SPEECH ENHANCEMENT USING A MODULATION DOMAIN KALMAN FILTER POST-PROCESSOR WITH A GAUSSIAN MIXTURE NOISE MODEL. Yu Wang and Mike Brookes
|
|
- Stuart Carpenter
- 5 years ago
- Views:
Transcription
1 14 IEEE Iteratioal Coferece o Acoustic, Speech ad Sigal Processig (ICASSP) SPEECH ENHANCEMENT USING A MODULATION DOMAIN KALMAN FILTER POST-PROCESSOR WITH A GAUSSIAN MIXTURE NOISE MODEL Yu Wag ad Mike Brookes Departmet of Electrical ad Electroic Egieerig, Exhibitio Road, Imperial College Lodo, UK {yw9, mike.brookes}@imperial.ac.uk ABSTRACT We propose a speech ehacemet algorithm that applies a Kalma filter i the modulatio domai to the output of a covetioal ehacer operatig i the time-frequecy domai. We show that the predictio residual sigal of the spectral amplitude errors at the output of the baselie ehacer do ot follow a Gaussia distributio. Accordigly, the Kalma filter used i our ehacemet algorithm combies a colored oise model with a Gaussia mixture model of the residual oise. We evaluate the performace of the speech ehacemet algorithm o the core TIMIT test set ad demostrate that it gives cosistet performace improvemets over the baselie ehacer ad over a previously proposed Kalma filter post-processor. Idex Terms speech ehacemet, post-processig, Kalma filter, Gaussia mixture model, modulatio domai 1. INTRODUCTION Over the past decades, may speech ehacemet algorithms have bee proposed i order to elimiate or reduce uwated backgroud oise. Ehacemet algorithms i the time-frequecy domai, such as [1] ad [], ca be effective i reducig the oise ad improvig the sigal-to-oise ratio (SNR) but they also distort the speech ad itroduce spurious artefacts kow as musical oise. I order to solve this problem, several post-processig methods have bee proposed that either filter the time-frequecy gai fuctio used withi the ehacer or else act directly o its output sigal. I [3], media filterig is applied to time-frequecy cells that are idetified as havig a low probability of cotaiig speech eergy i order to elimiate the isolated peaks that characterise musical oise ad i [4] musical oise i frames with low SNR is atteuated by smoothig the gai fuctio of the baselie ehacer. Other techiques, such as cepstral smoothig [5] ad Kalma filterig (KF) [6] have also bee used to post-process the miimum mea square error () spectral amplitude estimator [] to reduce the musical oise ad improve the quality of the ehaced speech. The use of a KF was itroduced i [7] for speech ehacemet assumig the oise was white ad i [8] this was later exteded to colored oise. Recet iterest i performig speech ehacemet i the modulatio domai [9, 1, 11] icludes the algorithm described i [1] which applies the KF to the short-time modulatio domai spectral amplitudes. The assumptio made i the KF is that the predictio residual sigals of both speech ad oise are Gaussia distributed with zero-mea. However, we have foud that the predictio residual sigal of the spectral amplitude errors i the ehaced speech sigal do ot follow a Gaussia distributio. Therefore, extedig the algorithm i [6], we propose i this paper a KF post-processor i the modulatio domai usig a Gaussia mixture model (GMM) of the oise which is colored due to the overlap betwee frames. The rest of the paper is orgaized as follows: Sec. gives the motivatio ad the probabilistic derivatio of the GMM Kalma filter for colored oise ad also describes the update procedure for the model parameters. I Secs 3 ad 4, we evaluate the proposed algorithm ad give our coclusios.. GMM KALMAN FILTER.1. Distributio of predictio error I the covetioal KF, the predictio residual sigal of both speech ad oise are assumed Gaussia distributed. However, after processig oisy speech by a ehacer, most of the statioary oise has bee removed leavig behid some residual oise together with musical oise artefacts especially where the iput oise power was high [13]. Because the musical oise is characterized by isolated spectral peaks i the spectrogram, it is difficult to predict i the modulatio domai. As a result, the predictio errors associated with the musical oise may be very large, ad the overall distributio of the predictio errors of the oise i the ehaced speech does ot follow a Gaussia distributio. To illustrate this, we show i Fig. 1 the distributio of the ormalized predictio error of the spectral amplitude errors i the ehaced speech i each time-frequecy bi together with a fitted sigle Gaussia distributio (i red) ad a 3-mixture GMM (i /14/$ IEEE 774
2 probability desity predictio error 3 compoets Sigle compoet oisy speech STFT ehaced speech ISTFT oise estimatio speech LPC estimatio Kalma Filter & GMM parameters update Trai iitial GMM parameters LPC estimatio ormalized predictio error Fig. 1. Distributio of the ormalized predictio error of the oise spectral amplitudes i -ehaced speech. The predictio errors are ormalized by the RMS power of the oise predictor residual i the correspodig modulatio frame. gree). The histogram shows the distributio over all timefrequecy bis usig the TIMIT core test set [14] corrupted by additive car oise at SNRs betwee 1 ad +15 db usig the framig parameters from Sec. 3. The estimated oise amplitude trajectory i each frequecy bi is represeted by a autoregressive model whose predictio error is ormalized by the root-mea-square (RMS) level of the oise predictor residual i the correspodig modulatio frame. From the figure, we see that the overall predictio residual sigal is ot zero mea ad does ot follow a Gaussia distributio. Based o the empirical predictio errors, we have exteded the covetioal colored oise KF to icorporate a GMM oise distributio. We use N (µ, ) to deote a multivariate Gaussia distributio with mea vector µ ad covariace matrix ad use N (x; µ, ) for its probability desity at x... Derivatio of GMM Kalma Filter The diagram of the proposed algorithm is show i Fig.. Followig time-frequecy domai ehacemet, the spectral amplitude of the short-time Fourier trasform (STFT) at time frame ad frequecy bi k is give by Y,k = X,k + W,k where X,k is the amplitude of the clea speech sigal ad W,k is the oise arisig from a combiatio of acoustic oise ad the ehacemet artefacts. The output from the KF ˆX,k is combied with the oisy phase spectrum,k ad passed through a iverse-stft (ISTFT) to create the output speech ˆx(t). I this ad the ext subsectio we will give the derivatio of the GMM Kalma filter (GMMKF) ad the parameter update procedure. Because each frequecy bi, k, is processed idepedetly ad for clarity, we omit the frequecy idex below. Fig.. Diagram of the proposed GMM KF algorithm Our system model is z +1 = A z + Dq (1) y +1 = c T z +1 () where z =[x x N+1 w w M+1 ] T is the N +M dimesioal state vector for both the speech, x, ad oise, w, ad q =[u v ] T cotais the correspodig predictio residuals. The (N + apple M) (N + M) trasitio matrix, A, is Ta i the form A = where T T a ad T b are the b trasitio matrices apple for speech ad oise respectively ad T a a is give by T T a = where a I is the vector of liear predictio (LPC) coefficiets for the speech. T b ad b are the correspodig quatities for the oise. The (N + M) matrix D is all zero except for d 1,1 = d N+1, =1. Likewise the colum vector c is all zero except for c 1 = c N+1 =1. We represet the predictio residuals as a -elemet vector q with a Gaussia mixture distributio of J mixtures as q N (µ, ) (3) As i a covetioal Kalma filter, we assume that the state vector at time based o observatios up to time is Gaussia distributed z s N (z, P ). Followig the time update, the distributio of z +1 becomes a Gaussia mixture P j N (z +1, P +1 ) where z +1 = Az + Dµ P +1 = AP A T + D D T Applyig the costrait c T z +1 = y +1 chages the Gaussia mixture parameters as follows [15] k +1 = P +1 c(ct P +1 c) 1 (4) z = z +1 + k +1 (y+1 ct z +1 ) (5) P = P +1 k +1 ct P +1 (6) 775
3 Fially, we collapse the GMM ito a sigle Gaussia for the estimatio of the state vector at time +1 The parameters of each model ca ow be updated adaptively as [18] +1 = P j z = P = N (y +1; c T z +1, ct P +1 c) N (y +1; c T z +1, ct P +1 c) (7) +1 z (8) +1 (P z (z )T ) z z T (9) The quatity +1 i (7) represets the posterior probability that z +1 belogs to mixture j. Thus we ca use the ew Kalma filter to process the residual oise i the ehaced speech because the GMM ca be used to model the spectral amplitude errors i the ehaced speech..3. Update of parameters The spectral amplitudes, Y,k are divided ito overlappig modulatio frames ad autocorrelatio LPC aalysis [16] is performed i each modulatio frame to obtai a vector of speech predictio coefficiets, a, ad a residual power a. To obtai the correspodig oise coefficiets, the sequece of spectral amplitudes, Y,k is passed through a oise power spectrum estimator [17] before performig LPC aalysis to obtai the oise predictor coefficiets, b, ad the residual power b. Withi the oise GMM, (3), the speech residual compoet u N(, a) is idetical i all mixture compoets but the ormalized oise residual e = v / b is modeled as a Gaussia mixture e P j N (m, j ). We model the ormalized residual rather tha the residual itself so that the GMM parameters are idepedet of the speech ad oise amplitudes. I order to update the GMM parameters we apply the oise predictor coefficiets, b, from the curret modulatio frame to the sequece of estimated oise spectral amplitudes to obtai a oise predictio error e b for each acoustic frame. The probability that e +1 comes from model j is give by p +1 = P j N (e +1; m, N (e +1; m, ) ) (1) Because ow the probability of the model give the observatio error is kow, we ca update i each acoustic frame the effective umber of observatios (O ), the sum of the observatios (S ) ad the sum of the squared observatios (T ) as O +1 = p +1 + O ad T +1 = p +1 e +1 + T, S +1 = p +1e+1 + S, where is a forgettig factor. m +1 = S +1 /O +1 (11) +1 = T +1 /O +1 m +1 (1) +1 = O +1 Pj O +1 =(1 ) O +1 (13) To iitialize the model, we trai a GMM with parameters m, ad offlie o a large amout of data ad set O = m /(1 ), S = m O ad T = ( + m )O. To esure stability of the update procedure, we impose lower bouds o p ad to prevet them becomig zero. 3. IMPLEMENTATION AND EVALUATION 3.1. Stimuli of experimets I this sectio, we compare the performace of the proposed Kalma filter post-processor based o a GMM () with the baselie ehacer from [] ad the modulatiodomai Kalma filter post-processor () from [6]. The iitial GMM parameters are traied usig a subset i the traiig set of the TIMIT database ad usig speech corrupted by white oise. The remaiig algorithm parameters were chose to optimize performace o a developmet subset of the TIMIT traiig database. The umber of mixtures used is set as J =3ad we select a acoustic frame legth 16 ms with a 4 ms icremet which gives a 5 Hz samplig frequecy i the modulatio domai. The speech ad oise LPC models are determied from a modulatio frame of 18 ms (3 acoustic frames) with a 16 ms frame icremet ad the model orders i the ad algorithms for the speech ad oise are N =3ad M =4respectively. I the experimets, we use the core test set from the TIMIT database which cotais 16 male ad 8 female speakers each readig 8 distict seteces (totallig 19 seteces) ad the speech is corrupted by the F16 oise from the RSG-1 database [19] ad street oise from the ITU-T test sigals database [] at 1, 1,, 5, 1 ad 15 db global SNR. A Hammig widow is used i the STFT aalysis ad sythesis ad the forgettig factor is set as = Performace evaluatio The performace of the algorithms is evaluated usig both segmetal SNR (segsnr) ad the perceptual evaluatio of speech quality (PESQ) measure defied i ITU-T P.86. All the measuremet values are averaged over the 19 seteces i the TIMIT core test set. The average segsnr for the corrupted speech, baselie ehacer, the algorithm ad the proposed algorithm is show for F16 776
4 segsnr (db) Fig. 3. Average segmetal SNR of ehaced speech after processig by three algorithms versus the global SNR of the iput speech corrupted by F16 aircraft oise (: proposed Kalma Filter post-processor with a Gaussia Mixture oise model; : modulatio-domai Kalma filter postprocessor from [6]; : ehacer from []). segsnr (db) Fig. 4. Average segmetal SNR of ehaced speech after processig by three algorithms versus the global SNR of the iput speech corrupted by street oise. oise i Fig. 3 as a fuctio of the global SNR of the oisy speech. We see that at 15 db global SNR all the algorithms give the same improvemet i segsnr of about 3 db. However, at db global SNR the proposed algorithm outperforms both referece algorithms by about 4 db ad 6 db respectively. The equivalet graphs for street oise are show i Fig. 4. We see that the overall tred i the results is the same ad at db the proposed algorithm gives a additioal improvemet of 1.5 db. The correspodig graphs for PESQ are show i Fig. 5 for F16 oise ad i Fig. 6 for street oise. I Figs 5 ad 6, the average PESQ scores mirror the results see for the segsnr. However, at high SNRs the proposed algorithm is also able to improve the PESQ, ad we obtai a improvemet of approximately.15 ad.5 over the algorithm ad ehacer respectively over a wide rage of SNRs. I additio, iformal listeig tests also suggest that the proposed post-processig method is able to reduce the musical oise itroduced by the ehacer. 4. CONCLUSION I this paper we propose a ew post-processor i the modulatio domai usig a GMM for modelig predictio error of the oise i the output of a covetioal spectral amplitude ehacer. We have derived a KF that icorporates a GMM oise model ad have also preseted a method for adaptively updatig the GMM parameters. We have evaluated our proposed post-processor usig segsnr ad PESQ ad show that the proposed method results i cosistetly improved performace whe compared to both the baselie ehacer ad a modulatio-domai KF postprocessor. The improvemet i segmetal SNR is over 4 db at a global SNR of db while the PESQ score is icreased by about.15 across a wide rage of iput global SNRs. PESQ Fig. 5. Average PESQ quality of ehaced speech after processig by three algorithms versus the global SNR of the iput speech corrupted by F16 aircraft oise. PESQ Fig. 6. Average PESQ quality of ehaced speech after processig by three algorithms versus the global SNR of the iput speech corrupted by street oise. 777
5 5. REFERENCES [1] S. Boll. Suppressio of acoustic oise i speech usig spectral subtractio. IEEE Tras. Acoust., Speech, Sigal Process., 7():113 1, April [] Y. Ephraim ad D. Malah. Speech ehacemet usig a miimum-mea square error short-time spectral amplitude estimator. IEEE Tras. Acoust., Speech, Sigal Process., 3(6): , December [3] Zeto Goh, Kah-Chye Ta, ad T. G. Ta. Postprocessig method for suppressig musical oise geerated by spectral subtractio. IEEE Tras. Speech Audio Process., 6(3):87 9, May [4] T. Esch ad P. Vary. Efficiet musical oise suppressio for speech ehacemet system. I Proc. IEEE Itl. Cof. o Acoustics, Speech ad Sigal Processig (ICASSP), pages , April 9. [5] C. Breithaupt, T. Gerkma, ad R. Marti. Cepstral smoothig of spectral filter gais for speech ehacemet without musical oise. Sigal Processig Letters, IEEE, 14(1): , December 7. [6] Y. Wag ad M. Brookes. Speech ehacemet usig a robust Kalma filter post-processig i the modulatio domai. I Proc. IEEE Itl. Cof. o Acoustics, Speech ad Sigal Processig (ICASSP), pages , May 13. [7] K. Paliwal ad A. Basu. A speech ehacemet method based o Kalma filterig. I Proc. IEEE Itl. Cof. o Acoustics, Speech ad Sigal Processig (ICASSP), pages , April [8] J. D. Gibso, B. Koo, ad S. D. Gray. Filterig of colored oise for speech ehacemet ad codig. IEEE Tras. Sigal Process., 39(8): , [9] K. Paliwal, K. Wojcicki, ad B. Schweri. Siglechael speech ehacemet usig spectral subtractio i the short-time modulatio domai. Speech Commuicatio, 5(5):45 475, 1. [1] T. H. Falk, S. Stadler, W. B. Kleij, ad W. Y. Cha. Noise suppressio based o extedig a speechdomiated modulatio bad. I Proc. Iterspeech Cof., pages , August 7. [11] J. G. Lyos ad K. K. Paliwal. Effect of compressig the dyamic rage of the power spectrum i modulatio filterig based speech ehacemet. I Proc. Iterspeech Cof., pages , September 8. [1] S. So ad K. K. Paliwal. Modulatio-domai Kalma filterig for sigle-chael speech ehacemet. Speech Commuicatio, 53(6):818 89, July 11. [13] O. Cappe. Elimiatio of the musical oise pheomeo with the Ephraim ad Malah oise suppressor. IEEE Tras. Speech Audio Process., (): , April [14] Joh S. Garofolo, Lori F. Lamel, William M. Fisher, Joatha G. Fiscus, David S. Pallett, Nacy L. Dahlgre, ad Victor Zue. TIMIT acoustic-phoetic cotiuous speech corpus. Corpus LDC93S1, Liguistic Data Cosortium, Philadelphia, [15] Mike Brookes. The matrix referece maual [16] J. Makhoul. Liear predictio: A tutorial review. Proceedigs of the IEEE, 63(4):561 58, April [17] T. Gerkma ad R. C. Hedriks. Ubiased based oise power estimatio with low complexity ad low trackig delay. IEEE Tras. Audio, Speech, Lag. Process., (4): , May 1. [18] D. A. Reyolds, T. F. Quatieri, ad R. Du. Speaker verificatio usig adapted Gaussia mixture models. Digital Sigal Processig, 1(1 3):19 41,. [19] H. J. M. Steeeke ad F. W. M. Geurtse. Descriptio of the RSG.1 oise data-base. Techical Report IZF , TNO Istitute for perceptio, [] ITU-T P.51. Test sigals for use i telephoometry, August
SPEECH ENHANCEMENT USING AN MMSE SPECTRAL AMPLITUDE ESTIMATOR BASED ON A MODULATION DOMAIN KALMAN FILTER WITH A GAMMA PRIOR
SPEECH ENHANCEMENT USING AN MMSE SPECTRAL AMPLITUDE ESTIMATOR BASED ON A MODULATION DOMAIN KALMAN FILTER WITH A GAMMA PRIOR Yu Wag? Mike Brookes? Egieerig Departmet Uiversity of Cambridge Uited Kigdom
More informationSPEECH ENHANCEMENT USING A ROBUST KALMAN FILTER POST-PROCESSOR IN THE MODULATION DOMAIN. Yu Wang and Mike Brookes
SPEECH ENHANCEMENT USING A ROBUST KALMAN FILTER POST-PROCESSOR IN THE MODULATION DOMAIN Yu Wang and Mike Brookes Department of Electrical and Electronic Engineering, Exhibition Road, Imperial College London,
More informationA New Space-Repetition Code Based on One Bit Feedback Compared to Alamouti Space-Time Code
Proceedigs of the 4th WSEAS It. Coferece o Electromagetics, Wireless ad Optical Commuicatios, Veice, Italy, November 0-, 006 107 A New Space-Repetitio Code Based o Oe Bit Feedback Compared to Alamouti
More informationEncode Decode Sample Quantize [ ] [ ]
Referece Audio Sigal Processig I Shyh-Kag Jeg Departmet of Electrical Egieerig/ Graduate Istitute of Commuicatio Egieerig M. Bosi ad R. E. Goldberg, Itroductio to Digital Audio Codig ad Stadards, Kluwer
More informationAPPLICATION NOTE UNDERSTANDING EFFECTIVE BITS
APPLICATION NOTE AN95091 INTRODUCTION UNDERSTANDING EFFECTIVE BITS Toy Girard, Sigatec, Desig ad Applicatios Egieer Oe criteria ofte used to evaluate a Aalog to Digital Coverter (ADC) or data acquisitio
More informationMeasurement of Equivalent Input Distortion AN 20
Measuremet of Equivalet Iput Distortio AN 2 Applicatio Note to the R&D SYSTEM Traditioal measuremets of harmoic distortio performed o loudspeakers reveal ot oly the symptoms of the oliearities but also
More informationAn Adaptive Image Denoising Method based on Thresholding
A Adaptive Image Deoisig Method based o Thresholdig HARI OM AND MANTOSH BISWAS Departmet of Computer Sciece & Egieerig Idia School of Mies, Dhabad Jharkad-86004 INDIA {hariom4idia, matoshb}@gmail.com Abstract
More informationCHAPTER 5 A NEAR-LOSSLESS RUN-LENGTH CODER
95 CHAPTER 5 A NEAR-LOSSLESS RUN-LENGTH CODER 5.1 GENERAL Ru-legth codig is a lossless image compressio techique, which produces modest compressio ratios. Oe way of icreasig the compressio ratio of a ru-legth
More informationCompound Controller for DC Motor Servo System Based on Inner-Loop Extended State Observer
BULGARIAN ACADEMY OF SCIENCES CYBERNETICS AND INFORMATION TECHNOLOGIES Volume 6, No 5 Special Issue o Applicatio of Advaced Computig ad Simulatio i Iformatio Systems Sofia 06 Prit ISSN: 3-970; Olie ISSN:
More informationInternational Journal of Scientific & Engineering Research, Volume 4, Issue 10, October ISSN
Iteratioal Joural of Scietific & Egieerig Research, Volume 4, Issue, October-3 48 A ovel Noise estimatio by histogram equalizatio ad optimal filterig for robust speech ehacemet M.Shravai, P.Chadra Sekhar,
More informationAnalysis of SDR GNSS Using MATLAB
Iteratioal Joural of Computer Techology ad Electroics Egieerig (IJCTEE) Volume 5, Issue 3, Jue 2015 Aalysis of SDR GNSS Usig MATLAB Abstract This paper explais a software defied radio global avigatio satellite
More informationA Novel Small Signal Power Line Quality Measurement System
IMTC 3 - Istrumetatio ad Measuremet Techology Coferece Vail, CO, USA, - May 3 A ovel Small Sigal Power Lie Quality Measuremet System Paul B. Crilly, Erik Leadro Boaldi, Levy Ely de Lacarda de Oliveira,
More informationA New Design of Log-Periodic Dipole Array (LPDA) Antenna
Joural of Commuicatio Egieerig, Vol., No., Ja.-Jue 0 67 A New Desig of Log-Periodic Dipole Array (LPDA) Atea Javad Ghalibafa, Seyed Mohammad Hashemi, ad Seyed Hassa Sedighy Departmet of Electrical Egieerig,
More informationX-Bar and S-Squared Charts
STATGRAPHICS Rev. 7/4/009 X-Bar ad S-Squared Charts Summary The X-Bar ad S-Squared Charts procedure creates cotrol charts for a sigle umeric variable where the data have bee collected i subgroups. It creates
More informationObjectives. Some Basic Terms. Analog and Digital Signals. Analog-to-digital conversion. Parameters of ADC process: Related terms
Objectives. A brief review of some basic, related terms 2. Aalog to digital coversio 3. Amplitude resolutio 4. Temporal resolutio 5. Measuremet error Some Basic Terms Error differece betwee a computed
More informationRadar emitter recognition method based on AdaBoost and decision tree Tang Xiaojing1, a, Chen Weigao1 and Zhu Weigang1 1
Advaces i Egieerig Research, volume 8 d Iteratioal Coferece o Automatio, Mechaical Cotrol ad Computatioal Egieerig (AMCCE 7) Radar emitter recogitio method based o AdaBoost ad decisio tree Tag Xiaojig,
More informationMeasurements of the Communications Environment in Medium Voltage Power Distribution Lines for Wide-Band Power Line Communications
Measuremets of the Commuicatios viromet i Medium Voltage Power Distributio Lies for Wide-Bad Power Lie Commuicatios Jae-Jo Lee *,Seug-Ji Choi *,Hui-Myoug Oh *, Wo-Tae Lee *, Kwa-Ho Kim * ad Dae-Youg Lee
More informationImage Contrast Enhancement based Sub-histogram Equalization Technique without Over-equalization Noise
World Academy of Sciece, Egieerig ad Techology Iteratioal Joural of Electrical ad Computer Egieerig Image Cotrast Ehacemet based Sub-histogram Equalizatio Techique without Over-equalizatio Noise Hyusup
More informationA SELECTIVE POINTER FORWARDING STRATEGY FOR LOCATION TRACKING IN PERSONAL COMMUNICATION SYSTEMS
A SELETIVE POINTE FOWADING STATEGY FO LOATION TAKING IN PESONAL OUNIATION SYSTES Seo G. hag ad hae Y. Lee Departmet of Idustrial Egieerig, KAIST 373-, Kusug-Dog, Taejo, Korea, 305-70 cylee@heuristic.kaist.ac.kr
More informationTehrani N Journal of Scientific and Engineering Research, 2018, 5(7):1-7
Available olie www.jsaer.com, 2018, 5(7):1-7 Research Article ISSN: 2394-2630 CODEN(USA): JSERBR 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38
More informationImage Denoising using Fast Bilateral Filter
Iteratioal Joural of Curret Egieerig ad Techology E-ISS 77 4106, P-ISS 347 5161 015IPRESSCO, All Rights Reserved Available at http://ipressco.com/category/ijcet Research Article Riku A Joseph * ad Maju
More informationFREQUENCY-WARPED TIME-WEIGHTED LINEAR PREDICTION FOR GLOTTAL VOCODING
FREQUENCY-WARPED TIME-WEIGHTED LINEAR PREDICTION FOR GLOTTAL VOCODING Mau Airaksie 1, Bajibabu Bollepalli 1, Joui Pohjalaie 2, Paavo Alku 1 1 Aalto Uiversity, Filad 2 Uiversity of Passau, Germay ABSTRACT
More informationImage Contrast Enhancement based Sub-histogram Equalization Technique without Over-equalization Noise
World Academy of Sciece, Egieerig ad Techology 5 9 Image Cotrast Ehacemet based Sub-histogram Equalizatio Techique without Over-equalizatio Noise Hyusup Yoo, Yougjoo Ha, ad Hersoo Hah Abstract I order
More informationLETTER A Novel Adaptive Channel Estimation Scheme for DS-CDMA
1274 LETTER A Novel Adaptive Chael Estimatio Scheme for DS-CDMA Che HE a), Member ad Xiao-xiag LI, Nomember SUMMARY This paper proposes a adaptive chael estimatio scheme, which uses differet movig average
More informationA study on the efficient compression algorithm of the voice/data integrated multiplexer
A study o the efficiet compressio algorithm of the voice/data itegrated multiplexer Gyou-Yo CHO' ad Dog-Ho CHO' * Dept. of Computer Egieerig. KyiigHee Uiv. Kiheugup Yogiku Kyuggido, KOREA 449-71 PHONE
More informationCOMPRESSION OF TRANSMULTIPLEXED ACOUSTIC SIGNALS
COMPRESSION OF TRANSMULTIPLEXED ACOUSTIC SIGNALS Mariusz Ziółko, Przemysław Sypka ad Bartosz Ziółko Departmet of Electroics, AGH Uiversity of Sciece ad Techology, al. Mickiewicza 3, 3-59 Kraków, Polad,
More informationHarnessing oversampling in correlation-coded OTDR
Haressig oversamplig i correlatio-coded OTDR Ruoli Liao, 1 Mig Tag, 1,* Ca Zhao, 1 Hao Wu, 1 Sogia Fu, 1 Demig Liu, 1 ad Perry Pig Shum 1 Wuha Natioal Laboratory for Optoelectroics (WNLO) ad Natioal Egieerig
More informationThe Detection of Abrupt Changes in Fatigue Data by Using Cumulative Sum (CUSUM) Method
Proceedigs of the th WSEAS Iteratioal Coferece o APPLIED ad THEORETICAL MECHANICS (MECHANICS '8) The Detectio of Abrupt Chages i Fatigue Data by Usig Cumulative Sum (CUSUM) Method Z. M. NOPIAH, M.N.BAHARIN,
More informationPHY-MAC dialogue with Multi-Packet Reception
PHY-AC dialogue with ulti-packet Receptio arc Realp 1 ad Aa I. Pérez-Neira 1 CTTC-Cetre Tecològic de Telecomuicacios de Cataluya Edifici Nexus C/Gra Capità, - 0803-Barceloa (Cataluya-Spai) marc.realp@cttc.es
More informationLaboratory Exercise 3: Dynamic System Response Laboratory Handout AME 250: Fundamentals of Measurements and Data Analysis
Laboratory Exercise 3: Dyamic System Respose Laboratory Hadout AME 50: Fudametals of Measuremets ad Data Aalysis Prepared by: Matthew Beigto Date exercises to be performed: Deliverables: Part I 1) Usig
More informationOutline. Motivation. Analog Functional Testing in Mixed-Signal Systems. Motivation and Background. Built-In Self-Test Architecture
Aalog Fuctioal Testig i Mixed-Sigal s Jie Qi Dept. of Electrical & Computer Egieerig Aubur Uiversity Co-Advisors: Charles Stroud ad Foster Dai Outlie Motivatio ad Backgroud Built-I Self-Test Architecture
More informationEstimation of reflection location by the correlation coefficient function
Acoustics 8 Paris Estimatio of reflectio locatio by the correlatio coefficiet fuctio H. Shibayama a, T. Araya b, Y. Maabe b ad E. Oarura b a Shibaura Istitute of Techology, 3-7-5, Toyosu, Koto-u, 35-8548
More informationSummary of Random Variable Concepts April 19, 2000
Summary of Radom Variable Cocepts April 9, 2000 his is a list of importat cocepts we have covered, rather tha a review that derives or explais them. he first ad primary viewpoit: A radom process is a idexed
More informationFingerprint Classification Based on Directional Image Constructed Using Wavelet Transform Domains
7 Figerprit Classificatio Based o Directioal Image Costructed Usig Wavelet Trasform Domais Musa Mohd Mokji, Syed Abd. Rahma Syed Abu Bakar, Zuwairie Ibrahim 3 Departmet of Microelectroic ad Computer Egieerig
More informationCS 201: Adversary arguments. This handout presents two lower bounds for selection problems using adversary arguments ëknu73,
CS 01 Schlag Jauary 6, 1999 Witer `99 CS 01: Adversary argumets This hadout presets two lower bouds for selectio problems usig adversary argumets ëku73, HS78, FG76ë. I these proofs a imagiary adversary
More informationSubband Coding of Speech Signals Using Decimation and Interpolation
3 th Iteratioal Coferece o AEROSPACE SCIENCES & AVIATION TECHNOLOGY, ASAT- 3, May 6 8, 9, E-Mail: asat@mtc.edu.eg Military Techical College, Kobry Elkobbah, Cairo, Egypt Tel : +() 459 43638, Fax: +() 698
More information(2) The MOSFET. Review of. Learning Outcome. (Metal-Oxide-Semiconductor Field Effect Transistor) 2.0) Field Effect Transistor (FET)
EEEB73 Electroics Aalysis & esig II () Review of The MOSFET (Metal-Oxide-Semicoductor Field Effect Trasistor) Referece: Neame, Chapter 3 ad Chapter 4 Learig Outcome Able to describe ad use the followig:
More informationIntroduction to CPM-OFDM: An Energy Efficient Multiple Access Transmission Scheme
Joural of Commuicatio ad Computer 1 (015) 37-43 doi: 10.1765/1548-7709/015.01.007 D DAVID PUBLISHING Itroductio to CPM-OFDM: A Eergy Efficiet Multiple Access Trasmissio Scheme Mohammad Irfa, Sag Hoo Lee
More informationDelta- Sigma Modulator with Signal Dependant Feedback Gain
Delta- Sigma Modulator with Sigal Depedat Feedback Gai K.Diwakar #1 ad V.Vioth Kumar *2 # Departmet of Electroics ad Commuicatio Egieerig * Departmet of Electroics ad Istrumetatio Egieerig Vel Tech Uiversity,Cheai,
More information3. Error Correcting Codes
3. Error Correctig Codes Refereces V. Bhargava, Forward Error Correctio Schemes for Digital Commuicatios, IEEE Commuicatios Magazie, Vol 21 No1 11 19, Jauary 1983 Mischa Schwartz, Iformatio Trasmissio
More informationA Data Field method for speech enhancement incorporating Binary Time-Frequency Masking
Jiaju HUANG Yafei ZHANG Xiogwei ZHANG Tao ZHU Istitute of Commad Automatio PLA Uiversity of Sciece ad Techology. A Data Field method for speech ehacemet icorporatig Biary Time-Frequecy Maskig Abstract.
More informationPROJECT #2 GENERIC ROBOT SIMULATOR
Uiversity of Missouri-Columbia Departmet of Electrical ad Computer Egieerig ECE 7330 Itroductio to Mechatroics ad Robotic Visio Fall, 2010 PROJECT #2 GENERIC ROBOT SIMULATOR Luis Alberto Rivera Estrada
More informationA Study on Performance Analysis for Error Probability in SWSK Systems
556 Tae-Il Jeog, et al.: A STUDY ON PERFORMANCE ANALYSIS FOR ERROR PROBABILITY IN SWSK SYSTEMS A Study o Performace Aalysis for Error Probability i SWSK Systems Tae-Il Jeog, wag-seo Moo, ad Jog-Nam Kim,
More informationApplication of Improved Genetic Algorithm to Two-side Assembly Line Balancing
206 3 rd Iteratioal Coferece o Mechaical, Idustrial, ad Maufacturig Egieerig (MIME 206) ISBN: 978--60595-33-7 Applicatio of Improved Geetic Algorithm to Two-side Assembly Lie Balacig Ximi Zhag, Qia Wag,
More informationx y z HD(x, y) + HD(y, z) HD(x, z)
Massachusetts Istitute of Techology Departmet of Electrical Egieerig ad Computer Sciece 6.02 Solutios to Chapter 5 Updated: February 16, 2012 Please sed iformatio about errors or omissios to hari; questios
More informationComparison of Frequency Offset Estimation Methods for OFDM Burst Transmission in the Selective Fading Channels
Compariso of Frequecy Offset Estimatio Methods for OFDM Burst Trasmissio i the Selective Fadig Chaels Zbigiew Długaszewski Istitute of Electroics ad Telecommuicatios Pozań Uiversity of Techology 60-965
More informationINCREASE OF STRAIN GAGE OUTPUT VOLTAGE SIGNALS ACCURACY USING VIRTUAL INSTRUMENT WITH HARMONIC EXCITATION
XIX IMEKO World Cogress Fudametal ad Applied Metrology September 6, 9, Lisbo, Portugal INCREASE OF STRAIN GAGE OUTPUT VOLTAGE SIGNALS ACCURACY USING VIRTUAL INSTRUMENT WITH HARMONIC EXCITATION Dalibor
More informationJoint Power Allocation and Beamforming for Cooperative Networks
It. J. Commuicatios, etwork ad System Scieces,, 4, 447-45 doi:.436/ijcs..4753 Published Olie July (http://www.scirp.org/joural/ijcs) Joit Power Allocatio ad Beamformig for Cooperative etworks Sodes Maadi,,
More information4. INTERSYMBOL INTERFERENCE
DATA COMMUNICATIONS 59 4. INTERSYMBOL INTERFERENCE 4.1 OBJECT The effects of restricted badwidth i basebad data trasmissio will be studied. Measuremets relative to itersymbol iterferece, usig the eye patter
More informationMassachusetts Institute of Technology Dept. of Electrical Engineering and Computer Science Fall Semester, Introduction to EECS 2.
Massachusetts Istitute of Techology Dept. of Electrical Egieerig ad Computer Sciece Fall Semester, 006 6.08 Itroductio to EECS Prelab Exercises Pre-Lab#3 Modulatio, demodulatio, ad filterig are itegral
More informationCHAPTER 8 JOINT PAPR REDUCTION AND ICI CANCELLATION IN OFDM SYSTEMS
CHAPTER 8 JOIT PAPR REDUCTIO AD ICI CACELLATIO I OFDM SYSTEMS Itercarrier Iterferece (ICI) is aother major issue i implemetig a OFDM system. As discussed i chapter 3, the OFDM subcarriers are arrowbad
More informationMid-level representations for audio content analysis *Slides for this lecture were created by Anssi Klapuri
Mid-level represetatios for audio cotet aalysis *Slides for this lecture were created by Assi Klapuri Represetatios 1 Sources: Ellis, Rosethal, Mid-level represetatios for Computatioal Auditory Scee Aalysis,
More informationOPTIMIZATION OF RNS FIR FILTERS FOR 6-INPUTS LUT BASED FPGAS
OPTIMIZATION OF RNS FIR FILTERS FOR 6-INPUTS LUT BASED FPGAS G.C. Cardarilli, M. Re, A. Salsao Uiversity of Rome Tor Vergata Departmet of Electroic Egieerig Via del Politecico 1 / 00133 / Rome / ITAL {marco.re,
More informationApplication of Image Fusion to Wireless Image Transmission
Applicatio of Image Fusio to Wireless Image Trasmissio Liae C. Ramac ad Pramod K. Varshey EECS Departmet, Syracuse Uiversity 121 Lik Hall, Syracuse, NY 13244 Email: lcramac@mailbox.syr.edu, varshey@cat.syr.edu
More informationFault Diagnosis in Rolling Element Bearing Using Filtered Vibration and Acoustic Signal
Volume 8 o. 8 208, 95-02 ISS: 3-8080 (prited versio); ISS: 34-3395 (o-lie versio) url: http://www.ijpam.eu ijpam.eu Fault Diagosis i Rollig Elemet Usig Filtered Vibratio ad Acoustic Sigal Sudarsa Sahoo,
More informationCombined Scheme for Fast PN Code Acquisition
13 th Iteratioal Coferece o AEROSPACE SCIENCES & AVIATION TECHNOLOGY, ASAT- 13, May 6 8, 009, E-Mail: asat@mtc.edu.eg Military Techical College, Kobry Elkobbah, Cairo, Egypt Tel : +(0) 4059 4036138, Fax:
More informationNonlinear System Identification Based on Reduced Complexity Volterra Models Guodong Jin1,a* and Libin Lu1,b
6th Iteratioal Coferece o Electroics, Mechaics, Culture ad Medicie (EMCM 205) Noliear System Idetificatio Based o Reduced Complexity Volterra Models Guodog Ji,a* ad Libi Lu,b Xi a research istitute of
More informationPRACTICAL FILTER DESIGN & IMPLEMENTATION LAB
1 of 7 PRACTICAL FILTER DESIGN & IMPLEMENTATION LAB BEFORE YOU BEGIN PREREQUISITE LABS Itroductio to Oscilloscope Itroductio to Arbitrary/Fuctio Geerator EXPECTED KNOWLEDGE Uderstadig of LTI systems. Laplace
More informationSIDELOBE SUPPRESSION IN OFDM SYSTEMS
SIDELOBE SUPPRESSION IN OFDM SYSTEMS Iva Cosovic Germa Aerospace Ceter (DLR), Ist. of Commuicatios ad Navigatio Oberpfaffehofe, 82234 Wesslig, Germay iva.cosovic@dlr.de Vijayasarathi Jaardhaam Muich Uiversity
More informationDesign of FPGA- Based SPWM Single Phase Full-Bridge Inverter
Desig of FPGA- Based SPWM Sigle Phase Full-Bridge Iverter Afarulrazi Abu Bakar 1, *,Md Zarafi Ahmad 1 ad Farrah Salwai Abdullah 1 1 Faculty of Electrical ad Electroic Egieerig, UTHM *Email:afarul@uthm.edu.my
More informationCancellation of Multiuser Interference due to Carrier Frequency Offsets in Uplink OFDMA
Cacellatio of Multiuser Iterferece due to Carrier Frequecy Offsets i Upli OFDMA S. Maohar, V. Tiiya, D. Sreedhar, ad A. Chocaligam Departmet of ECE, Idia Istitute of Sciece, Bagalore 56001, INDIA Abstract
More informationCFAR DETECTION IN MIMO RADARS USING FUZZY FUSION RULES IN HOMOGENEOUS BACKGROUND
CFAR DETECTION IN MIMO RADARS USING FUZZY FUSION RULES IN HOMOGENEOUS BACKGROUND Faycal Khaldi 1 ad Faouzi Soltai 2 1,2 Départemet d électroique, Uiversité des Frères Metouri Costatie Costatie 25, Algeria
More informationSpeech Enhancement Using Spectral Subtraction and Cascaded-Median Based Noise Estimation for Hearing Impaired Listeners
Speech Ehacemet Usig Spectral Subtractio ad Cascaded-Media Based Noise Estimatio for Hearig Impaired Listeers Satosh K. Waddi, Prem C. Padey, ad Nitya Tiwari Departmet of Electrical Egieerig Idia Istitute
More informationTotal Harmonics Distortion Reduction Using Adaptive, Weiner, and Kalman Filters
Wester Michiga Uiversity ScholarWorks at WMU Master's Theses Graduate College 6-2016 Total Harmoics Distortio Reductio Usig Adaptive, Weier, ad Kalma Filters Liqaa Alhafadhi Wester Michiga Uiversity, liquaa.alhafadhi@yahoo.com
More informationFPGA Implementation of the Ternary Pulse Compression Sequences
FPGA Implemetatio of the Terary Pulse Compressio Sequeces N.Balaji 1, M. Sriivasa rao, K.Subba Rao 3, S.P.Sigh 4 ad N. Madhusudhaa Reddy 4 Abstract Terary codes have bee widely used i radar ad commuicatio
More informationA Novel Method of Artificial Bandwidth Extension Using Deep Architecture
INTERSPEECH 2015 A Novel ethod of Artificial Badwidth Extesio Usig Deep Architecture Bi Liu, Jiahua Tao, Zhegqi We, Ya Li, Daish Bukhari Natioal Laboratory of Patter Recogitio, Istitute of Automatio, Chiese
More informationWAVE-BASED TRANSIENT ANALYSIS USING BLOCK NEWTON-JACOBI
WAVE-BASED TRANSIENT ANALYSIS USING BLOCK NEWTON-JACOBI Muhammad Kabir McGill Uiversity Departmet of Electrical ad Computer Egieerig Motreal, QC H3A 2A7 Email: muhammad.kabir@mail.mcgill.ca Carlos Christofferse
More informationRecovering low frequencies for impedance inversion by frequency domain deconvolution
Recoverig low frequecies for impedace iversio by frequecy domai decovolutio Sia. Esmaeili*, CREWES, Uiversity of Calgary, sesmaeil@ucalgary.ca Gary. Frak. Margrave, CREWES, Uiversity of Calgary, margrave@ucalgary.ca
More informationE X P E R I M E N T 13
E X P E R I M E N T 13 Stadig Waves o a Strig Produced by the Physics Staff at Colli College Copyright Colli College Physics Departmet. All Rights Reserved. Uiversity Physics, Exp 13: Stadig Waves o a
More informationAC : USING ELLIPTIC INTEGRALS AND FUNCTIONS TO STUDY LARGE-AMPLITUDE OSCILLATIONS OF A PENDULUM
AC 007-7: USING ELLIPTIC INTEGRALS AND FUNCTIONS TO STUDY LARGE-AMPLITUDE OSCILLATIONS OF A PENDULUM Josue Njock-Libii, Idiaa Uiversity-Purdue Uiversity-Fort Waye Josué Njock Libii is Associate Professor
More informationAkinwaJe, A.T., IbharaJu, F.T. and Arogundade, 0.1'. Department of Computer Sciences University of Agriculture, Abeokuta, Nigeria
COMPARATIVE ANALYSIS OF ARTIFICIAL NEURAL NETWORK'S BACK PROPAGATION ALGORITHM TO STATISTICAL LEAST SQURE METHOD IN SECURITY PREDICTION USING NIGERIAN STOCK EXCHANGE MARKET AkiwaJe, A.T., IbharaJu, F.T.
More informationSpread Spectrum Signal for Digital Communications
Wireless Iformatio Trasmissio System Lab. Spread Spectrum Sigal for Digital Commuicatios Istitute of Commuicatios Egieerig Natioal Su Yat-se Uiversity Spread Spectrum Commuicatios Defiitio: The trasmitted
More informationx 1 + x x n n = x 1 x 2 + x x n n = x 2 x 3 + x x n n = x 3 x 5 + x x n = x n
Sectio 6 7A Samplig Distributio of the Sample Meas To Create a Samplig Distributio of the Sample Meas take every possible sample of size from the distributio of x values ad the fid the mea of each sample
More informationProblem of calculating time delay between pulse arrivals
America Joural of Egieerig Research (AJER) 5 America Joural of Egieerig Research (AJER) e-issn: 3-847 p-issn : 3-936 Volume-4, Issue-4, pp-3-4 www.ajer.org Research Paper Problem of calculatig time delay
More informationSymbol Error Rate Evaluation for OFDM Systems with MPSK Modulation
Symbol Error Rate Evaluatio for OFDM Systems with MPS Modulatio Yuhog Wag ad Xiao-Pig Zhag Departmet of Electrical ad Computer Egieerig, Ryerso Uiversity 35 Victoria Street, Toroto, Otario, Caada, M5B
More informationA Reduced Complexity Channel Estimation for OFDM Systems with Precoding and Transmit Diversity in Mobile Wireless Channels Hlaing Minn, Dong In Kim an
A Reduced Complexity Chael Estimatio for OFDM Systems with Precodig ad Trasmit Diversity i Mobile Wireless Chaels Hlaig Mi, Dog I Kim ad Vijay K. Bhargava Departmet of Electrical ad Computer Egieerig,
More informationA New Energy Efficient Data Gathering Approach in Wireless Sensor Networks
Commuicatios ad Network, 0, 4, 6-7 http://dx.doi.org/0.436/c.0.4009 Published Olie February 0 (http://www.scirp.org/joural/c) A New Eergy Efficiet Data Gatherig Approach i Wireless Sesor Networks Jafar
More informationMulti-Carrier Transmission over Mobile Radio Channels. Jean-Paul M.G. Linnartz Philips Research and TU/e
Multi-Carrier Trasmissio over Mobile Radio Chaels Jea-Paul M.G. Liartz Philips Research ad TU/e Outlie Itroductio to OFDM Discussio of receivers for OFDM ad MC-CDMA Itercarrier Iterferece, FFT Leakage
More informationDIGITALLY TUNED SINUSOIDAL OSCILLATOR USING MULTIPLE- OUTPUT CURRENT OPERATIONAL AMPLIFIER FOR APPLICATIONS IN HIGH STABLE ACOUSTICAL GENERATORS
Molecular ad Quatum Acoustics vol. 7, (6) 95 DGTALL TUNED SNUSODAL OSCLLATOR USNG MULTPLE- OUTPUT CURRENT OPERATONAL AMPLFER FOR APPLCATONS N HGH STABLE ACOUSTCAL GENERATORS Lesław TOPÓR-KAMŃSK Faculty
More informationThe Firing Dispersion of Bullet Test Sample Analysis
Iteratioal Joural of Materials, Mechaics ad Maufacturig, Vol., No., Ma 5 The Firig Dispersio of Bullet Test Sample Aalsis Youliag Xu, Jubi Zhag, Li Ma, ad Yoghai Sha Udisputed, this approach does reduce
More informationUsing Color Histograms to Recognize People in Real Time Visual Surveillance
Usig Color Histograms to Recogize People i Real Time Visual Surveillace DANIEL WOJTASZEK, ROBERT LAGANIERE S.I.T.E. Uiversity of Ottawa, Ottawa, Otario CANADA daielw@site.uottawa.ca, lagaier@site.uottawa.ca
More information28.3. Kaushik Roy Dept. of ECE, Purdue University W. Lafayette, IN 47907, U. S. A.
8.3 Novel Sizig Algorithm for Yield Improvemet uder Process Variatio i Naometer Techology Seug Hoo Choi Itel Corporatio Hillsboro, OR 974, U. S. A. seug.h.choi@itel.com Bipul C. Paul Dept. of ECE, Purdue
More informationTechnical Explanation for Counters
Techical Explaatio for ers CSM_er_TG_E Itroductio What Is a er? A er is a device that couts the umber of objects or the umber of operatios. It is called a er because it couts the umber of ON/OFF sigals
More informationPRACTICAL ANALOG DESIGN TECHNIQUES
PRACTICAL ANALOG DESIGN TECHNIQUES SINGLE-SUPPLY AMPLIFIERS HIGH SPEED OP AMPS HIGH RESOLUTION SIGNAL CONDITIONING ADCs HIGH SPEED SAMPLING ADCs UNDERSAMPLING APPLICATIONS MULTICHANNEL APPLICATIONS OVERVOLTAGE
More informationSHORT-TERM TRAVEL TIME PREDICTION USING A NEURAL NETWORK
SHORT-TERM TRAVEL TIME PREDICTION USING A NEURAL NETWORK Giovai Huiske ad Eric va Berkum Dept. of Civil Egieerig - Uiversity of Twete - 7500 AE Eschede - The Netherlads E-mail: g.huiske@ctw.utwete.l ad
More informationAUDIO QUALITY MEASUREMENTS IN COMUNICATION SYSTEMS
AUDIO QUALITY MEASUREMETS I COMUICATIO SYSTEMS Ivo Matelja Faculty of electrical egieerig, Uiversity of Split R. Bosovica bb, 000 Split, Croatia (ivo.matelja@fesb.hr) Abstract: ITU gives P-class recommedatios
More informationCascaded Feedforward Sigma-delta Modulator for Wide Bandwidth Applications
Tamkag Joural of Sciece ad Egieerig, Vol. 4, No., pp. 55-64 () 55 Cascaded Feedforward Sigma-delta Modulator for Wide Badwidth Applicatios Je-Shiu Chiag, Teg-Hug Chag ad Pou-Chu Chou Departmet of Electrical
More informationBANDWIDTH AND GAIN ENHANCEMENT OF MULTIBAND FRACTAL ANTENNA BASED ON THE SIERPINSKI CARPET GEOMETRY
ISSN: 2229-6948(ONLINE) DOI: 10.21917/ijct.2013.0095 ICTACT JOURNAL ON COMMUNICATION TECHNOLOGY, MARCH 2013, VOLUME: 04, ISSUE: 01 BANDWIDTH AND GAIN ENHANCEMENT OF MULTIBAND FRACTAL ANTENNA BASED ON THE
More informationAN ESTIMATION OF MULTILEVEL INVERTER FED INDUCTION MOTOR DRIVE
9 IJRIC. All rights reserved. IJRIC www.ijric.org E-ISSN: 76-3336 AN ESTIMATION OF MULTILEVEL INVERTER FED INDUCTION MOTOR DRIVE K.RAMANI AND DR.A. KRISHNAN SMIEEE Seior Lecturer i the Departmet of EEE
More informationA Wide-Beam Broadcasting Antenna Using a Curved Dipole on Reflector Plane
1th WSEAS Iteratioal Coferece o COMMUNICATIONS, Heraklio, Greece, July 3-5, 8 A Wide-Beam Broadcastig Atea Usig a Curved Dipole o Reflector Plae RANGSAN WONGSAN School of Telecommuicatio Egieerig, Istitute
More informationHigh Speed Area Efficient Modulo 2 1
High Speed Area Efficiet Modulo 2 1 1-Soali Sigh (PG Scholar VLSI, RKDF Ist Bhopal M.P) 2- Mr. Maish Trivedi (HOD EC Departmet, RKDF Ist Bhopal M.P) Adder Abstract Modular adder is oe of the key compoets
More informationMarkov Modulated Punctured Autoregressive Processes for Traffic and Channel Modeling *
Markov Modulated Puctured Autoregressive Processes for Traffic ad Chael Modelig * Trista Pei-chu Che ad Tsuha Che + Departmet of Electrical ad Computer Egieerig Caregie Mello Uiversity, PA 1513 {peichu,
More informationLecture 14. Design of audio WM Model of digital audio CO ( WAV-format): - dependent samples with frequency 44,1кHz, - amplitude of samples is
Lecture 14. Desig of audio WM Model of digital audio CO WAV-format: - depedet samples wit frequecy 44,1кHz, - amplitude of samples is quatized o 2 16 =65536 levels 16 bits Remark 1. Tere exists a lossy
More informationDecode-forward and Compute-forward Coding Schemes for the Two-Way Relay Channel
Decode-forward ad Compute-forward Codig Schemes for the Two-Way Relay Chael Peg Zhog ad Mai Vu Departmet of Electrical ad Computer Egieerig McGill Uiversity Motreal, QC, Caada H3A A7 Emails: peg.zhog@mail.mcgill.ca,
More information10GBASE-T. length of precoding response, and PMA training
1GBASE-T TxFE solutios, dpsnr vs legth of precodig respose, ad PMA traiig IEEE P82.3a Task Force Austi, May 18-2, 25 Gottfried Ugerboeck 1 Cotets Study of trasmit frot-ed solutios Simple : o digital filterig,
More informationMEASUREMENT AND CONTORL OF TOTAL HARMONIC DISTORTION IN FREQUENCY RANGE 0,02-10KHZ.
ELECTRONICS 00 September, Sozopol, BLGARIA MEASREMENT AND CONTORL OF TOTAL HARMONIC DISTORTION IN FREQENCY RANGE 0,0-0KHZ. Plame Agelov Agelov Faculty for Computer Sciece, Egieerig ad Natural Studies,
More informationAdaptive MMSE Rake-Equalizer Receiver Design with Channel Estimation for DS-UWB System
Adaptive MMSE Rake-Equalizer Receiver Desig with Chael Estimatio for DS-UWB System Departmet of Electric Egieerig, Diwa Uiversity No.87-1, Nashi Li, Madou Tow, Taia Couty 7153, Taiwa, R.O.C. cyj@dwu.edu.tw
More informationMandarin Singing-voice Synthesis Using an HNM Based Scheme *
JOURNAL OF INFORMATION SCIENCE AND ENGINEERING 27, 303-317 (2011) Madari Sigig-voice Sythesis Usig a HNM Based Scheme * Departmet of Computer Sciece ad Iformatio Egieerig Natioal Taiwa Uiversity of Sciece
More informationStatistical Index Development from Time Domain for Rolling Element Bearings
Iteratioal Joural of Performability Egieerig Vol.0, No.3, May 04, pp.33-34. RAMS Cosultats Prited i Idia Statistical Idex Developmet from Time Domai for Rollig Elemet Bearigs YUAN FUQING * ad UDAY KUMAR
More informationELEC 350 Electronics I Fall 2014
ELEC 350 Electroics I Fall 04 Fial Exam Geeral Iformatio Rough breakdow of topic coverage: 0-5% JT fudametals ad regios of operatio 0-40% MOSFET fudametals biasig ad small-sigal modelig 0-5% iodes (p-juctio
More information