Department of Electrical Engineering

Indian Institute of Technology Palakkad&nbsp;(IIT Palakkad or IIT PKD) is a public autonomous&nbsp;engineering&nbsp;and&nbsp;research&nbsp;institute located in&nbsp;Palakkad,&nbsp;Kerala. It was&nbsp;one of the five new&nbsp;IITs&nbsp;proposed in the&nbsp;2014 Union budget of India. The campus was inaugurated on 3 August 2015. The institute has 94 faculty, 978 students, and 63 non-teaching staffs.

IT Palakkad currently offers four-year Bachelor of Technology (B.Tech) and two years MS programmes in 8 different engineering disciplines including&nbsp;Computer Science, Electrical, Mechanical, Civil among others.

&nbsp;

IIT Palakkad

Automatic audio event recognition (AER) plays a major role in designing and building intelligent location and context-aware applications including audio surveillance, audio indexing and content retrieval, highlight extraction, drone and robotic navigation, machine health monitoring, audio-aware voice processing services, and urban sound pollution monitoring. In this paper, we present audio event recognition (AER) schemes using the Mel-frequency cepstral coefficients (MFCC) and machine classifiers such as multi-class support vectors machines (MC-SVM), fully connected feed-forward neural networks (FCFFNNs), and one-dimensional convolutional neural networks (1D-CNNs) that are capable of automatically recognizing seven sound classes including aircraft, construction, music, nature (wind and rain), speech, vehicle, and train. In this study, we created large scale audio database for both training and testing purposes. The performance of the three AER schemes are evaluated under different audio frame sizes (100 ms, 250 ms and 500 ms) using a wide variety of sounds recorded using different kinds of recording devices. Results show that the FCFFNN and 1D-CNN based AER schemes had the F1-score values of 95.72% and 96.34% for audio frame size of 250 ms whereas MC-SVM based AER scheme had the F1-score value of 85.84%. The 1D-CNN based AER scheme had a class-wise accuracy is greater than 84% for audio frame size of 250 ms whereas the FC-FFNN based scheme had a class-wise accuracy is greater than 80% for audio frame size of 250 ms. The computational analysis results show that the prediction time of 1D-CNN based scheme is faster than the FC-FFNN based AER scheme. © 2019 IEEE.

World University root node

Proceedings - 2019 7th International Conference on Digital Information Processing and Communications, ICDIPC 2019

Automatic Audio Event Recognition Schemes for Context-Aware Audio Computing Devices

In dual inverter fed open end winding induction motor drives (OEWIM), the reference voltage can be synthesized either by level shifted carrier based space vector modulation (SVM) techniques or by using decoupled SVM methods. In level shifted SVM, each phase leg is switched for half the fundamental cycle and therefore has lower switching loss compared to decoupled methods. Bus clamping sequence (CS) pulse width modulation (PWM) technique also results in reduced switching loss. In this paper, decoupled and level shifted SVM methods are compared in terms of switching loss, common mode voltage (CMV) switching and stator current ripple. From the analysis, it is observed that level shifted clamping sequence SVM for a dual inverter fed OEWIM results in reduced switching loss and reduced number of CMV switching at the cost of increased stator current ripple. Compared to decoupled clamping sequence, level shifted clamping sequence has nearly 50% reduction in switching loss and CMV. Also, clamping sequence applied to decoupled method results in improved harmonic performance but has moderately increased switching loss. Various SVM methods are simulated in MATLAB Simulink for an open-loop volts per hertz (V/f) controlled induction motor drive with fixed number of samples per sector.

1st IEEE International Conference on Sustainable Energy Technologies and Systems, ICSETS 2019

Evaluation of Modulation Methods on Switching Loss, Common Mode Voltage and Current Ripple for an Open End Winding Induction Motor Drive

Latency is the one of the critical performance metrics for 5G and beyond mobile networks, particularly for ultra-reliable and low-latency communications (URLLC). In URLLC applications, it is required that the transmitted packets reach the destination within a certain time and the packets that are unable to meet this strict latency requirement will be discarded. In this paper, we compute the waiting time in the packetized fronthaul at the Ethernet switch and compute packet loss rate (PLR) incurred due to the inability of the transmitted packets to meet the FH latency threshold. In addition, we derive the tractable closed-form solution for the waiting time distribution and verify it with simulation results. Our results show that PLR is affected mainly by packet size, spectral efficiency, switch speed and arrival rate. © 2019 IEEE.

IEEE International Symposium on Personal, Indoor and Mobile Radio Communications, PIMRC

Packet Loss in Latency-constrained Ethernet-based Packetized C-RAN Fronthaul

Phased array weather radars (PAWRs) are increasingly becoming viable because of their operational advantage in the agile scanning of large precipitation volumes without any mechanical motion and an efficient beam scheduling-and-tracking. Conventional PAWR estimates precipitation profiles through Fourier beamforming. However, this method is not robust to interference, clutter and mismatch in steering vectors. In this work, we offset these disadvantages by using advanced robust and adaptive beamformers in PAWR. The volumetric nature of the precipitation target makes direct application of these techniques to PAWR non-trivial. We validate our methods through data collected from the solid-state low-power X-band PAWR developed at IIT Palakkad. © 2019 URSI. All rights reserved.

2019 URSI Asia-Pacific Radio Science Conference, AP-RASC 2019

Robust beamforming for X-band phased array weather radar

This brief proposes robust adaptive filtering algorithms and their VLSI architectures for sparse system identification under impulsive noise. Several robust algorithms are derived by combining error nonlinear adaptive filtering algorithms with proportionate adaptation. We make a comparative study of the derived algorithms and their VLSI architectures in terms of convergence rate and hardware complexity to show that the hardware overhead is negligible for the achieved improvement in robustness. © 1993-2012 IEEE.

IEEE Transactions on Very Large Scale Integration (VLSI) Systems

Robust proportionate adaptive filter architectures under impulsive noise

State-dependent jump linear systems (SJLSs) are a set of linear systems whose switching is determined by a finite state random process with state-dependent transition rates. In this study, a SJLS is considered with state multiplicative noise and stochastic perturbations. In particular, the jump process is regarded to have dissimilar transition rates among sets of state evolution space. The aim of this study is to consider stochastic ℋ∞ control of such systems using state-feedback control input. Sufficient conditions for stochastic ℋ∞ are obtained by solving linear matrix inequalities, which are validated by a simulation example. © The Institution of Engineering and Technology 2018.

Fulltext

IET Control Theory and Applications

Stochastic ℋ∞ control of state-dependent jump linear systems with state-dependent noise

For a linear system having event based measurements and correlated noises, a state estimation algorithm is proposed. A general event based sampling is employed to obtain the measurements, where in the case of unavailability of measurements, event based strategy itself is used to obtain approximate state and covariance estimates. To deal with correlated noises, a two-step ahead prediction approach is employed to obtain recursive equations for estimated state and covariance. The obtained results are illustrated using a simulation example. © 2019

Journal of the Franklin Institute

Event based estimation with correlated noises

Wideband Rate Adaptation and Scheduling in OFDM Cellular Systems: Modeling, Analysis and Base Station Side Estimation

Most on-device and cloud processing based automatic speech recognition (ASR) systems had poor recognition performance due to the noisy speech signals corrupted by various kinds of background noises such as vehicle, train, aircraft, fan, wind, rain, air-conditioner, and machinery noises which are unavoidable realistic scenarios. In this paper, we propose a novel speech signal quality assessment (SSQA) method for automatically assessing the quality of a recorded speech signal before processing on-device and sending the recorded data to the cloud server. The proposed method is based on the spectrogram feature and two-dimensional convolutional neural networks (2D-CNNs). The proposed SSQA method is evaluated using a large scale of noise-free speech and noisy speech signals which are corrupted with various kinds of noises with different noise levels. Results show that the 2D-CNN based method had an average Se=90.92%, Sp=98.44% and OA =96.44%. The method had better results in detecting the noisy speech segments. Results showed that there is confusion in performing the manual labelling of noise-free and noisy speech segments. Therefore, the noise-free and noisy speech signals are given to the publicly available ASR system to obtain the corresponding text. Then the word error rate (WER) and character error rate (CER) metrics were used to know the level of noise wherein the ASR system fails to correctly recognize its text. In this way, the noise level is determined for each of the noises to label the recorded speech signal into acceptable and unacceptable speech segments. The proposed quality-aware ASR system has great potential in improving the lifetime of the battery of the portable ASR devices and reducing the bandwidth and speech recognition software utilization costs in the case of cloud processing based ASR system. © 2019 IEEE.

SSQA: Speech Signal Quality Assessment Method using Spectrogram and 2-D Convolutional Neural Networks for Improving Efficiency of ASR Devices

In this paper, specific emitter identification(SEI) problem has been investigated over fading channel. Different emitters are distinguished by the non-linear nature of the power amplifiers used by the emitters. This non-linearity is the fingerprint for the specific emitters. Based on this non-linearity SEI problem is solved. In this paper, we have considered several features like energy entropy, first and second order moment and correlation coefficient, which are calculated from the intrinsic mode functions(IMFs) obtained by using Variational mode decomposition(VMD). The SEI performance using EMD-EM2 method is compared with VMD-EM2 over fading channel, where entropy and first, second order moments are calculated from the Hilbert spectrum. The performance in terms of probability of correct identification for the second method is better than the first method for both single-hop and relaying case. Also the performance of VMD-CB is superior to VMD-EM2. © 2019 IEEE.