Search Results (54)

We present RoBaseLM-S (125 M) and RoBaseLM-M (260 M), two compact Romanian decoder-only language models trained from scratch on a 4.3 B-token curated corpus. Architecturally, they follow a modern LLaMA-style recipe with pre-norm RMSNorm, rotary position embeddings, SwiGLU feed-forward blocks, grouped-query attention, and 4 k-token context windows. We release both full-precision (FP16) and post-training 5-bit (Q5_K_M) checkpoints in GGUF format for lightweight local inference. The 5-bit variants fit under 500 MB and generate text in real time on a Jetson Nano 4 GB, enabling fully offline Romanian text generation on consumer-grade edge hardware. We evaluate the models intrinsically (multi-domain perplexity across news, literary prose, poetry, and heterogeneous web text) and extrinsically (LaRoSeDa sentiment classification and RO-STS sentence similarity). Relative to Romanian GPT-2–style baselines at similar parameter scales, RoBaseLM-S and RoBaseLM-M reduce perplexity substantially, e.g., from 30.7 to 15.9 on our held-out news split. The 5-bit post-training quantized checkpoints remain within FP16 performance across all reported tasks. To our knowledge, these are the first Romanian small language models explicitly optimized for long-context inference, post-training quantization, and low-power on-device deployment. Full article

This work investigates the deployment of Deep Learning (DL) models for network intrusion detection on resource-constrained IoT devices, using the public CICIoT2023 dataset. In particular, we consider the following DL models: Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), Recurrent Neural Network (RNN), Convolutional Neural Network (CNN), Temporal Convolutional Network (TCN), Multi-Layer Perceptron (MLP). Bayesian optimization is employed to fine-tune the models’ hyperparameters and ensure reliable performance evaluation across both binary (2-class) and multi-class (8-class, 34-class) intrusion detection. Then, the computational complexity of each DL model is analyzed—in terms of the number of Multiply–ACCumulate operations (MACCs), RAM usage, and inference time—through the STMicroelectronics Cube.AI Analyzer tool, with models being deployed on an STM32H7S78-DK board. To assess the practical deployability of the considered DL models, a trade-off score (balancing classification accuracy and computational efficiency) is introduced: according to this score, our experimental results indicate that MLP and TCN outperform the other models. Furthermore, Post-Training Quantization (PTQ) to 8-bit integer precision is applied, allowing the model size to be reduced by more than $90\%$ with negligible performance degradation. This demonstrates the effectiveness of quantization in optimizing DL models for real-world deployment on resource-constrained IoT devices. Full article

When using optical remote sensing methods for Kelvin wake imaging, the imaging is affected by sea-surface stochastic fluctuation, imaging noise, and weak reflectivity contrast, resulting in weak wake image signals. In order to better obtain wake optical remote sensing images, this article proposes a multi-objective parameter stochastic optimization method for a time-delayed integration optical remote sensing imaging system. By constructing the wake imaging mechanism framework integrating a hydrodynamic model, rough sea surface probability and statistics model, and Time-Delay Integration Charge-Coupled Device (TDI-CCD) imaging link model, a stochastic multi-objective optimization model with constraints is established. The multi-objective function of this model is specifically defined as follows: maximizing the digital number difference between the crest and trough of Kelvin wakes in imaging results, maximizing the F number, minimizing the integration stages, and minimizing the quantization bits. Meanwhile, a two-stage solution method based on sample average approximation (SAA), branch and bound method (B&B), and the complex method is designed. The model can be used to obtain optimized design results for remote sensing imaging parameters, providing theoretical and methodological support for the design of remote sensing imaging systems. Numerical simulation results show that the optimized parameter combination can achieve clear imaging of the Kelvin wake, and the core indicators meet the design requirements. Full article

Stochastic geometry provides a powerful analytical framework for evaluating interference-limited cellular networks with randomly deployed base stations (BSs). While prior studies have examined limited channel state information at the transmitter (CSIT) and low-resolution analog-to-digital converters (ADCs) separately, their joint impact in multi-user multiple-input multiple-output (MIMO) systems remains largely unexplored. This paper investigates a downlink cellular network in which BSs are distributed according to a homogeneous Poisson point process (PPP), employing zero-forcing beamforming (ZFBF) with limited feedback, and receivers are equipped with one-bit ADCs. We derive a tractable approximation for the achievable spectral efficiency that explicitly accounts for both the quantization error from limited feedback and the receiver distortion caused by coarse ADCs. Based on this approximation, we determine the optimal feedback rate that maximizes the net spectral efficiency. Our analysis reveals that the optimal number of feedback bits scales logarithmically with the channel coherence time but its absolute value decreases due to coarse quantization. Simulation results validate the accuracy of the proposed approximation and confirm the predicted scaling behavior, demonstrating its effectiveness for interference-limited multi-user MIMO networks. Full article

Recently, the demand for immersive videos has surged with the expansion of virtual reality, augmented reality, and metaverse technologies. As an international standard, moving picture experts group (MPEG) has developed MPEG immersive video (MIV) to efficiently transmit large-volume immersive videos. The MIV encoder generates atlas videos to convert extensive multi-view videos into low-bitrate formats. When these atlas videos are compressed using conventional video codecs, compression artifacts often appear in the reconstructed atlas videos. To address this issue, this study proposes a feature-extraction-based convolutional neural network (FECNN) to reduce the compression artifacts during MIV atlas video transmission. The proposed FECNN uses quantization parameter (QP) maps and depth information as inputs and consists of shallow feature extraction (SFE) blocks and deep feature extraction (DFE) blocks to utilize layered feature characteristics. Compared to the existing MIV, the proposed method improves the Bjontegaard delta bit-rate (BDBR) by −4.12% and −6.96% in the basic and additional views, respectively. Full article

Recent progress in computer vision and embedded systems has facilitated real-time monitoring of bioprocesses; however, lightweight and scalable solutions for resource-constrained settings remain limited. This work presents a modular framework for monitoring Chlorella vulgaris growth by integrating RGB image processing with multimodal sensor fusion. The system incorporates a Logitech C920 camera and low-cost pH and temperature sensors within a compact photobioreactor. It extracts RGB channel statistics, luminance, and environmental data to generate a 10-dimensional feature vector. A feedforward artificial neural network (ANN) with ReLU activations, dropout layers, and SMOTE-based data balancing was trained to classify growth phases: lag, exponential, and stationary. The optimized model, quantized to 8 bits, was deployed on an ESP32 microcontroller, achieving 98.62% accuracy with 4.8 ms inference time and a 13.48 kB memory footprint. Robustness analysis confirmed tolerance to geometric transformations, though variable lighting reduced performance. Principal component analysis (PCA) retained 95% variance, supporting the discriminative power of the features. The proposed system outperformed previous vision-only methods, demonstrating the advantages of multimodal fusion for early detection. Limitations include sensitivity to lighting and validation limited to a single species. Future directions include incorporating active lighting control and extending the model to multi-species classification for broader applicability. Full article

The design and application of broadband phased array RF links is a complex and highly precise endeavor. To achieve optimal performance, it is essential to compare and validate multiple schemes during the system design phase. Utilizing simulation models to simulate system structures and validate parameters can effectively reduce research and development time and costs. This article takes the broadband phased array RF system (RFS04) currently being developed by Nanhu Laboratory as a reference and constructs a behavioral-level signal simulation model. Through this model, the antenna pattern of RFS04 was generated, and the relationship between beam pointing accuracy and delay quantization bit number was analyzed. The 3 dB beam coverage range of the 18 GHz antenna array was calculated, and the synthesis scheme of multi-phased arrays was explored. Additionally, the correspondence between the angle measurement accuracy and signal-to-noise ratio of the RFS04 system was analyzed. This article also measured the delay module parameters of the RF system and developed a correction strategy for the delay control scheme. Through simulation calculations and laboratory testing, it has been proven that this strategy can effectively improve delay accuracy. After applying the modified delay control scheme to the RFS04 simulation model, the beam pointing accuracy during phased array antenna scanning was significantly enhanced. The model research and integrated simulation software construction of the broadband phased array RF system provide an efficient and accurate simulation tool for system design and optimization. Full article

Meeting the rising need for secure authentication in IoT and Industry 4.0, this work presents chemically synthesized ZnO nanostructured homojunctions as powerful and scalable physical unclonable functions (PUFs). By leveraging intrinsic variability from Li doping and the stochastic hydrothermal growth process, we systematically identified electrical parameters offering outstanding variability, stability, and reproducibility. ZnO devices outperform commercial diodes by delivering richer parameter diversity, lower costs, and superior environmental sustainability. Pushing beyond traditional approaches, we introduce multi-level quantization for boosted accuracy and entropy, demonstrate the normal distribution of challenge candidate parameters to support a novel method under development, and extract multiple parameters (8–10) per device instead of relying on a single-bit output. Parameter optimization and selection are performed upfront through a rigorous assessment of variability and inter-correlation, maximizing uniqueness and reliability. Thanks to their strong scalability and eco-friendliness, ZnO-based homojunctions emerge as a dynamic, future-proof platform for building low-cost, high-security, and sustainable digital identity systems. Full article

To meet the demands of real-time and high-definition (HD) image processing applications, denoising methods must be both computationally efficient and hardware friendly. Traditional image denoising techniques are typically simple, fast, and resource-efficient but often fall short in terms of denoising performance and adaptability. This paper proposes an adjustable-threshold denoising method along with a corresponding hardware implementation designed to support the real-time processing of large-array images commonly used in image signal processors (ISPs). The proposed technique employs a LeGall 5/3 wavelet with a row-transform structure and multilevel decomposition. A 2D Pyramid VisuShrink thresholding algorithm is introduced, where the threshold is derived from the median value of the HH sub-band using a multi-stage segmentation approach. To further optimize performance, a quantization strategy with fixed-point parameter design is applied to minimize storage requirements and computational errors. A specialized hardware architecture is developed to enable the real-time denoising of 4K images while adhering to constraints on speed and resource utilization. The architecture incorporates a finite state machine (FSM) and a reusable median calculation unit to efficiently share threshold-related storage and computational resources. The system is implemented and verified on an FPGA, achieving real-time performance at a maximum frequency of 230 MHz. It supports flexible input data formats with resolutions up to 4096×4096 pixels and 16-bit depth. Comprehensive comparisons with other real-time denoising methods demonstrate that the proposed approach consistently achieves better PSNR and SSIM across various noise levels and image sizes. In addition to delivering improved denoising accuracy, the hardware implementation offers advantages in processing speed and resource efficiency while supporting a wide range of large-array images. Full article

The lack of extensive, varied, and thoroughly annotated datasets impedes the advancement of artificial intelligence (AI) for medical applications, especially colorectal cancer detection. Models trained with limited diversity often display biases, especially when utilized on disadvantaged groups. Generative models (e.g., DALL-E 2, Vector-Quantized Generative Adversarial Network (VQ-GAN)) have been used to generate images but not colonoscopy data for intelligent data augmentation. This study developed an effective method for producing synthetic colonoscopy image data, which can be used to train advanced medical diagnostic models for robust colorectal cancer detection and treatment. Text-to-image synthesis was performed using fine-tuned Visual Large Language Models (LLMs). Stable Diffusion and DreamBooth Low-Rank Adaptation produce images that look authentic, with an average Inception score of 2.36 across three datasets. The validation accuracy of various classification models Big Transfer (BiT), Fixed Resolution Residual Next Generation Network (FixResNeXt), and Efficient Neural Network (EfficientNet) were 92%, 91%, and 86%, respectively. Vision Transformer (ViT) and Data-Efficient Image Transformers (DeiT) had an accuracy rate of 93%. Secondly, for the segmentation of polyps, the ground truth masks are generated using Segment Anything Model (SAM). Then, five segmentation models (U-Net, Pyramid Scene Parsing Network (PSNet), Feature Pyramid Network (FPN), Link Network (LinkNet), and Multi-scale Attention Network (MANet)) were adopted. FPN produced excellent results, with an Intersection Over Union (IoU) of $0.64$, an F1 score of $0.78$, a recall of $0.75$, and a Dice coefficient of 0.77. This demonstrates strong performance in terms of both segmentation accuracy and overlap metrics, with particularly robust results in balanced detection capability as shown by the high F1 score and Dice coefficient. This highlights how AI-generated medical images can improve colonoscopy analysis, which is critical for early colorectal cancer detection. Full article

In the realm of product design and development, efficient retrieval and reuse of 3D CAD models are vital for optimizing workflows and minimizing redundant efforts. Manual labeling of CAD models, while traditional, is labor-intensive and prone to inconsistency, highlighting the need for automated classification systems. Multi-view convolutional neural networks (MVCNNs) offer an automated solution by leveraging 2D projections to represent 3D objects, balancing high classification accuracy with computational efficiency. Despite their effectiveness, the computational demands of MVCNNs pose challenges in large-scale CAD applications. This study investigates the use of optimization strategies, precisely pruning and quantization, in the scenario of MVCNN applied to the classification of 3D CAD mechanical models. By using different pruning and quantization strategies, we evaluate trade-offs between classification accuracy, execution time, and memory usage. In our evaluation of pruning and quantization techniques, 8-bit quantization reduced the memory used by the model from 83.78 MB to 21.01 MB, with accuracy only slightly decreasing from 93.83% to 93.59%. When applying 25% structured pruning, the model’s memory usage was reduced to 47.16 MB, execution time decreased from 133 to 97 s, and accuracy decreased to 92.14%. A combined approach of 25% pruning and 8-bit quantization achieved even better resource efficiency, with memory usage at 11.86 MB, execution time at 99 s, and accuracy at 92.06%. This combination of pruning and quantization leads to efficient MVCNN model optimization, balancing resource usage and classification performance, which is especially relevant in large-scale applications. Full article

Dual-functional radar–communication systems are extensively employed for the detection and control of unmanned aerial vehicle groups and play crucial roles in scenario monitoring. In this study, we address the downlink precoding problem in large-scale multi-user multiple-input multiple-output dual-function radar–communication systems equipped with low-resolution quantized digital-to-analog converters. To tackle this issue, we develop a weighted optimization framework that minimizes the mean squared error between the transmitted symbols and their estimates while satisfying specific radar performance requirements. Due to the complexity introduced by discrete constraints, we decompose the original problem into three sub-problems to reduce computational burden. Furthermore, we propose a dynamic projection refinement algorithm within the alternating direction method of multiplier framework to efficiently solve these sub-problems. Numerical experiments demonstrate that our proposed method outperforms existing state-of-the-art techniques, particularly in terms of bit error rate in low signal-to-noise ratio scenarios. Full article

This study presents a comprehensive workflow for developing and deploying Multi-Layer Perceptron (MLP)-based soft sensors on embedded FPGAs, addressing diverse deployment objectives. The proposed workflow extends our prior research by introducing greater model adaptability. It supports various configurations—spanning layer counts, neuron counts, and quantization bitwidths—to accommodate the constraints and capabilities of different FPGA platforms. The workflow incorporates a custom-developed, open-source toolchain ElasticAI.Creator that facilitates quantization-aware training, integer-only inference, automated accelerator generation using VHDL templates, and synthesis alongside performance estimation. A case study on fluid flow estimation was conducted on two FPGA platforms: the AMD Spartan-7 XC7S15 and the Lattice iCE40UP5K. For precision-focused and latency-sensitive deployments, a six-layer, 60-neuron MLP accelerator quantized to 8 bits on the XC7S15 achieved an MSE of 56.56, an MAPE of 1.61%, and an inference latency of 23.87 $\mu$s. Moreover, for low-power and energy-constrained deployments, a five-layer, 30-neuron MLP accelerator quantized to 8 bits on the iCE40UP5K achieved an inference latency of 83.37 $\mu$s, a power consumption of 2.06 mW, and an energy consumption of just 0.172 $\mu$J per inference. These results confirm the workflow’s ability to identify optimal FPGA accelerators tailored to specific deployment requirements, achieving a balanced trade-off between precision, inference latency, and energy efficiency. Full article

Increasingly massive image data is restricted by conditions such as information transmission and reconstruction, and it is increasingly difficult to meet the requirements of speed and integrity in the information age. To solve the urgent problems faced by massive image data in information transmission, this paper proposes a high-quality image compression algorithm based on unsupervised learning. Among them, a content-weighted autoencoder network is proposed to achieve image compression coding on the basis of a smaller bit rate to solve the entropy rate optimization problem. Binary quantizers are used for coding quantization, and importance maps are used to achieve better bit allocation. The compression rate is further controlled and optimized. A multi-scale discriminator suitable for the generative adversarial network image compression framework is designed to solve the problem that the generated compressed image is prone to blurring and distortion. Finally, through training with different weights, the distortion of each scale is minimized, so that the image compression can achieve a higher quality compression and reconstruction effect. The experimental results show that the algorithm model can save the details of the image and greatly compress the memory of the image. Its advantage is that it can expand and compress a large number of images quickly and efficiently and realize the efficient processing of image compression. Full article

The accuracy of Direction of Arrival (DOA) estimation primarily depends on the precision of the data. When the receiver uses a low-precision analog-to-digital converter (ADC), traditional DOA estimation algorithms exhibit poor accuracy. To face the challenge of multi-target DOA estimation in scenarios with low-precision ADC quantized sampling, this paper proposes a novel DOA estimation algorithm for quantized signals based on classification problems. A deep learning network was constructed using Deep Neural Networks (DNNs) and Convolutional Neural Networks (CNNs), divided into the quantized signal recovery framework and the DOA estimation framework. The DNN network is utilized to recover signals that have undergone low-precision quantization, while the CNN network addresses the classification problem to estimate the DOA from received data with an unknown number of signal sources. A comprehensive analysis of the impact of signal-to-noise ratio (SNR), the number of array elements, and the number of quantization bits on the proposed algorithm was conducted. Simulation results indicate that the proposed algorithm exhibits superior DOA estimation performance in low-precision scenarios, characterized by reduced computational complexity, thereby facilitating real-time DOA estimation. Full article