key is flexile n llow key lengths of 128,192 or 256 its. Primitive four funtions re exeutee in sequene Nr -1 time ompose loop lle roun, SuByte, ShiftRow, MixColumn n ARounKey. The itertion numer of loop ( Nr ) my e 14, 12, or 10 onsisting on the key size. The opertion of the SuByte is yte nonliner sustitution whih perform inepenently on every yte of the sitution using sustitution tle. ShiftRow is irulr shifting opertion on the rows of the stte with vrious numers of ytes (offsets). Mixes opertion of the ytes in eh olumn through the MixColumn y the multiplition of the stte with fixe polynomil (3x 3 +x 2 +x+2, x 3 +x 2 +9x+e for enryption n 6009

2 eryption respetively) moulo x4+1. In ARounKey proess XOR opertion is performe y s t eh itertion loop roun key to the stte [10]. Roun #1 Roun # 9 Roun # 10 Plintext A roun key Sustitutive ytes Shift row Mix olumns A roun key Sustitutive ytes Shift row Mix olumns A roun key Sustitutive ytes Shift row A roun key Ciphertext Key w(0,3) Expn key w(4,7) w(36,39) w(40,43) Figure 1: Blok igrm for AES enryption n eryption In the following steps 4 4 rry ytes lle the Stte is estlishe. This mtrix hve four ytes rrnge in four rows, eh ontining N ytes, where N is the lok size ivie y 32 (numer of wors). Whole AES steps (Cipher n Inverse Cipher) re performe on the Stte mtrix, fter whih its finl vlue is opie to the output (Stte rry set is onverte k to the it sequene). The AES lgorithm onsist of sequenes of 128 its for eh input n output. Sometimes these sequenes will e referre to s loks where the numer of its they inlue, will e referre to s their length. The AES lgorithm ipher key is sequene of 128, 192 or 256 its. Ten rouns re on AES lgorithm enryption, s n e shown in Fig. (1). Initilly, the 128-it key is expne into eleven so-lle roun keys, eh of them 128 its in size. To ensure high seurity of the enryption eh roun ontin trnsformtion y utilize the orresponing key of ipher. The mentione sizes stte in ove is only permit sizes while the other sizes re not llowe. Input lok is orgnize into (4-y-4 ) mtrix, where every entry represents yte. 2- A signifint prolem in imges iphering shemes is the enryption/eryption time euse imges hve lrge mount Roun # 10 Roun # 1 Roun # 9 Plintext A roun key Inverse su ytes Inverse shift row Inverse mix ols. A roun key Inverse su ytes Inverse shift row Inverse mix ols. A roun key Inverse su ytes Inverse shift row A roun key Ciphertext of informtion. Attks re nother hllenge of ll enryption lgorithms. So, iphering sheme moifition shoul e one refully suh tht they o not provie ny hitus to the ttkers. One of fmous types of ttker is the reueroun ttker. The AES is lok enryption sheme hs set of trnsformtions exeute severl times (in itertions). Theoretilly, the reue-roun versry n rek the AES lgorithm in itertions if reue its rouns to 7. Therefore, reuing the rouns numer of the AES lgorithm mkes it hep ginst this ttks. ARounKey is one of the AES itertion trnsforms, where the EX-ORing opertion is onute etween sukey erive from the former sukey in the key sheule opertion n stte. This reson prevents us from reuing the numer of AES rouns. The MEAS gol is moifying the AES lgorithm through eresing the omputtion osts y eresing the exeution time of the MixColumn trnsformtion n repling the S-ox with new simple S-ox to mke the lgorithm more omptile with our gol. In ition, inresing the seurity level through enhnes the key sheule opertion. The etiles of propose moifition in MAES lgorithm re exist in (9). WIMAX STANDARDS (IEE ) Due to the growing request for Bron Wireless Aess (BWA) in unerserve mrkets hs een mjor hllenge for proviers of servies, euse of, the sene of relly glol stnr. A stnr enles ompnies to onstrut systems tht will effetively reh resientil mrkets n unerserve usiness in tehnique tht kup infrstruture uil outs omprle to le, fier n DSL [11]. The im of WiMAX is to supply Internet ess with highspee within severl kilometres in rius for the overge rnge. Theoretilly, WiMAX supplyes for spees out 70 Mps within rnge of 50 kilometres. The physil lyer of WiMAX epens on Orthogonl Frequeny Division Multiplexing (OFDM); it trnsmits informtion (suh s t, vieo, uio n imge) with high-spee, n is utilize y n ron ssortment of ommeril systems. The most vntge of the WiMAX stnr is of permit wireless onnetions etween Bse Sttion (BS) n susriers without neee tht they e in iret line of sight (LOS) with tht se sttion. his tehnology know y Non-Line-of-Sight (NLOS ) [12]. In this pper performne of ipher imge vi WiMAX with QPSK moultion n oing is stuie on the sis of Bit Error Rte, Signl to Noise Rtio. Physil lyer is use to set up the ommunition etween the evies n is responsile for trnsporting the t it sequene. It lso etermines the type of moultion s well s power of trnsmission. Two trnsmission type tehniques were use in WiMAX PHY-lyer OFDM n OFDMA. These tehniques operte on n of frequeny elow 11 GHz tht use FDD n TDD s uplexing tehnology. Slle OFDMA ws use in physil lyer of the IEEE e, where the size of the FFT is hngele n my e equl to: 128, 512, 1024, n 2048 [13] FFT ws opte through this work. The FFT vrile size permits to simulte or implement of the system in high n-with hnnel n rio stte onitions; this lyer hs een epte y WiMAX for 6010

3 portle n moile opertions [14]. Figure 2 presents PHYlyer where the PHY-lyer inlues vrious funtionl stges [6, 15]: Soure S / p IFFT Outer Enoer Aing CP Inner Enoer Ryleigh Chnnel Moultor AWGN OFDM through uiling polynomil from trnsmitte t symols efore trnsmission polynomil with respet to the originl symols. RS oe is kmown s RS(n, k,t) with l-it symols. Tht men, the enoer oupies k t symols for eh l it then s t symols s prity to uil n n-symol oewor. Where, n is enoing ytes numer (output); k is t ytes numer (input) efore enoing, n t is t ytes numer tht le to e orrete. An error orretions of ny RSoe re etermine y (n k), whih is reunny mesurement in the lok. Atully, RSoe n orrete up to t symols if the wrong lotion symols is not known, where t = (n k)/2. The RS oer foun from orgnize RS(n, k, t = 255, 239 n 8 respetively) oe employ GloisFiel speifie s GF(2 8 ). Priml n genertor polynomils whih use for the systemti oe n e expresse in the next two equtions [16]: Primitive Polynomil: Outer Deoer Inner Deoer Demoultor P / S FFT Removl CP p(x) = x 8 + x 4 + x 3 + x (1) Genertor Polynomil Figure 2: Arhiteture for the IEEE WiMAX Physil Lyer g(x) = (x + λ 0 )(x + λ 1 ). (x + λ 2t 1 ) (2) A. Rnomizer To vert long ontinuous sequene of ones n zeros the rnomizer exeute input t rnomiztion on eh llotion on eh urst. This is one y PseuoRnomBinrySequene (PRBS) genertor whih employs fifteen stge registers with polynomil genertor with XOR-gtes in feek onfigurtion s shown in Fig. (3). C. Convolution Enoer The inner inry onvolutionl enoer is fe from the outer RS lok oe. The genertor polynomils is use to erive two output oe its. Figure (4) present onvolutionl enoer ws opte in this work. X=171 OCT MS LSB Dt T T T T T T Dt in Dt out Figure 3: PRBS genertor for rnomiztion Y=133 OCT Figure 4: Convolution enoer of inry rte (1/2) B. Ree-Solomon enoer The feture of Ree-Solomon (RS) oe mke them pproprite to pplitions where errors ple in ursts. RS orretion of error is sheme of oing whih oings D. Punturing, Interleving n Moultion In orer to minimize the mount of trnsmitte t, punturing is use to elete its from the low-rte output loks of enoer onsistently. Punturing n e generte 6011

4 the oing in ifferent rtes tht importnt to hieve ifferent levels of system error preition, these rtes re 1/2, 2/3, 3/4 n 5/6. Blok interlever is use to interleve the enoe t. The it enoe numer per su hnnel is etermine interleving lok size, Nps. Two step permuttion were use in the IEEE interlever. Depening the size n on the sis of ifferent the t were moulte. Shemes like BPSK, QPSK, 16 QAM n 64 QAM my e use to moulte t oming from interlever. The moultion ws opte in this work is QPSK. E. OFDM n Cyli Prefix A multi-rrier moultion tehnique OFDM, tht hieve high effiieny of nwith tht results from orthogonlity etween rriers n multiple rriers shre t mong themselves. The roustness to hnnel fing (Seletive n flt fing) is the min vntge of this tehnique [19]. To preserve orthogonlly of the frequeny n minimize the ely ue to propgtion in multipth hnnel, yli prefix is joine in OFDM signls. To mke so, efore sening the signl, it is joine t the eginning of the signl. In the WiMAX stnr four urtion of yli prefix re ville. The rtio of CP to OFDM symol time, this rtio n e equl to 1/4, 1/8, 1/6 n 1/32. On the other hn, the hnnel s white no A white noise is e y the hnnel of ertin vrine to the useful signl. Frequeny seletive n Flt Ryleigh fing is envisione for further simultor versions. The reeiver performs silly the inverse of the trnsmitter opertion loks. SYSTEM MODEL In this pper, simultion moel se on MATLAB (version 2010) is introue whih onsist of genertion of sen t from the trete imge, enoing unit outer n inner oing is use, igitl moultor of QPSK is opte, OFDM (IFFT) with yli prefix insertion loks in the trnsmitter n yli prefix remover in reeiving, OFDM reeiver (FFT) followe y emoultor n eoer in the reeiver. Figure 4 esrie ll the mentione stges n Tle I liste the prmeters of the moel. Here, the oing prt ompose of three steps of Forwr Error Corretion (FEC), rnomiztion n interleving [6]. FEC is hieve in two stges through the outer n inner oing (Ree Solomon (RS) n Convolutionl Coe (CC), respetively). The omplementry proesses re pplie in the reverse orer in the reeiver en. Ryleigh fing hnnel with three tp points re implemente. Aitive white Gussin Noise (AWGN) with ifferent vrine using to hieve ifferent Signl to Noise Rtio (SNR) vlues. SIMULATION RESULTS In this work the performne of propose system will e evlute through stuying the effet of trnsmission system on the erypte imge. The evlution prmeters use re visul sene, histogrm evlution, orreltion oeffiients, entropy, n BER. Four tests imge in ifferent pplitions re use in this evlution in ifferent vlues of SNR. F. Visul sene The visul sene of originl seret, iphere trnsmitte, iphere reeive n reonstrute imges re viewe n ompre to hek the effet of hnnel on the imge for two vlues of SNR only, where the tests of other SNR vlues shown in ppenies. Figures (7-14) show the imges visul sene for four tests. As lerly shown the ifferene etween trnsmitte n reeive iphere imges is so little suh tht the humn eyes nnot sense this ifferene for ll tests n for ll vlues of SNR s shown in figures ( & ) of ll figures. on the other hn, reonstrute imge is losses with little vlues of SNR while this loss is erese with inrese the SNR vlues (see figures & of Figs. 5-12). Figure 5: Visul sene Test1 SNR 10 B; - Originl Seret imge; - reonstrute seret imge. 6012

5 Figure 7: Visul sene Test2 SNR 10 B; - Originl Seret imge; reonstrute seret imge. Figure 6: Visul sene Test1 SNR 20 B; - Originl Seret imge; reonstrute seret imge. Figure 8: Visul sene Test2 SNR 20 B; - Originl Seret imge; reonstrute seret imge. TABLE I: SYSTEM MODEL PARAMETER Speifition Punture vetor Coing RS outer Convolution NFFT 2048 Moultion QPSK OFDM symol urtion e-7 se Cyli Prefix gur intervl 1/4 Column 1 Column 2 Column 3 xx1 yyy1 zzz1 xxx2 yy2 zzz2 xxx3 yyy3 zz3 xxx4 yy4 zzzzz4 xxx5 yyyyy5 zz5 6013

6 Figure 9: Visul sene Test3 SNR 10 B; - Originl Seret imge; reonstrute seret imge. Figure 11: Visul sene Test4 SNR 10 B; - Originl Seret imge; reonstrute seret imge. Figure 10: Visul sene Test3 SNR 20 B; - Originl Seret imge; reonstrute seret imge. Figure 12: Visul sene Test4 SNR 20 B; - Originl Seret imge; reonstrute seret imge. G. Histogrm Anlysis An imge histogrm is representing istriution of imge pixels on intensity levels. Enrypte imge histogrm shoul e uniformly istriute to ope the sttisti ttks[5]. This ftor use to evlute in two iretion, one y ompring etween the histogrm of trnsmitte n reeive iphere imges, while the other y ompring etween histogrm of originl n extrte versions of seret imge. As shown in Figure 13, the histogrm of two versions of seret imges re muh omptile espeilly with 20 B s in 13-, whih men the enryption lgorithm ws more stronger ginst hnnels prolems. Figure 14 is lso onfirm this nlysis. 6014

7 () () () () () Figure 13: Histogrm nlysis of originl n extrte versions of seret imge; - Originl Seret Imge; - Extrte seret imge with 10 B; - Extrte seret imge with 20 B. 1 st, 2 n, 3 r n 4 th tests in 1 st, 2 n, 3 r n 4 th olumns respetively. () Figure 14: Histogrm nlysis of trnsmitte n reeive iphere imge; - Trnsmitte imge; - Reeive iphere imge with 10 B; - reeive iphere imge with 20 B. 1 st, 2 n, 3 r n 4 th olumns re represent 1 st, 2 n, 3 r n 4 th tests respetively 6015

8 H. Correltion oeffiients Generlly, juste pixels in ll types of imges highly orrelte with itself. The est enryption lgorithm tht introuing protete imge with very low enggement etween the neighoring pixels [19]. Firstly, 5000 pir of jent pixels ws rnomly selete from tests imges. Figure 15 shows the orreltion oeffiients of originl n reonstrute version of seret imges for four tests use. s lerly shown in Figure the orreltion oeffiient of originl n extrte imges ws so onvergene espeilly with SNR equl to 20 B (15.), where with 10 B there re more rnom pixel vlues (15.). Tle II shows the vlue of orreltion oeffiients. () () TABLE II: CORRELATION COEFFICIENTS VALUES imges SNR 10 B 12 B 14 B 16 B 18 B 20 B OI Test1 TCI RCI EI OI Test2 TCI RCI EI OI Test3 TCI RCI EI OI Test4 TCI RCI EI () Figure 15: Correltion oeffiients of originl n extrte versions of seret imge; - Originl Seret Imge; - Extrte seret imge with 10 B; - Extrte seret imge with 20 B. 1 st, 2 n, 3 r n 4 th olumns represent 1 st, 2 n, 3 r n 4 th olumns respetively. I. Entropy The onept of entropy omes from informtion theory n Ergoi theory. Shnnon entropy is efine s metri ssoite with informtion ontent of input signl. In imge proessing fiels, the entropy known s the rnomness size whih n e explin s the out rnge of the informtion soure [20]. Entropy vlues for 4 imge tests in ifferent vlues of SNR re shown in Tle IV elow 6016

9 From results of histogrm nlysis, orreltion oeffiients n entropy, we n fin tht the protete imge y enryption lgorithm is not muh ffete y hnnels effets. Also, the effets of hnnels on the protete imge is erese when the SNR inrese. CONCLUSIONS Figure 16: BER v. SNR An stuying of trnsmission hnnel effet on the iphering lgorithm ws introue in this pper. An imge enrypte y new moifie version of AES lgorithm n trnsmitte over hnnel in WiMAX stnr then reeive it n hek its fetures. This stuy is hieve through omprison etween trnsmitte iphere imge n reeive iphere imge n through omprison etween originl n reonstrute versions of seret imge. Severl prmeters use to hieve this omprison suh s visul sene, histogrm nlysis, orreltion oeffiients n entropy. The results of evlution prove tht the effets of trnsmission hnnel prolems is very low on the imge iphere y MAES lgorithm whih men this iphere lgorithm is strong ginst hnnel prolems espeilly if the SNR of the hnnel inrese. REFERENCES [1] S. H. Kmli & R. Shkerin, " A New Moifie Version of Avne Enryption Stnr Bse Algorithm for Imge Enryption" Interntionl Conferene on Eletronis n Informtion Engineering (ICEIE 2010), Vol. 1, pp [2] T. M. Triet & D. A. Du, "Applying the AES n Its Extene Versions in Generl Frmework for Hiing Informtion in Digitl Imges", CIS, Prt II, LNAI 3802, 2005, pp [3] C. Yi-heng, Z. Xue-heng, L. Zheng-lin, C. Xio-fei, H. Yu, " Dynmi Inhomogeneous S-Boxes esign for effiient AES msking mehnisms," The Journl of Chin Universities of Posts n Teleommunitions, Vol. 15, No. 2, 2008, pp [4] C.W. Hung, C.L. Yen, C.H. Ching, K.H. Chng n C.J. Chng, "The Five Moes AES Applitions in TABLE III: ENTROPY imges SNR 10 B 12 B 14 B 16 B 18 B 20 B OI Test1 TCI RCI EI OI Test2 TCI RCI EI OI Test3 TCI RCI EI OI Test4 TCI RCI EI Souns n Imges," Sixth Interntionl Conferene on Informtion Assurne n Seurity, 2010, pp [5] S.M. Wi & N. Zinl, "Deomposition y Binry Coes-Bse Speey Imge Enryption Algorithm for Multiple Applitions," IET Imge Proessing Vol. 9, No. (5), 2015, p.p [6] C.H. Lin, R.H. Hwng, J.J. Wu, J.F. Lee, n Y.D. Lin, "Integrtion of Sptil Reuse n Allotion for Downlink Brost in LTE-Avne n WiMAX Rely Networks," IEEE Trnstions on Vehiulr Tehnology, VOL. 64, NO. 11, NOVEMBER 2015, pp [7] J. Rkesh, W. Vishl, U. Dll, "A Survey of Moile WiMAX IEEE m Stnr," Interntionl Journl of Computer Siene n Informtion Seurity (IJCSIS), Vol. 8, No. 1, April pp [8] A. Bouh & F. T. Benimer, "Performne Stuy Of The MCCDMA As Physil Lyer For Moile Wimx Tehnology," Interntionl Journl of Computer Networking, Wireless n Moile Communitions (IJCNWMC),Vol. 3, No. 2, Jun 2013, pp [9] S. M. Wi, & N. Zinl, "High efinition imge enryption lgorithm se on AES moifition," Wireless Personl Communitions, Vol. 79, No.2, 2014, pp [10] Deprtment of Commere, N. I. O. S.. T., Informtion Tehnology Lortory (Itl). (2001). Avne Enryption Stnr (AES) (Fips Pu 197). Computer Seurity Stnr, Cryptogrphy Port Royl Ro, Springfiel, VA 22161, Ntionl Tehnil Informtion Servie (NTIS): 51. [11] V. Grewl, A.K. Shrm, "On Performne Enhnements of WiMx PHY Lyer with Turo Coing for Moile Environments," Interntionl Journl of Avne Siene n Tehnology, Vol. 31, June, [12] M. Rooyen, J.W. Oenl, n J. Jouert, "High Gin Diretionl Antenn for WLAN n WiMAX 6017

10 Applitions," IEEE Antenns n Wireless Propgtion Letters,DOI /LAWP , [13] M.M. Mtlgh, O.M. Hmmouri, B. Puel, "Crosslyer Cpity Optimiztion in WiMAX Orthogonl Frequeny Division Multiple Aess Systems with Multi-lss Qulity of Servies n Users Queue Sttus," IET Commun., Vol. 8, No. 14, 2014, pp [14] H. Zhi, Q. Go, C. Ling, R. Yu, n S. Liu, "A Duln High-gin Bse Sttion Antenn for WLAN n WiMAX Applitions," IEEE Antenns n Wireless Propgtion Letters, DOI /LAWP , [15] W. Xiumin, G. Tingting, L. Jun, S. Chen n H. Fngfei, "Effiient Multi-rte Enoer of QC-LDPC Coes Bse on FPGA forwimax Stnr," Chinese Journl of Eletronis, Vol. 26, No. 2, Mr [16] M.A. Mohme, F.W. Zki n R.H. Moseh, "Simultion of WiMAX Physil Lyer: IEEE e," Interntionl Journl of Computer Siene n Network Seurity IJCSNS, Vol.10 No.11, Novemer 2010, pp [17] Neh Rthore, Dr. Suhir Kumr, Ahishek Chouey "OFDM Performne evlution of WIMAX MAC lyer :IEEE e," Interntionl Journl of Eletronis & Communition Tehnology IJECT, Vol. 2, No. 3, Sept [18] A. G. Wy, A.S. Aullh n S. A. Alsey, "Turo n Convolution Coes Assiste Spe Time Blok Coe-Sptil Moultion," Future Communition Networks (ICFCN) Interntionl Conferene, 01 June [19] M. Viehi n B. Justus Ri, "Enhne MixColumn Design for AES Enryption," Inin Journl of Siene n Tehnology, Vol. 8, No. 35, DOI: /ijst/2015/v8i35/82302, Deemer 2015, pp [20] S.M. Wi, N. Zinl n A. Aulger, "Grey Sle Imge Hiing Metho Bse on Deomposition Opertion," IEEE Stuent Conferene on Reserh & Development 2013, Putrjy, Mlysi. 6018