Many important applications rely on Dynamic Facial Expression Recognition (DFER), including affective computing, mental health monitoring, and human–computer interaction. The computational cost of current state-of-the-art approaches is increased and important contextual clues are lost due to their reliance on Face Detection and Alignment (FDA) preprocessing. This paper proposes a novel FDA-free hybrid DFER framework TSM-Transformer that integrates EfficientNetV2-Lite for lightweight spatial feature extraction, Temporal Shift Modules (TSM) for parameter-free local motion encoding, and a Transformer-based temporal fusion mechanism for global sequence modeling. By processing full-frame video inputs, the proposed model preserves both facial and body cues, enhancing robustness under real-world conditions with variations in lighting, occlusions, head poses, and background complexity. Experimental evaluation on a multi-class emotion dataset demonstrates that the TSM-Transformer achieves state-of-the-art performance, with 91.38

With the rising demand for high-speed devices operating in ultra-wideband spectrum and 5G Networks, there is a need for low-power battery-operated devices. LNA, being the first block of the receiver chain, must be optimized. This research paper outlines that implementing the particle swarm optimization (PSO) algorithm improves LNA design. The recommended LNA design consists of two cascaded phases, and the initial stage of the circuit employs a current-reuse topology. Two transistors are cascaded in the subsequent phase to enhance the entire gain of the circuit. Particle Swarm Optimization (PSO) balances the aspect ratio, all passive variables with voltage gain and NF, and basic parameters of interest at the target frequency. The proposed LNA circuit is simulated in a standard 45 nm CMOS process for a bandwidth of 1 GHz–20 GHz. Forward gain (S21) of 16.8 dB over the entire band, input return loss (S11) of -19.2 dB, and (NF)min of 1.7 dB were observed with a power consumption of 3.6mW. A process corner simulation analyzed the adaptability of the LNA, and the findings showed an approximate ten per cent deviation from the theoretical value. The chip footprint is about 3.52 mm2, with a core area of 645 µm by layout.

Moving object detection and segmentation in the field of computer vision is gaining much interest amongst researchers for its wide variety of application. Moving object detection aims at captivating the moving objects in a video sequence whereas moving object segmentation focuses on segmenting the objects in motion from a stationary background in a video sequence. This paper introduces both these computer vision tasks in detail along with its various applications and challenges. A comprehensive literature review in this field is provided in the next section. Further, the various deep learning approaches for moving object detection and segmentation are detailed with their advantages and disadvantages. The various available dataset for surveillance applications along with their properties are elaborated. Next section explains the numerous evaluation metrics that are used to evaluate the performance of the model. The qualitative results are detailed along with graphs and figures. At last, the study is concluded with future research directions in this field.

This paper presents a high-gain, single-stage cascode low noise amplifier (LNA) tailored for S-band applications, implemented in a 180-nm CMOS process. The design employs a common-source cascode topology with inductive degeneration, enabling a careful trade-off between gain, noise figure, and input matching. Schematic simulations indicate a peak gain of 53.2 dB and a low noise figure of 1.51-1.74 dB across the S-band, with excellent impedance matching. The amplifier is unconditionally stable and occupies a compact core area of only 0.175 mm(2). Operating with a power consumption of 45 mW, these results demonstrate that the proposed LNA provides a competitive, area-efficient, and robust solution for modern S-band wireless systems, achieving performance levels comparable to more complex multi-stage architectures while maintaining simplicity and design efficiency. The design also achieves excellent reverse isolation (S12 < -48 dB), output matching (S22 = -12.75 dB), and unconditional stability (K > 1), further demonstrating the robustness of the proposed LNA.

Electrical utilities attach significant emphasis to power load forecasting to design the generating, transmission, and distribution systems. Additional uses of load forecasting in the power network include demand and supply management, determination of power plant operational resource needs, planning of spinning reserve, and scheduling power unit activities. Additionally, there is the option to use a plethora of other applications. The researcher used two distinct machine learning approaches, including Recurrent Neural Network (RNN) and Long Short-Term Memory (LSTM), also known to forecast the expected future demand for electrical power in this study. Historical load information’s are used to assess the efficiency of implemented measures. The Root Mean Squared Error (RMSE) and Mean Absolute Percentage Error (MAPE) are analytical metrics used to assess the accuracy of the methods being used.

Recommendation systems are ubiquitous in various domains, facilitating users in finding relevant items according to their preferences. Identifying pertinent items that meet their preferences enables users to target the right items. To predict ratings for more accurate forecasts, recommender systems often use collaborative filtering (CF) approaches to sparse user-rated item matrices. Due to a lack of knowledge regarding newly formed entities, the data sparsity of the user-rated item matrix has an enormous effect on collaborative filtering algorithms, which frequently face lazy learning issues. Real-world datasets with exponentially increasing users and reviews make this situation worse. Matrix factorization (MF) stands out as a key strategy in recommender systems, especially for CF tasks. This paper presents a neural network matrix factorization (NNMF) model through machine learning to overcome data sparsity challenges. This approach aims to enhance recommendation quality while mitigating the impact of data sparsity, a common issue in CF algorithms. A thorough comparative analysis was conducted on the well-known MovieLens dataset, spanning from 1.6 to 9.6 M records. The outcomes consistently favored the NNMF algorithm, showcasing superior performance compared to the state-of-the-art method in this domain in terms of precision, recall, ℱ1_score , MAE, and RMSE.

Bluetooth, Wi-Fi are famous data transfer devices for 5G and beyond networks for better connectivity LNA play a vital role as first block of the receiver chain. This paper proposed a CMOS low noise amplifier (CLNA) novel architecture for wideband matching in RF environments and IoT operated devices to achieve high gain (S21) and low noise figure (NF), an improved cascode source degenerated topology with an extra transistor is utilized to provide biased voltage to conventional CS transistor at 2.4 GHz center frequency with (1–5 GHz) BW is adopted due to diverse field of operation. An input output matching circuit is designed to achieve better input and output reflection coefficient (S11 and S22). The Proposed circuit utilizes CMOS 0.18 μm node to achieve S21 as 22.56 dB, S11 of − 19.65 dB, S12 of − 49.23 dB, NF of 1.1 dB, IP1dB of − 18.12 dB, IIP3 of − 9.38dBm at the desired frequency is achieved at 1.8 V.

The designs of first- and second-order digital low-pass filters with infinite impulse response (IIR) are presented in this letter, utilizing a meta-heuristic optimization technique. Firstly, the analog transfer functions of the first and second- order filters are considered, followed by the application of an L-1-norm-based multi-verse optimization algorithm to directly emulate their magnitude-frequency response in the digital domain. The obtained magnitude-frequency response shows superior matching with the analog counterpart for different cut-off frequencies of the first- and second-order filters, as well as varying quality factors for the second-order filter. In comparison to the filter's magnitude-frequency response obtained through traditional bilinear transform and advanced operators, the proposed technique accurately manifests the analog magnitude-frequency response in the digital domain.

Interleave division multiple access (IDMA), which is based on interleaving sequences, may be regarded as code domain non orthogonal multiple access (NOMA) and may confront numerous challenges in supporting big diverse data traffic and a high number of users. Further, full duplex wireless communication can support double of the data rate in comparison to half duplex network. However, the throughput can be limited due to self-interference (SI). To overcome such interference and other environmental hazards, a newly proposed technique i.e., intelligent reflecting surfaces (IRS) in which several reflecting elements are mounted on the single surface can be utilized to design the wireless environment, resulting in improved performance through constructive reflections. In this article, the amalgamated system i.e. intelligent reflecting surface assisted full duplex IDMA communication system has been proposed to incorporate all the advantages of NOMA, IDMA and IRS. To validate the performance of the proposed system, the outage probability in full duplex mode is examined. All the simulations have been carried out in MALAB to calculate the outage and error probability.

This paper reports finite element model (FEM) simulation and fabrication of a square shaped diaphragm along with microtunnel for MEMS acoustic sensor which can be used for measurement of wide operational frequency range and high sound pressure level (SPL) 100 dB-180 dB measurement in launching vehicle and aircraft. The structure consists of a piezoelectric ZnO layer sandwiched between two aluminum electrodes on a thin silicon diaphragm. There is a microtunnel in the structure which relates the cavity to the atmosphere for pressure compensation. The microtunnel decides the lower cut-off frequency of device. Analytical and simulation approaches are used to optimize microtunnel dimension and simulation approach for diaphragm structure optimization. The change in displacement, stress, sensitivity and resonance frequency due to different diaphragm sizes with diaphragm thickness variation is also analyzed. The optimized diaphragm structure of 1750 x 1750 & mu;m2 and microtunnel of 100 & mu;m wide and 24 & mu;m deep have been fabricated using bulk micromachining technique. The fabricated device response has been tested using LDV and sensitivity measurement system.

The human body is regulated by a variety of glands that produce hormones, which are responsible for making us feel good or bad. Cortisol is a hormone that falls under the category of feel-bad hormones as it increases stress levels in the body. Stress can come in different forms such as acute, episodic acute, and chronic and can have adverse effects on a person’s physical, mental, and emotional health. It can lead to the development of diseases such as diabetes, heart ailments, depression, asthma, obesity, Alzheimer’s disease, gastrointestinal problems, and anxiety, among others. Early diagnosis and prevention of such diseases can be aided by monitoring and predicting stress levels. Various techniques are reported in the literature to measure stress, including wearable devices, behavioural coding, self-reporting, physiological measuring tools, heart rate variability analysis, psychosocial approach, perceived stress scale, and measuring salivary and hair cortisol. This chapter focuses on digital monitoring techniques for measuring stress levels, such as using intelligent wireless sensor systems, personal digital assistants, mobile applications, bioelectronics, digital signal processing, and other such technologies. The role of recent computational techniques such as machine learning, deep learning, and the Internet of Things in real-time stress detection has also been discussed, highlighting potential directions for further research in this area.

The factors affecting the stock market are large in numbers that make accurate predictions a challenging task. There is an overwhelming addition of data on the Internet, and some of this data like current market sentiments along with technical indicators can help in better stock prediction. In this paper, the state-of-the-art machine learning models ARIMA, SVR, LSTM, and XGBoost along with the ensembles of these models using weighted averaging and boosting techniques have been studied and compared on a 1-year and 5-year timeline. The data have been collected for two companies, HDFC and Sun Pharma, listed in the Nifty 50 stocks on the National Stock Exchange (NSE) in India.. The study shows the ensemble usefulness for different models, technical indicators, and sentimental analysis (Wikipedia hits and Google News mentions) in the stock prediction.

As the use of multimedia devices is rising, power management is becoming a major challenge. Various types of compressors have been designed in this study. Compressor circuits are designed using several circuits of XOR-XNOR gates and multiplexers. XOR-XNOR gate combinations and multiplexer circuits have been used to construct the suggested compressor design. The performance of the proposed compressor circuits using these low-power XOR-XNOR gates and multiplexer blocks has been found to be economical in terms of space and power. This study proposes low-power and high-speed 3-2, 4-2, and 5-2 compressors for digital signal processing applications. A new compressor has also been proposed that is faster and uses less energy than the traditional compressor. The full adder circuit, constructed using various combinations of XOR-XNOR gates, has been used to develop the proposed compressor. The proposed 3-2 compressor shows average power dissipation 571.7 nW and average delay 2.41 nS, 4-2 compressor shows average power dissipation 1235 nW and average delay 2.7 nS while 5-2 compressor shows average power dissipation 2973.50 nW and average delay 3.75 nS.

Non-cooperative scenarios in cognitive wireless sensor network (CWSN) are often encountered with shadowing and hidden terminal issues. Cooperative spectrum sensing (CSS) can solve these issues but at the expense of large overhead. It is necessary to optimise the energy of battery-operated unlicensed users, also known as secondary users (SUs). The effect of fading and noise uncertainty is often overlooked when determining CSS performance. Energy efficiency (EE) is a comprehensive parameter that gives a complete picture of the overall performance of the CSS. There are several parameters that affect the EE of CSS. In this paper, the detector threshold is optimised to maximise the EE of the system. A system model is proposed to determine the EE of centralised CSS over different fading channels in noisy reporting conditions. An iterative algorithm is presented which determines the optimum detector threshold for which the EE is maximum. Results show that the optimum value of the detector threshold from the analytical model matches with simulation data.

Low power demand of electronics industry and communication systems are driven by technology scaling and marketing. Compressors are the most often used modules in the architecture of an Arithmetic Circuits. Compressor is the basic combinational digital logic circuit for performing numerous arithmetic operations. The design criterion in any Compressor circuit involves the XOR-XNOR and multiplexer circuits. It is a prominent component in the designing of integrated circuit that performs arithmetic operations. The proposed Compressor is design in order to reduce the transistor count which results in curtailing power dissipation and delay. The performance parameter are recorded and tabulated. This compressor has less delay and low power dissipation. The temperature effect has also shown a linear change in power dissipation in lower temperature ranges which shows the robustness of the proposed circuits. The proposed compressor can be used in numerous applications such as multiplier, DSP microprocessor and data processing systems.

This paper exhibits review on different low-noise amplifier (LNA) topologies for wideband frequency applications. The LNA design metric includes gain, noise figure (NF), power dissipation, bandwidth, and linearity with broadband wideband input impedance matching. These all specified parameters required a tradeoff for optimization of LNA. This paper includes all-important LNA topologies and provides good insight to optimize LNA performance. In this paper, FOM is used to compare all the specifications on single platform. In addition, a performance summary of result has been given which will help the creation of new idea. Moreover, researchers also utilize the capabilities of soft computing like particle swarm optimization (PSO) and firefly algorithms (FA) in the area of LNA optimization and added at the end of Sect. 2.

Tracking accurate movement of stock market is a difficult job for investors in recent times due to the non- linear behavior and volatility of the stocks, which gets affected because of different reasons. Now, advancement in technology, introduction of artificial intelligence, improved computation power and programming methods have enabled researchers to forecast the market more efficiently. This paper is an attempt to create new variables with the help of artificial neural network (ANN) and random forests (RF) algorithm, to be used to forecast closing price of the next day. For this, open, high, closing and low price are used to create new variables, which can be treated as inputs to the model. The indicators used in this model are RMSE, MBE and MAPE and their low values are indicative of the fact that the proposed machine learning methods are suitable for the prediction of stock prices of the companies under consideration.

Stress is a mental illness that impacts facets of life and can cause crucial health problems like sleep disorders and depression. In order to stay informed about one’s mental health, it is important to analyze one’s vitals. Wearable IoT devices collect and send the physiological parameters to an edge device for further processing and monitoring stress levels. Additionally, person-specific stress monitoring systems outperform generic ones, but they have limitations because person-specific models are not very adaptive to a wide range of people. Furthermore, creating a generic stress model is difficult because of different stress handling capacities. In this paper, we have proposed an IoT and Machine learning-based stress monitoring system. The proposed approach is a hybrid model that gives a relatively accurate prediction. We have demonstrated the proposed model on the WSEAD dataset and comparative analysis has been done with state-of-the-art methods.