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

Photoacoustic Spectroscopy (PAS) is a potential method for the noninvasive detection of blood glucose. However random blood glucose testing can help to diagnose diabetes at an early stage and is crucial for managing and preventing complications with diabetes. In order to improve the diagnosis, control, and treatment of Diabetes Mellitus, an appropriate approach of noninvasive random blood glucose is required for glucose monitoring. A polynomial kernel-based ridge regression is proposed in this paper to detect random blood glucose accurately using PAS. Additionally, we explored the impact of the biological parameter BMI on the regulation of blood glucose, as it serves as the primary source of energy for the body’s cells. The kernel function plays a pivotal role in kernel ridge regression as it enables the algorithm to capture intricate non-linear associations between input and output variables. Using a Pulsed Laser source with a wavelength of 905 nm, a noninvasive portable device has been developed to collect the Photoacoustic (PA) signal from a finger. A collection of 105 individual random blood glucose samples was obtained and their accuracy was assessed using three metrics: Root Mean Square Error (RMSE), Mean Absolute Difference (MAD), and Mean Absolute Relative Difference (MARD). The respective values for these metrics were found to be 10.94 (mg/dl), 10.15 (mg/dl), and 8.86%. The performance of the readings was evaluated through Clarke Error Grid Analysis and Bland Altman Plot, demonstrating that the obtained readings outperformed the previously reported state-of-the-art approaches. To conclude the proposed IoT-based PAS random blood glucose monitoring system using kernel-based ridge regression is reported for the first time with more accuracy.

The chronic wounds have been a significant public health concern. The continuous monitoring using a smart point-of-care system can improve the chronic wound care quality. The proposed study implements the MobileNetV2 deep learning model for automated wound segmentation and estimates the area. The model has been implemented on a Jetson nano board with integrated camera module for point-of-care system. The model achieves 0.95 dice score for segmentation of images. The advancement could lead to more efficient and accurate chronic wound management, ultimately enhancing patient outcomes

This paper analyzes the concept of hybridization to leverage the advantages of multiple technologies. Specifically, we have designed a hybrid multiplexer using memristors and Carbon Nano Tube Field Effect Transistors (CNTFETs). The nonvolatility, high density, and scalability of memristors render them very promising for memory-based analog and digital applications. due to its reduced propagation delay, low power consumption, and multi-threshold functioning, CNTFETs are preferred in digital circuit design. Through its integration, this work aims to maximize the benefits of both technologies. With the concept, we have designed hybrid 2:1, 4:1, and 8:1 multiplexers (MUX) capable of operating in the sub-threshold region not seen in conventional CMOS. The proposed 8:1 MUX offered delay, power dissipation, and power-delay product as 12.005×10?8s,3.35?W, and 42.97 J respectively. To further reduce leakage current, we employed an advanced adaptation of the traditional power gating technique using CNTFET, with a power dissipation of 1.263?W. Comparative analysis with baseline CMOS technology indicates that the proposed hybrid approach significantly enhances the device specifications. All circuits were designed using 45 nm technology in Cadence Virtuoso

Accurate blood flow velocity measurement using Doppler ultrasound is critical for diagnosing cardiovascular diseases. The Doppler angle is a crucial step for accurately measuring blood velocity. The proposed study proposes a low-complexity deep learning framework integrated with Explainable AI (XAI) technique for Doppler angle estimation. For image segmentation, we evaluated several models, including VGG19, ResNet50, MobileNetV3, ResNet18, and EfficientNet-B0, assessing their Dice score accuracy and computational efficiency. The application of GradCAM XAI provides insights into model decision-making, crucial for enhancing diagnostic precision. Finally, a polynominial regression model facilitates Doppler angle estimation for assessments of blood flow dynamics. For best care scenario, we achieved a Dice score of 0.96 for segmented images and a mean error of -3.2 degrees for Doppler angle estimation. Overall, we propose a low-complexity model that enhances visual interpretation by segmenting arterial positions, uses for XAI for interpretability, automates Doppler angle estimation and is suitable for implementation on mobile devices

This paper introduces a software and hardware-integrated method for handwritten digit recognition using a customized Convolutional Neural Network (CNN architecture. The process comprises two concurrent tracks hardware implementation and software model development. The CNN was first designed and trained with Python frameworks before being transformed into a hardware-compatible format. It was then put on a Raspberry Pi (RPB) Board and converted into an ASIC design using Cadence tools. The conversion of floatingpoint to fixed-point representation of neural network parameters, such as weights and biases, allows for a smooth integration into a single hardware module. The RPB is used to verify the model's performance, and the Vivado platform is used for further verification, it uses comprehensive test cases to guarantee correctness and resilience. Moreover, a comparative study of the execution timings of hardware implementations (RPB) and software platforms (CPU, GPU) is carried out, emphasizing the design efficiency and interface performance. This method shows that it is feasible to implement neural networks with optimal performance on hardware with limited resources

This work presents the development of a custom convolutional neural network (CNN) architecture for image classification, efficiently implemented on a hardware platform such as the Raspberry Pi and ported to ASIC design using Cadence tools. The project adopts a two-stream approach: one dedicated to the Python-based modelling framework and the other focused on the ASIC design of the neural network. Key parameters such as including network weights and biases, are extracted from the Python model and converted to a fixed-point format for seamless integration into a unified hardware module with well-defined inputs and outputs. The CNN model was extensively validated on the Raspberry Pi hardware, while the hardware module was rigorously tested on Vivado using comprehensive test cases. The results were analyzed to ensure the correctness and robustness of the design. Additionally, the Python model's interface timing is evaluated on different software (CPU, GPU) and hardware (Raspberry Pi) platforms to assess the Python model performance

This extensive research delves into the intersection of MRI imaging and deep learning, in the task of identifying and categorizing brain tumors. In addition to models like VGG16 and ResNet101 a designed Convolutional Neural Network (CNN) was developed and thoroughly evaluated showcasing a range of techniques utilized. Augmentation methods were purposefully applied to enrich the dataset, enhancing the models' robustness and adaptability. Evaluation metrics, including the F1 score, recall, accuracy, and precision gave a general picture of the model's performance. Furthermore, leveraging Explainable AI (XAI) techniques such as LIME unveiled insights, into the decision-making processes underlying the models enhancing their interpretability and trustworthiness. The study findings ultimately underscore the potential of learning in revolutionizing automated brain tumor diagnosis and classification poised to enhance patient care pathways and medical diagnostic capabilities significantly. Index Terms-MRI image, deep learning, VGG16, ResNET101, CNN, LIME

The emerging market for hardware neuromorphic systems has fulfilled the growing demand for fast and energy-efficient computer architectures. Memristor-based neural networks are a viable approach to meet the need for low-power neuromorphic devices. Spiking neural networks (SNNs) are widely recognized as the best hardware solution for mimicking the brain's efficient processing capabilities. To build the SNN model, we have designed an energy-efficient hybrid Leaky Integrated and Fire (LIF) neuron model using Carbon Nano Tube Field Effect Transistors (CNTFET) and memristors. This hybrid neuron operates at 3.89 MHz, with 1.047nW and 0.257fJ of power and energy per spike with a constant power supply (Vdd) and an excitation voltage of 0.5V, under the ideal conditions. When the intrinsic constraints of CNTFETs and memristors, such as parasitic elements and hysteresis effects, are taken into consideration, the operating frequency is lowered to 3.45 MHz (an 11.5% decrease), and energy consumption rises to 0.317 fJ per spike (a 23.3% increase). Despite these limitations, our design outperforms with existing works. On the other hand the development of in situ, Spike Timing Dependent Plasticity (STDP) learning through memristors as synapses results in a computational challenge. In this paper, we adopt a potent technique capable of carrying out both learning and inference. The weight modulation is accomplished using a linear memristor model, resulting in high speed and reduced power consumption. We intend to apply the winner-takes-all (WTA) mechanism within the SNN architecture, which incorporates recurrently connected proposed neurons in the output layer, for real-time pattern recognition. The proposed design has been implemented and the performance metrics superseded the existing works in terms of power, energy, and accuracy. Furthermore, the design is capable of classifying 50×104 images per second

This paper presents a comprehensive overview of the potential applications for Photo-Acoustic (PA) imaging employing functional nanoparticles. The exploration begins with an introduction to nanotechnology and nanomaterials, highlighting the advancements in these fields and their crucial role in shaping the future. A detailed discussion of the various types of nanomaterials and their functional properties sets the stage for a thorough examination of the fundamentals of the PA effect. This includes a thorough chronological review of advancements, experimental methodologies, and the intricacies of the source and detection of PA signals. The utilization of amplitude and frequency modulation, design of PA cells, pressure sensor-based signal detection, and quantification methods are explored in-depth, along with additional mechanisms induced by PA signals. The paper then delves into the versatile applications of photoacoustic imaging facilitated by functional nanomaterials. It investigates the influence of nanomaterial shape, size variation, and the role of composition, alloys, and hybrid materials in harnessing the potential of PA imaging. The paper culminates with an insightful discussion on the future scope of this field, focusing specifically on the potential applications of photoacoustic (PA) effect in the domain of biomedical imaging and nanomedicine. Finally, by providing the comprehensive overview, the current work provides a valuable resource underscoring the transformative potential of PA imaging technique in biomedical research and clinical practice.

This paper presents the design of a synergistic arithmetic logic unit (SALU) that combines a complementary metal-oxide–semiconductor (CMOS) and memristors using the modified gate diffusion input (m_GDI) technique. The proposed design aims to reduce power consumption, delay, and the number of transistors, which are critical parameters in digital VLSI design. We have incorporated the m_GDI technique to implement the SALU, resulting in a lower number of transistors and reduced power consumption along with the delay and power delay product (PDP). The whole proposed architecture has been constructed using a basic m_GDI cell that consists of one p-type (pMOS) and two memristors. Combinational arithmetic circuits like full adder and full subtractor have been designed using the XOR-based m_GDI approach, and their performance will double that of current methods. A 1-to-8 de multiplexer circuit has also been utilized to perform 8 different operations for the ALU. Using Cadence Virtuoso tools in 45-nm technology, we have simulated the proposed design and evaluated its performance with existing designs. The proposed design has the potential to be a viable option for low-power and high-performance digital VLSI design applications. It delivers improved power, delay, and PDP characteristics while using fewer components. The power consumption, delay, and PDP of the proposed architecture are 9.71E?6 W, 2.141E?9 s, and 2.0789E?14 J, respectively. Furthermore, the design comprises just 33 pMOS transistors and 66 memristors, which is a lower count and unique combination in contrast to previous techniques.

The accurate diagnosis of medical conditions from low-light images, particularly black-and-white x-rays, is impeded by challenges such as noise, constrained visibility, and a lack of detail. Existing enhancement methods often exacerbate these issues by introducing detail loss, color oversaturation, or higher noise levels. This paper proposes a novel U-Net-based Convolutional Neural Network (CNN) specifically developed to address these challenges in low-light black-and-white medical images. Our designed architecture employs skip connections within the U-Net framework to effectively balance noise reduction with detail information preservation. This makes it possible for the network to learn hierarchical image representations while retaining important features for diagnosis. The trained network accomplishes real-time image enhancement, enabling immediate visual improvement during diagnosis and perhaps assisting radiologists in making faster and more accurate findings. Our approach illustrates a significant improvement in image quality and outperforms traditional methods in terms of noise reduction and detail preservation. This study holds significant potential to improve medical image analysis and diagnosis, potentially leading to enhanced patient care and earlier interventions.

The air pressure system (APS) is an integral component of Scania trucks and other heavy machinery. Because the brakes on these vehicles use air pressure, keeping the APS in good working order is crucial. Automakers can save money on repairs and boost vehicle efficiency with predictive maintenance. This can be done manually or using an automated system. Predictive maintenance that is performed manually requires human interaction and, as a result, introduces room for error. When humans are involved, there is always a chance that something may be missed or misunderstood, which might compromise the reliability of the maintenance procedures. Several benefits may be gained by employing automatic predictive maintenance strategies, such as artificial intelligence (AI), to investigate the underlying reasons for failure in the APS of Scania trucks. The company relies heavily on the dataset since it pinpoints the faulty parts. Predicting the root cause of failure is made more difficult if the dataset has missing values and unbalanced class issues. To overcome these issues, the data are preprocessed by many resampling techniques such as under-sampling, over-sampling, and the synthetic minority over-sampling technique (SMOTE), and imputation techniques such as KNNImputer and SimpleImputer for mean, mode, and constant strategies, multivariate imputation by chained equations (MICE), and principal component analysis (PCA), to balance the entire data set. After preprocessing, implementation of eight different machine learning algorithms, namely random forest, decision tree, gradient boosting, logistic regression, k-nearest neighbors classifier, AdaBoost classifier, CatBoost classifier, and XGB classifier, is carried out, and then the cost, accuracy metrics, and confusion matrices are analyzed. The results from the experimental analysis show that the XGB classifier is the best model, with accuracy of 99.6241% along with cost-effectiveness.

This paper presents an innovative enhancement aimed at revolutionizing disease detection and providing users with a reliable source of information for accurate diagnoses of their symptoms. Our open-source initiative combines a user-friendly interface design with advanced machine learning models, establishing a new benchmark for accuracy and enabling integration with even higher-performing models. We address the pervasive challenges of misinformation and misdiagnosis associated with online symptom searches, presenting a significant advancement in disease detection. Leveraging cutting-edge machine learning techniques, our system analyzes user-input symptoms against a comprehensive medical knowledge database, providing accurate and reliable information on potential diseases or conditions. Major challenges such as data quality, quantity, model interpretability, integration with healthcare systems, continual model improvement, and bias are tackled with the proposed methodology. This work includes the integration of higher-performing models, open-source principles fostering collaboration, and continuous improvement of diagnostic accuracy. Additionally, efforts to enhance model interpretability through visualization and explanation methods are proposed. Overall, our work represents a significant step towards a more reliable and accurate healthcare technology, with potential implications for the broader field of medical diagnostics.

The use of neural modeling tools is becoming increasingly common in the exploration of human brain behavior, enabling effective simulations through event-driven methodologies. As a result, years of study and advancements in the field of neurotechnology have led to the creation of several artificial neural network approaches that mimic biological neural networks. The event-driven approach provides an effective method for mimicking large-scale spiking neural networks (SNNs), by taking advantage of the brain’s sparse processing. This paper investigates SNN employing a leaky integrate-and-fire neuron model with latency estimation through FNS. A three-layer feedforward network (FFN) is constructed, incorporating design parameters from Config Wizard. Notably, our study sheds light on the impact of synchrony within a simple FFN. Through the incorporation of biologically plausible delay effects, our model offers novel insights that complement the existing literature. Neural activity is organized in CSV format files, facilitating the reconstruction of electrophysiological-like signals. FNS enables a comprehensive exploration of interactions within and between populations of spiking neurons. In the near future, we intend to use these findings in situations where this particular class of neural networks and digital signal processing (DSP) applications can be combined to create potent nonlinear DSP techniques.

The rapid development of Internet of Things (IoT) technology is driving a transformation in the healthcare sector. This paradigm change provides new opportunities for real-time, ongoing physical parameter monitoring, particularly in remote situations, providing an ideal setting for research and development. IoT device deployment has become widespread, enabling the growth of an automated data exchange ecosystem. However, our capacity to carry out remote monitoring has been constrained by our past dependence on specialized electronic equipment for assessing vital signs such as heart rate (beats per minute [BPM]) and oxygen saturation (SpO2). To address this issue, we developed an innovative technology that makes use of internet connectivity to allow for remote vital sign measurement and monitoring. The main focus of this article is the use of IoT technology to measure and track vital physiological indicators, notably heart rate and oxygen saturation, regardless of a person’s location. In addition, our study aims to create a system that can send out real-time notifications in the event of serious medical emergencies, increasing the likelihood that life-saving actions can be taken in a timely manner. © The Minerals, Metals & Materials Society 2024.

Quantum-dot Cellular Automata (QCA) technology is emerging nanotechnology for designing low-power digital circuits and various high-performance calculations at the nanoscale dimension, as it is termed as an emerging technology in Digital Image Processing (DIP) due to having advantages like less area occupancy, low energy dissipation, and high speed as compared with conventional transistor-based technologies. This paper demonstrates the design & implementation of median filter (MF) using QCA technology. The MF plays an important role in DIP for the reduction in noise. The proposed QCA-based MF is designed in a single layer with less cell count and low latency. The MF is designed by using Compare and Selective Module (CSM). The proposed 1-bit, 2-bit & 4-bit CSM architectures occupy the area of 0.17,0.52&3.25?m2 and use 118, 380 & 1963 QCA cells, respectively. The proposed CSM is further extended to a larger bit size. The QCA Designer-E simulation tool has been used to design, and verify all the proposed architectures. The energy dissipation has been simulated using a coherent vector engine setup. The total energy dissipation of 1-bit, 2-bit & 4-bit CSM architecture is 2.56×10-2,1.35×10-1&5.19×10-1eV, and the average energy dissipation is 2.31×10-3,1.22×10-2&4.71×10- 2eV, respectively. The total & average energy dissipation per cycle of the proposed MF is 41.72×10-1&38.26×10-2eV, respectively.

A hybrid nanogenerator (HNG) offers both high output performance and flexibility by utilizing the synergy between piezoelectric and triboelectric mechanisms. Achieving high output performance, reproducibility, and mechanical stability in a HNG device is still a major challenge. Here, we demonstrate the design and fabrication of a flexible HNG device based on the composite of a lead-free piezoelectric ceramic and the triboelectric polymer poly(dimethylsiloxane) (PDMS). The piezoelectric ceramic oxide (BiKBa)(FeTi)O (BKBFT/MPB-piezo) exhibits improved piezoelectric properties at the morphotropic phase boundary (MPB) in the BiFeO-BiKTiO-BaTiO ternary phase diagram. We find that the composite 90 wt % PDMS-10 wt % MPB-piezo offers optimum device performance and flexibility. Interestingly, the incorporation of multiwalled carbon nanotubes (MWCNTs), a conducting filler, significantly enhances the device’s performance without the aid of electric field poling. MWCNTs form nanoelectrical bridges that aid in charge transfer and improve the composite’s structural homogeneity. The 89 wt % PDMS-10 wt % MPB-piezo-1 wt % MWCNT composite displays a peak-to-peak open-circuit voltage (V), short-circuit current (I), and power density (U) of 22 V, 1.8 ?A, and 72 nW/cm, respectively. Furthermore, we show the capability of the composite to be used as a wearable human pulse sensor.

A new Decision Based Adaptive Denoising Filter (DBADF) algorithm & hardware architecture are proposed for restoring the digital image that is highly corrupted with impulse noise. The proposed DBADF detects only the corrupted pixels and that pixel is restored by the noise-free median value or previous value based upon the noise density in the image. The proposed DBADF uses a 3 × 3 window initially and adaptively goes up to 7 × 7 window based on the noise corruption more than 50% by impulse noise in the current processing window. The proposed architecture was found to exhibit better visual qualitative and quantitative evaluation based on PSNR, IEF, EKI, SSIM, FOM, and error rate. The DBAMF architecture also preserves the original information of digital image with a high density of salt & pepper noise, when compared to many standard conventional algorithms. The proposed architecture has been simulated using the VIRTEX7 FPGA device and the reported maximum post place and route frequency are 149.995MHz and the dynamic power consumption is 179mW.

Ultrasound images often get distorted by impulse noise during data acquisition and processing in the Back-end of the system, which overlay the finer details of the scanned body parts. Generally, a portable low-cost USG system doesn't have an impulse noise-cleaning module which hinders detections of smaller details in the images. A Depth Invariant Impulse Noise Removal (DIINoR) algorithm for real-time impulse noise removal from the corrupt USG image is proposed in this paper. In this decision-based algorithm, the corrupt pixel is first detected depending on the homogeneity of the processing window and is restored with the median of the window or previous pixel value. Testing of the DIINoR algorithm on different USG images establishes that the denoised images have superior quantitative performance compared to those of existing schemes which proves its suitability for the real-time fixed and random valued impulse noise cleaning in the Back-end of the portable USG system.

The wearable gadgets facilitate the continuous real-time monitoring of personal health. The VLSI industry is attempting to incorporate more functional modules that run at high speed and consume less power inside the prescribed space. Several methodologies and procedures are designed to implement practical VLSI circuits to meet the market requirements. The Comparator is a basic arithmetic unit in high-end Processors for IoT-based applications. The concept of hybridization facilitates a promising solution to realize high-speed, low-power systems with a minimum number of transistors as compared to the CMOS technology. On the other hand, m-GDI-based logic circuit design is very popular for high-speed & low-power applications. These strategies demonstrate the trade-off between several parameters. In this paper, we have designed a novel CMOS-Memristors-based hybrid 16-bit magnitude compactor using the modified Gate Diffusion Input (m-GDI) technique. The proposed comparator operates at Vdd=1V and offers 12.71nw, 2.44ps, and 31.02zJ of power dissipation, delay, and PDP respectively. All the prescribed circuits have been designed and simulated using a 45nm Generic Process Design Kit (GPDK) in the Cadence Virtuoso tool.

A Real-Time Impulse Noise Removal (RTINR) algorithm and its hardware architecture are proposed for denoising images corrupted with fixed valued impulse noise. A decision-based algorithm is modified in the proposed RTINR algorithm where the corrupted pixel is first detected is restored with median or previous pixel value depending on the number of corrupted pixels in the image. The proposed RTINR architecture has been designed to reduce the hardware complexity as it requires 21 comparators, 4 adders, and 2 line buffers which in turn improve the execution time. The proposed architecture results better in qualitative and quantitative performance in comparison to different denoising schemes while evaluated based on PSNR, IEF, MSE, EKI, SSIM, FOM. The proposed architecture has been simulated using the VIRTEX7 FPGA device and the reported maximum post place route frequency is 360.88 MHz. The proposed RTINR architecture is capable of denoising images of size 512 × 512 at a frame rate of 686. The architecture has also been synthesized using UMC 90 nm technology where 103 mW power is consumed at a clock frequency of 100 MHz with a gate count of 2.3K (NAND2) including two memory buffers.

Quantum-dot Cellular Automata (QCA) lead to fundamental changes in nanoscale technology. It promises small area, low power, & high-speed structures for digital circuit design. This paper presents efficient low power structures of reversible multiplexer & demultiplexer (RMD) modules based on the QCA technology. The simulation result shows that the proposed RMD module has utilized less area & low power consumption. The simulation, layout, & energy dissipation of the proposed RMD module have been carried out using the QCA Designer-E simulation tool.

-

-

With the rapid development of Very Large-Scale Integration (VLSI) technology, it is important to achieve a robust design with low power consumption. CMOS design has been affected by several problems over the past few years. Increasing the dissipation of power is a major problem in CMOS devices and circuits. Reversible computing can solve this issue, and reversible logic circuits serve as the foundation of quantum computing. Quantum-dot Cellular Automata (QCA) can be such a nanoscale technology and thus emerges as a promising alternative to the traditional CMOS VLSI. This work focuses on the design of a reversible multiplexer and demultiplexer in the quantum dot cell automata (QCA) framework. Experimentation reveals that the new reversible mux and demux is superior to the traditional reversible modules. The simulation, layout energy dissipation of the proposed RMD, RM module has been carried out using the QCA Designer-E simulation tool.

This paper proposes a high-throughput median finding architecture where the sorting of an incoming pixel is executed by a high-speed Compare and Select (CS) module. In this work, four clock pulses are required to populate the 4 × 4 window as four pixels are read at a time from the incoming grey image.This median finding process is carried out by parallel and pipeline median architecture. The proposed median finding process requires two read operations to take eight input pixels and generates four output pixels with a latency of seven clock cycles. The proposed architecture has been implemented on Xilinx Virtex–VII FPGA. The proposed architecture is synthesized using the SoC Encounter along with Faraday 90 nm standard cell library. The maximum operating frequency is 950.57 MHz, the total gate count is 4540,area is 0.40543mm2 and the dissipated power is 0.92617 mW. The high-throughput, high-speed and low-power-dissipation nature of the proposed architecture make it suitable for computationally extensive Internet of Things (IoT) applications.

Existing architectures for the median filter are based on sorting algorithm where comparators are used in serial. This paper proposes a new high-speed architecture of two dimensional (2-D) median filter where compare and select modules are used in parallel to sort the incoming numbers. The hardware implementation results show that the proposed architecture (PA) operates at 26% and 34% higher frequency in Virtex 4 and Virtex 7 FPGA device, respectively, in comparison with the architectures reported. The PA is synthesized using the RTL Compiler of Cadence along with Faraday 180 nm standard cell library. The maximum operating frequency of the PA is 1.06 GHz with a total gate count of 917. The complete chip layout has been done using the SoC encounter tool. The area of the final chip is 0.13928 mm with a power consumption of 0.168 mW analysed using prime-power.