Department of Electronics and Communication Engineering

Drop-casted polypyrrole (PPY) nanomaterial-based point-of-care Traumatic Brain Injury (TBI) immunosensing platforms reported previously demand trained manpower at field-test, due to poor adhesion between nanomaterial and electrode surface, limiting the point-of-care purpose. The usage of conventional clean-room-based physical and chemical vapor deposition techniques in creating strong adhesion is limited on account of cost and process complexity. Addressing this technical gap, we report a novel low-cost clean-room-free technique that can effectively electrodeposit the PPY simultaneously onto the working areas of array of Interdigitated microelectrodes (ID?Es) from the precursor solution. Through optimization of deposition cycles and molar concentration ratio of monomer and oxidizing agents, a high-quality nanomaterial was electrodeposited on ID?Es' surface. Further, by using the electrodeposited PPY as a bioelectrical transducer, the TBI-specific UCHL1 and GFAP target analytes were simultaneously detected in terms of variation of DC-Resistance and AC-Capacitance parameters, recorded through chemiresistive I-V and chemicapacitive C-F responses of bioelectrodes, respectively. Such simultaneous multianalyte-detection in terms of multiple parameters increases the diversity of decision-making parameters by several folds, inherently aids in enhancing the diagnostic accuracy of TBI test kit. Here, the efficiency of the electrodeposited PPY-based chemiresistive and chemicapacitive immunosensing platforms in detecting TBI-specific target analytes simultaneously in real-time human-plasma samples was analyzed in terms of sensitivity, resolution, LoD, RoD, long-term stability (30 weeks), and the same is compared with drop-cast PPY-based immunosensing platform. Notably, the electrodeposited PPY sensing platforms showed superior performance in terms of sensitivity, LoD, device variability and long-term stability without demanding any trained manpower in the field

Automated early breast cancer detection has been credited as a lifesaver. In this work, an innovative approach based on the persistent homology diagram (PHD) is proposed. Every mammogram is processed using topological data analytic methods to generate its PHD and then the resized PHDs are analyzed for similarity using Earth mover's distance (EMD). The mammogram corpora obtained from SRM-Chennai Hospital with requisite clearance are analyzed for preliminary results. EMD from our earlier investigations has shown promising results when implemented independently on mammograms. We believe that knowledge of the topological structures obtained using the persistent diagrams can help identify the important structures and signatures in a mammogram and focus on a relevant region of interest. This additional processing layer can provide some interesting insights to offer while implementing an automated disease-tagging web service for breast cancer. The PHD will form the rationale for devising the strategy aimed to resolve the issue of missed- and misdiagnosis of breast cancer resulting in poor clinical prognosis at the community level. Furthermore, a web service-based or mobile-health approach promises to provide fruganomic point of care disease-tagging to the stakeholders at the bottom-end of the healthcare ecosystems residing in remote locations across the Indian subcontinent. The development of multimodal multisensory computational platforms incorporating digital signals from PHD-based image analytics of mammograms and novel biomarkers will form the rationale for large-scale screening of breast cancer patients at the community level.

In dentistry, endodontics is the study of dental pulp and tissues surrounding the roots. Endodontic treatment is otherwise called root canal treatment. The importance of endodontics focuses on several therapies to protect human teeth from cavities or infections, injuries, and various oral diseases like oral cancer and periodontal disease. Over 3.5 billion people are affected by various oral diseases, 10% of the global population is affected by periodontal diseases, and 530 million children suffer from tooth decay. There are different types of root canal morphology and configurations in which multiple abnormalities exist, such as C-shaped canals, fusion of roots, dens invaginatus, distolingual root, taurodontism, root dilaceration, etc. AI plays a vital role in endodontic applications. Using AI for the 98prediction and diagnosis of periapical lesions, root fractures can be detected. Nowadays, AI is used to determine working length measurements, predict dental pulp stem cells, and guide retreatment procedures. Therefore, AI provides successful outcomes and improvements in diagnosis and prediction in root canal applications in day-to-day practices. This review chapter summarizes different deep learning techniques that can be implemented in various endodontic applications in detail to understand the pros and cons

The ring oscillator (RO) is widely used to address different hardware security issues. For example, the RO-based physical unclonable function (PUF) generates a secure and reliable key for the cryptographic application, and the RO-based aging sensor is used for the efficient detection of recycled ICs. In this paper, a CMOS inverter with two voltage control signals is used to design a configurable RO (CRO). With its control signals, the proposed CRO can both accelerate and lower the impact of aging on the oscillation frequency. This vital feature of the proposed CRO makes it suitable for use in PUFs and RO-based sensors. The performance of both the proposed modified architecture, i.e., CRO PUF and CRO sensor, is evaluated in 90 nm CMOS technology. The aging tolerant feature of the proposed CRO enhances the reliability of CRO PUF. Similarly, the aging acceleration property of CRO improves the rate of detection of recycled ICs. Finally, both the proposed architectures are area and power-efficient compared to standard architectures

We propose herein a metamaterial (MM) dual-band THz sensor for various biomedical sensing applications. An MM is a material engineered to have a particular property that is rarely observed in naturally occurring materials with an aperiodic subwavelength arrangement. MM properties across a wide range of frequencies, like high sensitivity and quality factors, remain challenging to obtain. MM-based sensors are useful for the in vitro, non-destructive testing (NDT) of samples. The challenge lies in designing a narrow band resonator such that higher sensitivities can be achieved, which in turn allow for the sensing of ultra-low quantities. We propose a compact structure, consisting of a basic single-square split ring resonator (SRR) with an integrated inverted Z-shaped unit cell. The projected structure provides dual-band frequencies resonating at 0.75 THz and 1.01 THz with unity absorption at resonant peaks. The proposed structure exhibits a narrow bandwidth of 0.022 THz and 0.036 THz at resonances. The resonant frequency exhibits a shift in response to variations in the refractive index of the surrounding medium. This enables the detection of various biomolecules, including cancer cells, glucose, HIV-1, and M13 viruses. The refractive index varies between 1.35 and 1.40. Furthermore, the sensor is characterized by its performance, with an average sensitivity of 2.075 THz and a quality factor of 24.35, making it suitable for various biomedical sensing applications

Hyperspectral unmixing (HU) is crucial for extracting material information from hyperspectral images (HSI) obtained through remote sensing. Although linear unmixing methods are widely used due to their simplicity, they only address linear mixing effects. Nonlinear mixing models, while more complex, often focus solely on the nonlinear aspects affecting individual pixels. However, in practice, light reflected from materials within a pixel experiences linear and nonlinear interactions, necessitating a hybrid mixing model (HMM) that leverages spatial and spectral information. This work proposes a novel deep learning-based autoencoder (AE) with dual-stream decoders to enhance spectral unmixing. Our approach employs multitask learning (MTL) to process spatial and spectral information concurrently. Specifically, one decoder stream performs linear unmixing from HSI patches, while the other stream utilizes fully connected layers to capture and model the nonlinear interactions within the data. By integrating linear and nonlinear information, our method improves the accuracy of unmixing the mixed spectrum. We validate the effectiveness of our architecture on three real-world HSI datasets and compare its performance against various baseline methods. Experimental results consistently demonstrate that our approach outperforms existing methods, as evidenced by superior spectral angle distance (SAD) and mean squared error (MSE) metrics

The Smart Monitoring System for Pond Management and Automation in Aquaculture integrates state-of-the-art technology to optimize oxygen levels, monitor shrimp health, and automate feeding in aquaculture settings. By employing dissolved oxygen sensors, the system dynamically adjusts aeration fans to maintain optimal oxygen concentrations, ensuring an ideal environment for shrimp growth. Shrimp health is closely monitored using imaging sensors and advanced algorithms, which promptly alert users to any abnormalities, allowing for timely intervention. The automated feeding mechanism, incorporating feeders and weight sensors, manages feeding schedules and quantities, tailoring them to the specific needs of the shrimp population. A central monitoring system, operated through microcontrollers or PLCs, enables user interaction, real-time monitoring, and data storage for informed decision-making. Robust security measures, including authentication and encryption, safeguard sensitive data, while the integration of renewable energy sources enhances sustainability and reduces operational costs. This holistic approach represents a significant advancement in aquaculture technology, fostering resource efficiency, promoting shrimp health, and ultimately enhancing the overall success of aquaculture operations.

This study analyses the vertically stacked GAA Multi-Bridge-Channel FETs like Nanosheet at the device level for CMOS applications. Studies are carried out to validate the impact of geometric deviations concerning thickness and width of the FET's performance. The study also investigates the process parameter variation on DC metrics like threshold voltage (Vth), subthreshold swing (SS), ON-time (ION), OFF-time (IOFF), ION/IOFF, and DIBL. The device achieves better performance by optimizing Nanosheet width (NW) and thickness (NT) variability which ensures scaling flexibility. The CADENCE tool is used to investigate the device's performance in terms of circuit applications. Various circuits like CMOS inverter transient response, switching characteristics, voltage transfer characteristics (VTC) and noise margins are evaluated. The CMOS inverter energy delay product (EDP) and power delay product (PDP) are also analyzed. The PDP and EDP increase by 2.51x and 3.06x with rise of NW. The CMOS inverter noise margins (NMs) are calculated towards digital circuit applications. The proposed Nanosheet FET has good electrostatic integrity due to its GAA nature; thus, it is a strong contender for low-power applications for future technology nodes.

Cardiovascular diseases (CVD) are a predominant health concern globally, emphasizing the need for advanced diagnostic techniques. In our research, we present an avant-garde methodology that synergistically integrates ECG readings and retinal fundus images to facilitate the early disease tagging as well as triaging of the CVDs in the order of disease priority. Recognizing the intricate vascular network of the retina as a reflection of the cardiovascular system, alongwith the dynamic cardiac insights from ECG, we sought to provide a holistic diagnostic perspective. Initially, a Fast Fourier Transform (FFT) was applied to both the ECG and fundus images, transforming the data into the frequency domain. Subsequently, the Earth Mover’s Distance (EMD) was computed for the frequency-domain features of both modalities. These EMD values were then concatenated, forming a comprehensive feature set that was fed into a Neural Network classifier. This approach, leveraging the FFT’s spectral insights and EMD’s capability to capture nuanced data differences, offers a robust representation for CVD classification. Preliminary tests yielded a commendable accuracy of 84%, underscoring the potential of this combined diagnostic strategy. As we continue our research, we anticipate refining and validating the model further to enhance its clinical applicability in resource limited healthcare ecosystems prevalent across the Indian sub-continent and also the world at large.

We demonstrate the stable resistive switching (RS) and interesting neuromorphic features of Ag/Ni-HfO?/P??-Si memristors. This unique technique stacks a Ni-HfO? resistive switching (RS) layer on top of a P??-Si layer, considerably improving the stability, switching efficiency, and synaptic characteristics of memristors. A detailed physical model describes the RS filamentary process, which involves Ag+ ions migrating and forming electrical filaments with applied voltage, shifting the memristor consistent response from low-resistance and high-resistance phases. The memristor maintains consistent RS properties for 96 hours with low deterioration, because of the strong Ni-HfO? layer that improves switching stability. The memristor chip performs successfully in both voltage sweeping and pulse mode processes. The pulse-mode endurance results show that the low-resistance state (LRS) and high-resistance state (HRS) are stable after 100 cycles, with SET and RESET reaction times of 960 and 1636 ms, correspondingly. These findings show the memristors capacity for quick, energy-efficient switching. Furthermore, the memristor shows synaptic action, which resembles biological activities for example short-term (STP) and long-term plasticity (LTP). The conductivity regulation, like neurotransmitter release and synaptic weight correction, is accomplished by ion migration during voltage pulses. Also, the paired-pulse facilitation (PPF) reveals the memristors capacity to simulate synaptic activities, with a PPF index of 130%. The variations in pulse height and width indicate the progressive change from STP to LTP. Thus, the new device design indicates potential in neuromorphic computing, combining robust resistive switching with sophisticated synaptic properties to simulate essential brain activities such as memory retention and adaptation. These findings indicate that Ag/Ni-HfO?/P??-Si memristors have potential consistent switching efficiency and synaptic abilities serve as promising contenders for future artificial intelligence and computer hardware applications

Cuffless BP measurement system using a single lead electrocardiogram (ECG) and a photoplethysmography (PPG) measurement has been done. The method used is based on the fact that changes in blood pressure (BP) cause changes in Pulse Transit Time (PTT), which usually gets measured as the time difference between the R peak of electrocardiogram (ECG) signal and the peak of a Photoplethysmography (PPG) signal. We have tested the developed BP measurement system with three healthy volunteers and calibrated n subject specific model that predicts blood pressure based on PTT. Systolic Blood Pressure (SBP) and Diastolic Blood Pressure (DBP) are measured with a maximum mean error and standard deviation of±2.88 mmHg±1.96 mmHg and±4.7 mmHg±2.5 mmHg respectively

In seismic data processing, denoising is one of the essential steps to identifying the earth’s subsurface layer information. The noise present in the seismic data are categorized into two types: random and erratic noise. The random noise is distributed uniformly over the seismic data. The erratic noise attenuation is always challenging due to the unknown distribution of high-amplitude peaks over seismic data. The existing double sparsity dictionary learning (DSDL) method performs with analytical and adaptive transforms; both the transforms include iterative algorithms with K-SVD; it is computationally costly, and the dictionary is initialized with trained data. To address these limitations, we propose a novel method of dictionary learning with regularization (DLDR) to denoise both random and erratic noise from seismic data. In double regularization, we used with ?1-norm and nuclear norm. The denoised data is applied to the alternating direction method of multipliers (ADMM) to improve denoising while preserving the signal features from seismic data while reducing the computational cost. We evaluated the performance of the proposed method using signal-to-noise ratio (SNR), mean squared error (MSE), and local similarity map. The numerical results demonstrated that the proposed method resulted in higher SNR, lower MSE, and less signal leakage from seismic data. The method gives precise interpretation from the denoised seismic data

This manuscript presents an essential overview of the implementation of memristor-based logic circuits. Different memristor models such as Linear Drift, Simmons Tunnel, Yakopcic, N on-Linear ion Drift, TEAM, and VTEAM have been taken into consideration and a comparison of results has been done adopting all these models. The logic gates such as NAND, NOT, NOR, AND, OR along with some combinational circuits such as a l-bit comparator, and 2xl multiplexer were designed and implemented using MRL (Memristor Ratioed Logic) in the L TSpice simulation tool. The primary target of this paper is affirming the attributes of memristor at the device level. Here the logic circuits are implemented using Memristor and NMOS devices. All the circuits designed in this paper operate at a supply voltage of lV using HP memristor model. Moreover, the power consumed by the designed circuits has been analyzed and found to be in the order of p W

This paper examines the integration of Intelligent Reflecting Surfaces (IRS) in beyond 5G (B5G) communication networks, where the IRS reflects signals with adjustable phase shifts. By optimizing these phase shifts, called passive beamforming, substantial improvements in communication performance can be achieved. We maximize energy efficiency in the uplink communication, utilizing the IRS. However, including an IRS introduces complexities, particularly in channel estimation. To address this, we examine two innovative approaches to minimize the channel estimation overhead: the first leverages a grouping strategy for the reflecting elements. In contrast, the second approach utilizes positioned-based phase optimization. Simulation results confirm that the IRS significantly enhances energy efficiency compared to the traditional system

Tera Hertz photoconductive antennas (THz PCAs) have significantly advanced the THz research by offering room-temperature operation, broad bandwidth and relatively low cost as both emitters and detectors. However, the primary limitation has been their low power output due to inefficient conversion. This article demonstrates a substantial improvement in efficiency (?200%) by incorporating sub-micron photonic structures on the surface. These photonic structures enhance pump beam coupling, leading to increased photocarrier generation. They also facilitate efficient carrier recombination after THz emission, thereby suppressing carrier screening. Experimental and numerical studies confirm the enhanced photocarrier generation and controlled transport through defect-free paths, further reducing screening effects. The integration of photonic structures into large area emitters (LAEs) holds the potential to develop emitters and detectors suitable for real-world THz systems, overcoming the limitations of the current commercial LAEs that rely on plasmonic structures or antireflection coatings. This innovation has the potential to revolutionise THz technology, enabling the development of more powerful and efficient THz sources and detectors. This can lead to advancements in various fields, including wireless communication, imaging and sensing and spectroscopy. © Indian Academy of Sciences 2025

It is important in the rising demands to have efficient anomaly detection in camera surveillance systems for improving public safety in a complex environment. Most of the available methods usually fail to capture the long-term temporal dependencies and spatial correlations, especially in dynamic multi-camera settings. Also, many traditional methods rely heavily on large labeled datasets, generalizing poorly when encountering unseen anomalies in the process. We introduce a new framework to address such challenges by incorporating state-of-the-art deep learning models that improve temporal and spatial context modeling. We combine RNNs with GATs to model long-term dependencies across cameras effectively distributed over space. The Transformer-Augmented RNN allows for a better way than standard RNNs through self-attention mechanisms to improve robust temporal modeling. We employ a Multimodal Variational Autoencoder-MVAE that fuses video, audio, and motion sensor information in a manner resistant to noise and missing samples. To address the challenge of having a few labeled anomalies, we apply the Prototypical Networks to perform few-shot learning and enable generalization based on a few examples. Then, a Spatiotemporal Autoencoder is adopted to realize unsupervised anomaly detection by learning normal behavior patterns and deviations from them as anomalies. The methods proposed here yield significant improvements of about 10% to 15% in precision, recall, and F1-scores over traditional models. Further, the generalization capability of the framework to unseen anomalies, up to a gain of +20% on novel event detection, represents a major advancement for real-world surveillance systems

Resource-constrained Internet of Things (IoT) edge devices demand lightweight, energy efficient, and secure cipher designs with CMOS technology scaling to enhance hardware security. This work proposes and demonstrates for the first time the potential and challenges of using NCFETs for energy efficient and secure S-box design used in lightweight ciphers exploring the Feistel network structure at VDD = 0.5 V. Performance benchmarking is performed for the proposed NCFET-based S-box design of a Feistel network SLIM cipher with a baseline CMOS SLIM cipher and other existing NCFET PRESENT Cipher with Substitution and Permutation (SPN) networks. The proposed NCFET S-box design exploits the unique steep slope device characteristics and increases non-linearity in power traces caused by the extra gate capacitance of the NCFETs along with the highly secure Feistel network structure to enhance overall energy efficiency and DPA attack resiliency. A thorough DPA resiliency analysis of the proposed S-box design with performance metrics such as SNR, MTD, and SPD performance comparison with the baseline CMOS design and other state-of-the-art S-box designs has been performed. Performance benchmarking of the proposed S-box design of an ultra-lightweight NCFET-based SLIM cipher design with an equivalent baseline CMOS design shows ~4.25× lower energy consumption, a 16× increase in the attacker effect ratio, a ~3.7× reduction in signal-to-noise ratio (SNR) values, a 16× increase in the minimum traces to disclosure (MTD) value, and a ~13.4× higher security power delay (SPD) value at VDD = 0.5 V.

This paper proposes a cooperative reinforcement learning-based framework to jointly optimize active and passive beamforming in a passive Intelligent Reflecting Surface (PIRS)-assisted wireless communication system for green and secured communications. The framework employs two Deep Deterministic Policy Gradient (DDPG) agents: one at the Base Station (BS) for active beamforming control and the other at the PIRS for phase shift adjustments in passive beamforming. The BS agent optimizes beamforming for both Line-of-Sight (LOS) and Non-Line-of-Sight (NLOS) paths, while the PIRS agent adjusts phase shifts to improve the constructive contribution of the reflected signals. The user assesses the combined direct and reflected signals, using a secure rate (Rsec) based reward to guide the learning process of both agents. Through channel state information (CSI) from BS-PIRS, PIRS-user, and BS-user links, the agents learn coordinated actions to maximize the secure rate, boosting signal strength for the intended user and reducing eavesdropping risks. Simulations reveal that the proposed framework achieves substantial secured data rate efficiency gains with BS antenna configurations of 4, 8, and 16. However, further increases in antenna count require BS power adjustments for optimal performance. This joint optimization approach significantly improves secure rate and signal quality, positioning it as a valuable solution for next-generation wireless networks, such as 6G, that demand high data rates, enhanced security, and reliable connectivity

Diabetes management requires frequent blood glucose monitoring, yet invasive procedures impede testing. A noninvasive approach to detect random blood glucose (RBG) is crucial for early diagnosis and timely treatment. This work leverages Photoacoustic Spectroscopy for the detection of RBG due to its high sensitivity, specificity, and real-time monitoring capabilities. Therefore PAS has been implemented with a shallow dense neural network using a hybrid loss function (logcosh + huber loss) to estimate RBG. The augmentation of blood glucose is obtained by integrating biological parameters of a person like Body Mass Index, Age, and Spo2 with photoacoustic signal values. The intended hardware setup integrates with a Raspberry Pi 4 enabling real-time monitoring through the Thingspeak cloud platform. Testing with 105 in vivo samples demonstrated accuracies of 2.86 mg/dl (RMSE), 8.77 mg/dl (MAD), and 8.49% (MARD). Overall, an IoT-based PAS portable device is designed to provide smart healthcare services and quality care improvement

Precise segmentation of retinal blood vessels (RBVs) is pivotal in ophthalmology research, aiding in detecting diverse retinal abnormalities. This study proposes a contrast-limited adaptive histogram equalisation (CLAHE) technique to improve retinal image quality and visibility of microvascular structures. We aimed to determine the complexity of blood vessels using fractal dimensions (FD) and compare different metrics for their effectiveness. We employed the UNet architecture to separate blood vessels, and our results on the DRIVE retinal fundus image standard dataset showed an impressive accuracy rate of 97.24%, surpassing traditional filtering methods. Box counting, information, capacity, correlation, and probability dimensions are used in the FD analysis to help us understand the complex and irregular structures of retinal blood vessels. These metrics are valuable for detecting and monitoring retinal diseases in clinical settings. Our comparison with other techniques reveals promising results, particularly in the capacity and information dimensions, with statistical significance (P < 0.05). The potential of fractal dimensions as a screening tool for diabetic retinopathy underscores their importance in epidemiological studies