Lightweight Hybrid Research Articles

Studies on Al-Si based hybrid aluminium metal matrix nanocomposites

A lightweight Hybrid Aluminium Metal Matrix Nanocomposite (HAMNC) is proposed for brake rotor application in this work. Hypereutectic Aluminium-Silicon (Al-Si) grade LM30 alloy is opted as the matrix with Boron Carbide (nB4C) and Rutile (nTiO2) nanoparticles as reinforcements. The brake rotor samples are produced using ultrasonic squeeze-assisted stir-casting, which is cost-effective and feasible for mass production. The hybrid nanocomposite specimens are fabricated by maintaining 1 wt% of nB4C and by varying TiO2 nanoparticles from 0 wt% to 1.00 wt% in steps of 0.25 wt%. The formulations are subjected to microstructure study and dry sliding wear test. The Field Emission Scanning Electron Micrographs (FESEM) reveal that the stir-casting method combining ultrasonic and squeeze effects aids in the uniform distribution of nano reinforcements. The tribological findings show that the hybrid nanocomposite containing 0.75 wt%. of nTiO2 has an 82.76% lower specific wear rate compared to the parent LM30 alloy. Further, the HAMNC brake rotor is 60% lighter than the brake rotors made of Grey Cast Iron making it a potential replacement material.

A Lightweight Hybrid Model with Location-Preserving ViT for Efficient Food Recognition.

Food-image recognition plays a pivotal role in intelligent nutrition management, and lightweight recognition methods based on deep learning are crucial for enabling mobile deployment. This capability empowers individuals to effectively manage their daily diet and nutrition using devices such as smartphones. In this study, we propose an Efficient Hybrid Food Recognition Net (EHFR-Net), a novel neural network that integrates Convolutional Neural Networks (CNN) and Vision Transformer (ViT). We find that in the context of food-image recognition tasks, while ViT demonstrates superiority in extracting global information, its approach of disregarding the initial spatial information hampers its efficacy. Therefore, we designed a ViT method termed Location-Preserving Vision Transformer (LP-ViT), which retains positional information during the global information extraction process. To ensure the lightweight nature of the model, we employ an inverted residual block on the CNN side to extract local features. Global and local features are seamlessly integrated by directly summing and concatenating the outputs from the convolutional and ViT structures, resulting in the creation of a unified Hybrid Block (HBlock) in a coherent manner. Moreover, we optimize the hierarchical layout of EHFR-Net to accommodate the unique characteristics of HBlock, effectively reducing the model size. Our extensive experiments on three well-known food image-recognition datasets demonstrate the superiority of our approach. For instance, on the ETHZ Food-101 dataset, our method achieves an outstanding recognition accuracy of 90.7%, which is 3.5% higher than the state-of-the-art ViT-based lightweight network MobileViTv2 (87.2%), which has an equivalent number of parameters and calculations.

Traffic prediction for 5G: A deep learning approach based on lightweight hybrid attention networks

The maturity of 5G technology provides a guarantee for increasingly large communication networks, while the resources required for communication and computation are also increasing, and reasonable resource allocation can improve the efficiency of network communication and reduce the consumption of communication resources. Existing deep learning methods have been able to predict network traffic to a certain extent, so as to solve the communication efficiency and resource consumption problems in the field of integrated sensing, communication and computation (ISCC) through rational resource allocation. However, the following problems still exist: (1) The feature learning ability of the prediction model is insufficient, and the prediction accuracy needs to be improved. (2) Powerful and complex deep learning methods lead to an increase in the prediction cost of the model. To address these problems, this paper proposes a deep learning method based on a lightweight hybrid attention network. In order to capture the key features of 5G data more effectively, an efficient hybrid attention mechanism (EHA) is proposed. After this attention is applied to convolution, the key information can be well enhanced. We use depthwise separable convolution in feature extraction, which greatly improves the efficiency of lightweight convolution layer (LC) in feature extraction. Combined with the efficient hybrid attention mechanism (EHA), the proposed model has better lightweight properties. Experimental results show that the model proposed in this paper has lower RMSE and MAE values on the three datasets, as well as fewer parameters and computational effort compared to the baseline scheme.

Low-latency DAE: A Configurable Lightweight Hybrid Data and Address Encryption Engine for IoT Real-time NVM Protection

Low-latency DAE: A Configurable Lightweight Hybrid Data and Address Encryption Engine for IoT Real-time NVM Protection

LH-IDS: Lightweight Hybrid Intrusion Detection System Based on Differential Privacy in VANETs

LH-IDS: Lightweight Hybrid Intrusion Detection System Based on Differential Privacy in VANETs

Image Super‐Resolution Reconstruction Based on the Lightweight Hybrid Attention Network

Image Super‐Resolution Reconstruction Based on the Lightweight Hybrid Attention Network

An Effective and Lightweight Hybrid Network for Object Detection in Remote Sensing Images

An Effective and Lightweight Hybrid Network for Object Detection in Remote Sensing Images

A Collision Proof Energy Efficient Lightweight Hybrid Cryptosystem for Fog

A Collision Proof Energy Efficient Lightweight Hybrid Cryptosystem for Fog

A hybrid TCN-GRU model for classifying human activities using smartphone inertial signals.

Recognising human activities using smart devices has led to countless inventions in various domains like healthcare, security, sports, etc. Sensor-based human activity recognition (HAR), especially smartphone-based HAR, has become popular among the research community due to lightweight computation and user privacy protection. Deep learning models are the most preferred solutions in developing smartphone-based HAR as they can automatically capture salient and distinctive features from input signals and classify them into respective activity classes. However, in most cases, the architecture of these models needs to be deep and complex for better classification performance. Furthermore, training these models requires extensive computational resources. Hence, this research proposes a hybrid lightweight model that integrates an enhanced Temporal Convolutional Network (TCN) with Gated Recurrent Unit (GRU) layers for salient spatiotemporal feature extraction without tedious manual feature extraction. Essentially, dilations are incorporated into each convolutional kernel in the TCN-GRU model to extend the kernel's field of view without imposing additional model parameters. Moreover, fewer short filters are applied for each convolutional layer to alleviate excess parameters. Despite reducing computational cost, the proposed model utilises dilations, residual connections, and GRU layers for longer-term time dependency modelling by retaining longer implicit features of the input inertial sequences throughout training to provide sufficient information for future prediction. The performance of the TCN-GRU model is verified on two benchmark smartphone-based HAR databases, i.e., UCI HAR and UniMiB SHAR. The model attains promising accuracy in recognising human activities with 97.25% on UCI HAR and 93.51% on UniMiB SHAR. Since the current study exclusively works on the inertial signals captured by smartphones, future studies will explore the generalisation of the proposed TCN-GRU across diverse datasets, including various sensor types, to ensure its adaptability across different applications.

Pioneering Security: A Hybrid Logic Framework for the Future of IoT Protection

In a complicated situation, the Internet of Things simplifies the process of connecting a wide range of things. Because of its openness and lack of human control, the Internet of Things is open to assaults like denial of service (DoS) and man-in-the-middle attacks. Furthermore, any device that can connect to the internet may be hacked. These attacks put the network connections and the actual equipment at danger. IoT security and privacy will so be compromised. Due to its limited power, bandwidth, and storage, the Internet of Things requires a security solution that does not overload it. This study aims to preserve consumers' trust in the Internet of Things (IoT) by safeguarding their data against unauthorised access and maintaining the privacy of their personal information. With the ultimate objective of presenting a unique hybrid and optimised lightweight logical security architecture to ensure data privacy and integrity while making effective use of available resources, all research are carried out with this purpose in mind. The Hybrid Lightweight Security Framework (HLSF), which this study suggests, offers secrecy and integrity in addition to authentication. The structure consists of three distinct steps. The first step is registration, the second is authentication, and the third is transit security, which protects data while it is being transported. Compared to other current frameworks, the results reveal that HLSF performs better in terms of precision, recall, and accuracy when applied to an IoT situation.

Securing wireless sensor networks with novel hybrid lightweight cryptographic protocols

WSN are essential in the Internet of Things (IoT) framework as they enable the gathering and transmission of data in different fields. The limited resources of sensor nodes in these networks require security protocols that are lightweight, striking a balance between data protection and efficiency. This research paper introduces a new hybrid cryptographic protocol called AES-128-CCM + Diffie-Hellman (ACDH), which aims to tackle the security issues encountered by WSN. Here present thorough assessment of the ACDH protocol, with a specific emphasis on critical factors including Security Strength, Energy Efficiency and Communication Overhead. The comparative study examines the ACDH protocol in relation to three well-established lightweight cryptographic protocols like Lightweight IPSec, BLAKE2, and Simon. We analyze and highlight the differences between these protocols. These protocols function as standards for evaluating the efficiency of ACDH in a practical WSN setting. The evaluation unequivocally showcases the supremacy of the proposed ACDH model across various crucial aspects. ACDH demonstrates exceptional Security Strength, guaranteeing strong resilience against established cryptographic attacks and providing a high degree of data safeguarding. It demonstrates exceptional Energy Efficiency, enabling sensor nodes to preserve energy resources while upholding a high standard of security ACDH reduces Resource Utilization, optimizing the utilization of memory and processing power. The protocol’s Communication Overhead is significantly minimal, guaranteeing the most efficient utilization of restricted bandwidth. The ACDH hybrid model consistently surpasses its counterparts in performance, making it the optimal selection for securing WSNs while adhering to the resource limitations of sensor nodes. The results of our research highlight the capability of ACDH as an innovative solution for developing lightweight cryptographic protocols that can improve the security and efficiency of Wireless Sensor Networks in the growing IoT environment.

MH-LViT: Multi-path Hybrid Lightweight ViT Models with Enhancement Training

MH-LViT: Multi-path Hybrid Lightweight ViT Models with Enhancement Training

Lightweight Hybrid CAE-ELM and Enhanced Smote Based Intrusion Detection for Networks

Lightweight Hybrid CAE-ELM and Enhanced Smote Based Intrusion Detection for Networks

A lightweight hybrid CNN-LSTM explainable model for ECG-based arrhythmia detection

Objective:Electrocardiogram (ECG) is the most frequent and routine diagnostic tool used for monitoring heart electrical signals and evaluating its functionality. The human heart can suffer from a variety of diseases, including cardiac arrhythmias. Arrhythmia is an irregular heart rhythm that in severe cases can lead to stroke and can be diagnosed via ECG recordings. Since early detection of cardiac arrhythmias is of great importance, computerized and automated classification and identification of these abnormal heart signals have received much attention for the past decades. Methods:This paper introduces a light Deep Learning (DL) approach for high accuracy detection of 8 different cardiac arrhythmias and normal rhythms. To employ DL techniques, the ECG signals were preprocessed using resampling and baseline wander removal techniques. The classification was performed using an 11-layer network employing a combination of Convolutional Neural Network (CNN) and Long Short Term Memory (LSTM). Results:In order to evaluate the proposed technique, ECG signals are chosen from the two physionet databases, the MIT-BIH arrhythmia database and the long-term AF database. The proposed DL framework based on the combination of CNN and LSTM showed promising results than most of the state-of-the-art methods. The proposed method reaches the mean diagnostic accuracy of 98.24%. Conclusion:A trained model for arrhythmia classification using diverse ECG signals were successfully developed and tested. Significance:This study presents a lightweight classification technique with high diagnostic accuracy compared to other notable methods, making it a potential candidate for implementation in Holter monitor devices for arrhythmia detection. Finally, we used SHapley Additive exPlanations (SHAP), the most popular Explainable Artificial Intelligence (XAI) method to understand how our model make predictions. The results indicate that those features (ECG samples) that have contributed the most to a prediction are consonant with clinicians’ decisions. Therefore, the use of interpretable models increases the trust of clinicians in AI and thus leads to decreasing the number of misdiagnoses of cardiovascular diseases.

A hybrid lightweight breast cancer classification framework using the histopathological images

A crucial element in the diagnosis of breast cancer is the utilization of a classification method that is efficient, lightweight, and precise. Convolutional neural networks (CNNs) have garnered attention as a viable approach for classifying histopathological images. However, deeper and wider models tend to rely on first-order statistics, demanding substantial computational resources and struggling with fixed kernel dimensions that limit encompassing diverse resolution data, thereby degrading the model’s performance during testing. This study introduces BCHI-CovNet, a novel lightweight artificial intelligence (AI) model for histopathological breast image classification. Firstly, a novel multiscale depth-wise separable convolution is proposed. It is introduced to split input tensors into distinct tensor fragments, each subject to unique kernel sizes integrating various kernel sizes within one depth-wise convolution to capture both low- and high-resolution patterns. Secondly, an additional pooling module is introduced to capture extensive second-order statistical information across the channels and spatial dimensions. This module works in tandem with an innovative multi-head self-attention mechanism to capture the long-range pixels contributing significantly to the learning process, yielding distinctive and discriminative features that further enrich representation and introduce pixel diversity during training. These novel designs substantially reduce computational complexities regarding model parameters and FLOPs, which is crucial for resource-constrained medical devices. The outcomes achieved by employing the suggested model on two openly accessible datasets for breast cancer histopathological images reveal noteworthy performance. Specifically, the proposed approach attains high levels of accuracy: 99.15 % at 40× magnification, 99.08 % at 100× magnification, 99.22 % at 200× magnification, and 98.87 % at 400× magnification on the BreaKHis dataset. Additionally, it achieves an accuracy of 99.38 % on the BACH dataset. These results highlight the exceptional effectiveness and practical promise of BCHI-CovNet for the classification of breast cancer histopathological images.

Lightweight hybrid lithium-ion battery thermal management system based on 3D-printed scaffold

Electric vehicles have been developed rapidly to alleviate energy shortages and environmental pollution. However, battery thermal management is still challenge to the complex structure, heavy weight, and limited heat dissipation under harsh conditions. To address these issues, a honeycomb hybrid thermal management system, which integrates the multi-layered liquid-cooled channels into a 3D-printed metal scaffold to enhance the thermal exchange effect. The impacts of the coolant inlet and outlet arrangement, channel layer number, porosity, coolant flow rate, and continuous charge-discharge cycle on heat dissipation were numerically analyzed at 35 °C and discharge rate of 3C to evaluate the system cooling performance. The results show that the heat dissipation efficiency of the proposed system is greatly improved compared with that of the system with pure phase change materials. An 85 % porosity of 3D-printed metal scaffold performs excellent heat dissipation and low system weight, the system weight can be reduced by 60 % compared to a conventional system with aluminum solid cold plate with a same volume. The system's maximum temperature and temperature difference are kept at 48.1 °C and 2.6 °C, respectively. Additionally, the system performance was validated in the continuous charge-discharge cycle test. This work will help develop the lightweight BTMS with excellent thermal performance for electric vehicles.

Driver distraction detection using semi-supervised lightweight vision transformer

The continuously increasing number of traffic accidents necessitates addressing distracted driving, which is responsible for numerous fatalities. Enhancing driver behavior recognition, particularly through developing a highly reliable Advanced Driver Assistance System (ADAS), holds substantial potential for cultivating safer transportation systems. Inspired by the success of Convolutional Neural Networks (CNNs), various networks have been proposed to enhance the detection accuracy of distracting behaviors. However, existing models have too many parameters and rely heavily on extensive labeled data, leading to time-consuming labeling and large models. To address these limitations, we propose a distracted behavior detection method based on a lightweight vision transformer trained using pseudo-label-based semi-supervised learning to learn discriminative representations from labeled and unlabeled data while ensuring a small model and speedy inference. Specifically, we create strong and weak augmented versions of the input minibatch and employ a hybrid lightweight transformer model in a teacher-student network. Pseudo-labels are generated from the teacher's predictions on weakly augmented data. The student model aligns with these labels on strongly augmented input, with teacher parameters evolving through exponential moving average. Our method presents a real-time, accurate solution for distracted driver detection, with the potential to significantly enhance road safety by reducing accidents. The effectiveness of the proposed approach is demonstrated through a comparative evaluation against alternative fully-supervised and semi-supervised methods. Furthermore, our method is evaluated in naturalistic driving settings with varying lighting and complex backgrounds. Experiments conducted on two publicly available driver distraction detection datasets show that our method outperforms current state-of-the-art approaches.

Study on the interfacial microstructure evolution of aluminum and CFRTP on FSLW joints: influence of pin length

Polymer/metal hybrid lightweight structures are increasingly employed in automotive, aerospace and electronics industries. Friction stir welding (FSW) is applied for the effective welding of metals and thermoplastic materials. For the welding of dissimilar materials, due to the significant property differences, it is hard to achieve high strength welded joints with smooth appearance. In this article, the influence of pin length is studied by experiments for the friction stir lap welding (FSLW) of Al alloy and carbon-fiber-reinforced thermoplastic (CFRTP). Tensile tests, OM, SEM and EBSD are carried out to evaluate the welding performance. Results indicate that the joint with smooth surface topography is obtained at the pin length of 2.5 mm with a peak temperature of 250.3 °C and the CFRTP matrix PEI thermal decomposition temperature residence time of 47.6 s. From OM and SEM micro analysis, the width of the bonding area is 3.02 mm and a thin uniform interaction layer is formed in the joint. EBSD confirms that the Al alloy in the stirring zone is refined after recrystallization. It is difficult to form a stacked layered structure in the bottom of joint, so the property of the joint is improved. The tensile strength is 32.9 MPa, which is 63.1% of CFRTP.

A lightweight GAN-based fault diagnosis method based on knowledge distillation and deep transfer learning

Generative adversarial networks (GANs) have shown promise in the field of small sample fault diagnosis. However, it is worth noting that generating synthetic data using GANs is time-consuming, and synthetic data cannot fully replace real data. To expedite the GAN-based fault diagnostics process, this paper proposes a hybrid lightweight method for compressing GAN parameters. First, three modules are constructed: a teacher generator, a teacher discriminator, and a student generator, based on the knowledge distillation GAN (KD-GAN) approach. The distillation operation is applied to both teacher generator and student generator, while adversarial training is conducted for the teacher generator and the teacher discriminator. Furthermore, a joint loss function is proposed to update the parameters of the student generator by combining distillation loss and adversarial loss. Additionally, the proposed KD-GAN method is combined with deep transfer learning (DTL) and leverages real data to enhance the diagnostic model’s performance. Two numerical experiments are performed to demonstrate that the proposed KD-GAN-DTL method outperforms other GAN-based fault diagnosis methods in terms of computational time and diagnostic accuracy.

A wind speed forecasting method based on EMD-MGM with switching QR loss function and novel subsequence superposition

The ultra-short-term forecasting of wind speed is of great significance to the stable power supply of the power system. Current wind speed forecasting methods aim to improve forecasting precision while disregarding model training speed and model deployment complexity. This research proposes a lightweight hybrid model named SLF-EMD-MGM-NS for wind speed forecasting. EMD-MGM is designed as the network’s fundamental structure for reducing the hybrid model’s training time and ensuring the hybrid model has high forecasting precision. The study presents the switching loss function (SLF) mechanism. When the quantile is 0.5, an MSE-based loss function is employed for training all subsequences. When the quantile is 0.5 and 0.95, first use the wind speed fluctuation threshold to select primary subsequence, and then use the Log-Cosh-based loss function for training primary subsequences. The SLF mechanism can increase point prediction precision and interval prediction boundary stability. Moreover, a novel subsequence superposition (NS) mechanism is proposed for getting high confidence level and narrow-width interval prediction results. The NS mechanism superimposes the interval prediction results of the fluctuation subsequence with the point prediction results of the model to generate the final interval prediction results. According to the experimental results, the SLF-EMD-MGM-NS model has a high confidence level, acceptable prediction results, a narrow-width interval prediction result, and a significantly shorter training time than the other hybrid models.