Faculty Publications
Permanent URI for this communityhttps://idr.nitk.ac.in/handle/123456789/18736
Publications by NITK Faculty
Browse
15 results
Search Results
Item Estimation of evolutionary spectrum based on STFT and modified group delay(2003) Narasimhan, S.V.; Pavanalatha, S.This paper proposes a new estimator for Evolutionary Spectrum (ES) based on short time Fourier transform (STFT) and modified group delay (MGD). Here, the STFT enables crossterm suppression and the MGD preserves the frequency resolution of the rectangular window. It is applicable to deterministic and random signals generated by time varying systems. The proposed method provides signal to noise ratio enhancement due to the use of MGD. The results indicate that for linear chirp signals and for time varying random process, its frequency resolution is close to that of WVD and better than Evolutionary periodogram (EP) and STFT. Further, its noise immunity is better than those of EP and STFT.Item Autoregressive modeling of the Wigner-Ville distribution based on signal decomposition and modified group delay(2004) Nayak, M.B.; Narasimhan, S.V.An autoregressive modeling of the Wigner-Ville distribution (WVD), based on signal decomposition (SD) by a perfect reconstruction filter bank (PRFB) and the modified magnitude group delay function (MMGD), has been proposed. The SD and MMGD, respectively, reduce the existence of crossterms (without any time smoothing) and the Gibb's ripple effect (due to truncation of the WVD kernel, without applying any window), significantly. In view of this, the modeling is not affected by either the crossterms or the Gibb's ripple and the window that would have been used. The proposed method represents actual time-frequency information parsimoniously and compared to the existing WVD modeling methods, its performance is significantly better in terms of both time and frequency resolution (as there is no time and frequency smoothing) and noise immunity/variance and is computationally efficient. © 2003 Elsevier B.V. All rights reserved.Item Estimation of evolutionary spectrum based on short time Fourier transform and modified group delay(Elsevier, 2004) Narasimhan, S.V.; Pavanalatha, S.This paper proposes a new estimator for evolutionary spectrum (ES) based on short time Fourier transform (STFT) and modified group delay function (GDFM). The STFT due to its built-in averaging suppresses the crossterms and the GDFM preserves the frequency resolution of the rectangular window as it reduces the Gibbs ripple without using any window function. The new estimator is applicable to random signals as the GDFM removes the effect of the zeros due to input noise driving the time-varying system and provides the system information effectively. The GDFM also provides signal-to-noise ratio enhancement as it removes the zeros due to the associated noise. The performance of the method is illustrated for linear chirp signals, frequency shift keying and for time-varying random process which indicate that its frequency resolution is better than evolutionary periodogram (EP) and STFT and nearer to that of Wigner Ville distribution. Further, its noise immunity is better than those of EP and STFT. © 2004 Elsevier B.V. All rights reserved.Item Harmonic wavelet transform signal decomposition and modified group delay for improved Wigner-Ville distribution(2004) Narasimhan, S.V.; Shreyamsha Kumar, B.K.S.A new approach for the Wigner-Ville Distribution (WVD) based on signal decomposition by harmonic wavelet transform (SDHWT) and the modified magnitude group delay function (MMGD) has been proposed. The SDHWT directly provides subband signals and the WVD of these components are concatenated to get the overall WVD without using antialias and image rejection filtering. The SDHWT and the MMGD remove the existence of crossterms (CT) and the ripple effect due to truncation of the WVD kernel without applying any window, respectively. Since there is no time and frequency smoothing, the proposed method has a better performance in terms of both time and frequency resolution and desirable properties of a time-frequency representation (TFR) than the Pseudo WVD (PWVD). Further, it has a relatively better noise immunity compared to that of PWVD. In the WVD, for signal decomposition, the use of SDHWT, compared to that of a filter bank, provides almost similar results but has a significant (72%) computational advantage. © 2004 IEEE.Item Virtual sensor based Feedback Active Noise Control for neonates in NICU(IEEE Computer Society, 2013) Veena, S.; Pavithra, S.; Lokesha, H.; Narasimhan, S.V.; More, N.V.In this paper, a novel approach based on Feedback Active Noise Control with Virtual sensing has been proposed to reduce the Neonatal Intensive Care Unit (NICU) noise for infants in incubator. In NICU, the noise is due to infant monitoring medical equipments located in the vicinity of incubators. To address this, the algorithms in the literature are based on feedforward approach which requires a good reference to achieve effective attenuation. In an NICU environment, these algorithms require more than one reference microphone, which increases the complexity of the algorithm. This is eliminated by the proposed Feedback Active Noise Control approach as it generates its own reference from the error signal. The baby's ears must fall in the zone of silence (ZOS) to ensure noise reduction at its ears. But the volume of ZOS is inversely proportional to noise frequency and for frequencies above 1 KHz the ZOS is less than an inch. This constraint is overcome by the virtual sensing technique, which focuses the ZOS at baby's ears. The algorithm is evaluated with noise record from Neonatal Intensive Care Unit and it has resulted in 7dB more reduction at the baby's ears compared to existing algorithms. © 2013 IEEE.Item An improved system blind identification method based on second-order cyclostationary statistics and the properties of group delay, has been proposed. This is achieved by applying a correction to the estimated phase (by the spectral correlation density of the system output) for the poles, in the group delay domain. The results indicate a significant improvement in system blind identification, in terms of root mean square error. Depending upon the signal-to-noise ratio, the improvement in percentage normalized mean square error ranges between 20 and 50%.(Improved system blind identification based on second-order cyclostationary statistics: A group delay approach) Giridhar, P.V.S.; Narasimhan, S.V.2000Item A new approach has been proposed for improving the performance of the Wigner-Ville distribution. This approach is based on signal decomposition and modified magnitude group delay function. Signal decomposition achieved by perfect reconstruction filter bank reduces significantly the existence of crossterms. The Gibbs ripple effect is due to truncation of the Wigner-Ville distribution kernel. The modified magnitude group delay function overcomes this effect without applying any window. Compared to those of Pseudo Wigner-Ville distribution and its versions, the proposed method has significantly improved performance in both time and frequency resolution as there is no time and frequency smoothing. Further, this method obeys better the desirable properties of time-frequency representation and has a better noise immunity. © 2003 Elsevier B.V. All rights reserved.(Improved Wigner-Ville distribution performance by signal decomposition and modified group delay) Narasimhan, S.V.; Nayak, M.B.2003Item A new approach for channel blind identification based on second order cyclostationary statistics and the group delay has been proposed. In this, two methods are proposed. In both the methods, the correction is applied to the basic phase estimate for both the poles and zeros, in the group delay domain. The basic phase estimate is derived from the spectral correlation density (SCD) of the system output. In the first method, the phase correction is based on magnitude group delay. In the second method, not only the phase correction but also an improved system magnitude estimate of better variance and frequency resolution is derived based on modified magnitude group delay. The results indicate a significant improvement in performance for both the methods. For the first method in the absence of noise, the percentage normalized mean square error is reduced by about 85% over that of the existing non-parametric method. The second method in the presence of noise (SNR=5 dB), provides a reduction of 74% over the existing non-parametric method and 57% over the existing combined parametric and non-parametric methods. © 2005 Elsevier B.V. All rights reserved.(Channel blind identification based on cyclostationarity and group delay) Narasimhan, S.V.; Hazarathaiah, M.; Giridhar, P.V.S.2005Item Improved Wigner-Ville distribution performance based on DCT/DFT harmonic wavelet transform and modified magnitude group delay(2008) Narasimhan, S.V.; Haripriya, A.R.; Shreyamsha Kumar, B.K.A new Wigner-Ville distribution (WVD) estimation is proposed. This improved and efficient WVD is based on signal decomposition (SD) by DCT or DFT harmonic wavelet transform (DCTHWT or DFTHWT) and the modified magnitude group delay (MMGD). The MMGD processing can be either in fullband or subband. The SD by DCTHWT provides better quality low leakage decimated subband components. The concatenation of WVDs of the subbands results in an overall WVD, significantly free from crossterms and Gibbs ripple. As no smoothing window is used for the instantaneous autocorrelation (IACR), MMGD removes or reduces the Gibbs ripple preserving the frequency resolution achieved by the DCT/DFT HWT. The SD by DCTHWT compared to that of DFTHWT, has improved frequency resolution and detectability. These are due to the symmetrical data extension and the consequential low leakage (bias and variance). As the zeros due to the associated white noise are removed by the MMGD effectively in subband domain than in fullband, the proposed WVD based on subband has a better noise immunity. Compared to fullband WVD, the subband WVD is computationally efficient and achieves a significantly better: frequency resolution, detectability of low-level signal in the presence of high-level one and variance. The SD-based methods, however cannot bring out the frequency transition path from band to band clearly, as there will be gap in the contour plot at the transition. For the proposed methods, the heart rate variability (HRV) real data is also considered as an example. © 2007 Elsevier B.V. All rights reserved.Item Improved phase estimation based on complete bispectrum and modified group delay(2008) Narasimhan, S.V.; Basumallick, N.; Chaitanya, R.In this paper, a new method for extracting the system phase from the bispectrum of the system output has been proposed. This is based on the complete bispectral data computed in the frequency domain and modified group delay. The frequency domain bispectrum computation improves the frequency resolution and the modified group delay reduces the variance preserving the frequency resolution. The use of full bispectral data also reduces the variance as it is used for averaging. For the proposed method at a signal to noise ratio of 5dB, the reduction in root mean square error is in the range of 1.5-7 times over the other methods considered. © 2008 Springer-Verlag London Limited.