Search Results (1,282)

Deep learning accelerators play a crucial role in enhancing computation-intensive AI applications. Optimizing system resources—such as shared caches, on-chip SRAM, and data movement mechanisms—is essential for achieving peak performance and energy efficiency. This paper explores the trade-off between last-level cache (LLC) and scratchpad memory (SPM) usage in accelerator-based SoCs. To evaluate this trade-off, we introduce a high-speed simulator for estimating the timing performance of complex SoCs and demonstrate the benefits of SPM utilization. Our work shows that dynamic reconfiguration of the LLC into an SPM with prefetching capabilities reduces cache misses while improving resource utilization, performance, and energy efficiency. With SPM usage, we achieve up to 13× speedup and a 10% reduction in energy consumption for CNN backbones. Additionally, our simulator significantly outperforms state-of-the-art alternatives, running 3000× faster than gem5-SALAM for fixed-weight convolution computations and up to 64,000× faster as weight size increases. These results validate the effectiveness of both the proposed architecture and simulator in optimizing deep learning workloads. Full article

Herein, we present a novel method for junction temperature monitoring of GaN HEMT devices to achieve real-time temperature perception at different locations on the device surface. Through sputtering patterned Ti/Pt thermistor strips on the surface of a GaN HEMT device to construct an on-chip array junction temperature monitoring unit, the thermal distribution of the device during operation is fully reflected. The developed temperature monitoring unit exhibited a desirable temperature coefficient of resistance of 0.183%/°C in the range of 25 °C to 205 °C. Comparison with the thermal imager shows that the integrated temperature monitoring unit can accurately reflect the real-time temperature with a monitoring accuracy of more than 95%, which helps to improve the long-term reliability of GaN power devices under actual application conditions of high frequency and high power density. Full article

Mycoplasma pneumoniae (MP) is the main culprit of community-acquired pneumonia. Commonly used laboratory testing methods have many shortcomings. Serological diagnosis has low sensitivity, causing false negatives, while a quantitative real-time polymerase chain reaction (qPCR) requires large equipment and professional staff. To make up for these shortcomings, we proposed a label-free, low-cost, and small-sized ion-sensitive field-effect transistor (ISFET) array based on a low-buffered loop-mediated isothermal amplification (LAMP) assay. A complementary metal oxide semiconductor (CMOS)-based ISFET array with 512 × 512 sensors was used in this system, which responds specifically to H⁺ with a sensitivity of 365.7 mV/pH. For on-chip amplification, a low-buffered LAMP system designed for the conserved sequences of two genes, CARDS and gyrB, was applied. The rapid release of large amounts of H⁺ in the low-buffered LAMP solution led to a speedy increase in electrical signals captured by the ISFET array, eliminating the need for a sophisticated temperature cycling and optical system. The on-chip results showed that the device can accurately complete MP detection with a detection limit of about 10³ copies/mL (approximately 1 copy per reaction). In the final clinical validation, the detection results of eight throat swab samples using the ISFET sensors were fully consistent with the clinical laboratory diagnostic outcomes, confirming the accuracy and reliability of the ISFET sensors for use in clinical settings. And the entire process from sample lysis to result interpretation takes about 60 min. This platform has potential to be used for the point-of-care testing (POCT) of pathogen infections, providing a basis for the timely adjustment of diagnosis and treatment plans. Full article

Circularly polarized light (CPL) detection sensors have significant potential for applications in quantum communication and biosensing. In this work, we propose a three-layer complementary chiral metasurface (TCCM) for on-chip integration in the mid-infrared range (2–6 μm). The TCCM consists of an Al nanorod layer, a SiO₂ dielectric layer, and an Al nanoslit layer, with strong circular dichroism (CD) achieved through the symmetry breaking of the inclined rectangular rods. Finite-difference time-domain (FDTD) simulation results demonstrate that the electric fields excited by left circularly polarized (LCP) light and right circularly polarized (RCP) light exhibit different bonding and antibonding modes, which explains the CD mechanism. The CD response and spectral tunability are influenced by the angle and length of the inclined rectangular rods. Through simulation optimization of structural parameters, a maximum CD value of 0.72 is achieved. Compared to traditional multilayer chiral metasurfaces, the TCCM simplifies the fabrication process. These findings provide valuable insights and practical strategies for the development of compact infrared devices, particularly in optical communication, chiral sensing, and full-Stokes polarization detection. Full article

Extracellular vesicles (EVs) and their cargo are increasingly suggested as innovative biomarkers correlated to the diagnosis, progression and therapy of diseases like cancer. Several techniques have been developed for the specific separation of the different classes of EVs that give solutions enriched in vesicles, but still containing other unwanted components. New methods for a more efficient, reliable and automated isolation of EVs are therefore highly desirable. Here, microparticles with surfaces endowed with positive ions were exploited to separate vesicles from complex biological matrices. First, flat silicon oxide surfaces functionalized with different divalent cations were tested for their efficiency in terms of small EV capture. Small EVs pre-purified via serial ultracentrifugations were employed for these analyses. The two better-performing cations, i.e., Cu²⁺ and Ni²⁺, were then selected to functionalize magnetic microbeads to be inserted in microfluidic chips and evaluated for their efficiency in capturing EVs and for their release of biomarkers. The best protocol setup was explored for the capture of EVs from cell culture supernatants and for the analysis of a class of biomarkers, i.e., microRNAs, via RT-PCR. The promising results obtained with this on-chip protocol evidenced the potential automation, miaturization, ease-of-use and the effective speed of the method, allowing a step forward toward its integration in simple and fast biosensors capable of analyzing the desired biomarkers present in EVs, helping the spread of biomarker analysis in both clinical settings and in research. Full article

In this paper, we demonstrate a broadband and simultaneous waveguide array light source based on water-soluble CdSe/ZnS quantum dots (QDs). We initially measure the fluorescence intensity for various cladding solution concentrations along the fiber axis to assess their impact on the propagation loss; the experimental results show that the fluorescent intensity decreases with fiber length, with higher concentrations showing a more pronounced decrease. Then, we showcase a synchronous QD light source in an optofluidic chip that fluoresces in red, green, and blue (RGB) within a microfluidic channel. Finally, a 3 × 3 QD array of a fluorescent display on a single PDMS chip is demonstrated. The QD waveguide represents a compact and stable structure that is readily manufacturable, making it an ideal light source for advancing high-throughput biochemical sensing and on-chip spectroscopic analysis. Full article

With the rapid development of lightweight network models and efficient hardware deployment techniques, the demand for real-time semantic segmentation in areas such as autonomous driving and medical image processing has increased significantly. However, realizing efficient semantic segmentation on resource-constrained embedded platforms still faces many challenges. As a classical lightweight semantic segmentation network, ENet has attracted much attention due to its low computational complexity. In this study, we optimize the ENet semantic segmentation network to significantly reduce its computational complexity through structural simplification and 8-bit quantization and improve its hardware compatibility through the optimization of on-chip data storage and data transfer while maintaining 51.18% mIoU. The optimized network is successfully deployed on hardware accelerator and SoC systems based on Xilinx ZYNQ ZCU104 FPGA. In addition, we optimize the computational units of transposed convolution and dilated convolution and improve the on-chip data storage and data transfer design. The optimized system achieves a frame rate of 130.75 FPS, which meets the real-time processing requirements in areas such as autonomous driving and medical imaging. Meanwhile, the power consumption of the accelerator is 3.479 W, the throughput reaches 460.8 GOPS, and the energy efficiency reaches 132.2 GOPS/W. These results fully demonstrate the effectiveness of the optimization and deployment strategies in achieving a balance between computational efficiency and accuracy, which makes the system well suited for resource-constrained embedded platform applications. Full article

In this article, we present experimental and simulation results of a novel high-performance cascaded directional coupler-based triplexer. The device is designed to combine the wavelengths of 1530 nm, 1653.7 nm, and 2003 nm for use in spectroscopy systems targeting the detection of ammonia, methane, and carbon dioxide gases, respectively. The triplexer’s functions focus on enhancing the coupling efficiency and selectivity, while facilitating the on-chip integration of diode lasers. The experimental results demonstrate that the coupling efficiency is 82%, 73%, and 91% for the respective wavelengths of 1530 nm, 1653.7 nm, and 2003 nm. The results highlight the triplexer’s capability as a multifunctional beam combiner and an adaptable power source, essential for advanced gas sensing techniques and integrated couplers. Full article

Kidney transplantation is currently the treatment of choice for patients with end-stage kidney diseases. Although significant advancements in kidney transplantation have been achieved over the past decades, the host’s immune response remains the primary challenge, often leading to potential graft rejection. Effective management of the immune response is essential to ensure the long-term success of kidney transplantation. To address this issue, immunosuppressives have been developed and are now fully integrated into the clinical management of transplant recipients. However, the considerable inter- and intra-patient variability in pharmacokinetics (PK) and pharmacodynamics (PD) of these drugs represents the primary cause of graft rejection. This variability is primarily attributed to the polymorphic nature (genetic heterogeneity) of genes encoding xenobiotic-metabolizing enzymes, transport proteins, and, in some cases, drug targets. These genetic differences can influence drug metabolism and distribution, leading to either toxicity or reduced efficacy. The main objective of the present review is to report an historical overview of the pharmacogenetics of immunosuppressants, shedding light on the most recent findings and also suggesting how relevant is the research and investment in developing validated NGS-based commercial panels for pharmacogenetic profiling in kidney transplant recipients. These advancements will enable the implementation of precision medicine, optimizing immunosuppressive therapies to improve graft survival and kidney transplanted patient outcomes. Full article

All-optical computing is an emerging information processing technology. As a cutting-edge technology in the field of photonics, it effectively leverages the unique advantages of photons to achieve rapid computation. However, the lack of a fully functional and programmable design has slowed the progress of this type of optical computing system, especially in optical logic computing. In this paper, we design and propose a programmable photonic logic array based on all-optical computing methods. By efficiently combining on-chip photonic devices such as micro-ring resonators, we have realized a complete set of reconfigurable all-optical logic computation functions, including basic logic such as IS&NOT, AND, and OR, as well as combined logic, such as XOR and XNOR. To the best of our knowledge, the proposed architecture not only introduces three structurally similar standard logic units but also allows for their multiple-level cascading to form a large-scale photonic logic array, enabling multifunctional logic computation. Furthermore, using two independent wavelengths to represent the high and low levels of logic can effectively reduce cross-talk and overlap between signals, decreasing the dependence on the strength of the optical signal and the decision threshold. Simulation results by Photonic Integrated Circuit Simulator (INTERCONNECT) demonstrate the effectiveness and feasibility of the proposed programmable photonic logic array. Full article

Type 1 diabetes mellitus (T1D) is a chronic autoimmune disease caused by the immune-mediated destruction of insulin-producing pancreatic beta cells, resulting in the lifelong need for exogenous insulin. Over the last few years, overweight and obesity have recently emerged as growing health issues also afflicting patients with T1D. In this context, the term “double diabetes” has been coined to indicate patients with T1D who have a family history of type 2 diabetes mellitus (T2D) and/or patients with T1D who are affected by insulin resistance and/or overweight/obesity and/or metabolic syndrome. At the same time, the use of second-generation incretin analogs semaglutide and tirzepatide has substantially increased on a global scale over the last few years, given the remarkable clinical benefits of these drugs (in terms of glucose control and weight loss) in patients with T2D and/or overweight/obesity. Although the glucagon-like peptide-1 (GLP-1) receptor agonists and the novel dual GIP (glucose-dependent insulinotropic polypeptide)/GLP-1 receptor agonist tirzepatide are currently not approved for the treatment of T1D, a growing body of evidence over the last few years has shown that these medications may serve as valid add-on treatments to insulin with substantial efficacy in improving glucose control, promoting weight loss, preserving residual beta-cell function and providing other beneficial metabolic effects in patients with T1D, double diabetes and latent autoimmune diabetes in adults (LADA). This manuscript aims to comprehensively review the currently available literature (mostly consisting of real-world studies) regarding the safety and therapeutic use (for different purposes) of semaglutide and tirzepatide in patients with T1D (at different stages of the disease), double diabetes and LADA. Full article

On-chip microfluidics are advanced in vitro models that simulate lung tissue’s native 3D environment more closely than static 2D models to investigate the complex lung architecture and multifactorial processes that lead to pulmonary disease. Current microfluidic systems can be restrictive in the quantities of biological sample that can be retrieved from a single micro-channel, such as RNA, protein, and supernatant. Here, we describe a newly developed large-scale airway-on-chip model that employs a surface area for a cell culture wider than that in currently available systems. This enables the collection of samples comparable in volume to traditional cell culture systems, making the device applicable to any workflow utilizing these static systems (RNA isolation, ELISA, etc.). With our construction method, this larger culture area allows for easier handling, the potential for a wide range of exposures, as well as the collection of low-quantity samples (e.g., volatiles or mitochondrial RNA). The model consists of two large polydimethylsiloxane (PDMS) cell culture chambers under an independent flow of medium or air, separated by a semi-permeable polyethylene (PET) cell culture membrane (23 μm thick, 0.4 μm pore size). Each chamber carries a 5 × 18 mm, 90 mm² (92 mm² with tapered chamber inlets) surface area that can contain up to 1–2 × 10⁴ adherent structural lung cells and can be utilized for close contact co-culture studies of different lung cell types, including airway epithelial cells, fibroblasts, smooth muscle cells, and endothelial cells. The parallel bi-chambered design of the chip allows for epithelial cells to be cultured at the air–liquid interface (ALI) and differentiation into a dense, multi-layered, pseudostratified epithelium under biological flow rates. This millifluidic airway-on-chip advances the field by providing a readily reproducible, easily adjustable, and cost-effective large-scale fluidic 3D airway cell culture platform. Full article

This paper introduces an analog differential optoelectronic receiver (ADOR) integrated with digital slicers for short-range LiDAR systems, consisting of a spatially modulated P⁺/N-well on-chip avalanche photodiode (APD), a cross-coupled differential transimpedance amplifier (CCD-TIA) with cross-coupled active loads, a continuous-time linear equalizer (CTLE), a limiting amplifier (LA), and dual digital slicers. A key feature is the integration of an additional on-chip dummy APD at the differential input node, which enables the proposed ADOR to outperform a traditional single-ended TIA in terms of common-mode noise rejection ratio. Also, the CCD-TIA utilizes cross-coupled PMOS-NMOS active loads not only to generate the symmetric output waveforms with maximized voltage swings, but also to provide wide bandwidth characteristics. The following CTLE extends the receiver bandwidth further, allowing the dual digital slicers to operate efficiently even at high sampling rates. The LA boosts the output amplitudes to suitable levels for the following slicers. Then, the inverter-based slicers with low power consumption and a small chip area produce digital outputs. The fabricated ADOR chip using a 180 nm CMOS process demonstrates a 20 dB dynamic range from 100 μA_pp to 1 mA_pp, 2 Gb/s data rate with a 490 fF APD capacitance, and 22.7 mW power consumption from a 1.8 V supply. Full article

This paper presents a compact on-chip multiple-input multiple-output (MIMO) antenna designed for future communication systems, featuring frequency-agile elements. The antenna achieves enhanced decoupling and reduced cross-section through the integration of a metasurface, which also introduces frequency agility. Designed for the millimeter-wave band using low-loss BenzoCycloButene (BCB) polymer, the antenna is manufactured with microelectronic processes, and the dimensions are 7.54 × 7.54 × 0.055 mm³. Simulations and measurements demonstrate excellent frequency agility around 60 GHz, with gains of 6.5 to 9 dBi. As a proof of concept, open and short circuits were used for switching, with future designs aiming to incorporate diodes for a full dynamic reconfiguration. This work highlights the potential for compact, high-performance, and frequency-reconfigurable on-chip antennas in next-generation millimeter-wave systems. Full article