The SM4 cryptographic algorithm is one of the standards of commercial cryptographic algorithm in China, which is used widely in the data encryption and other application scenarios for the advantages of less resources and fast speed. The attackers attack the first and last round of SM4 algorithm generally in the side channel attack. The number of the invalid random rounds before and after the SM4 cryptographic algorithm can make the side channel information collected by the attacker invalid in order to resist the side channel attack; At the same time, the multistage pipelining method is adopted to realize the SM4 cryptographic algorithm at the high speed, that is, four rounds of operations are realized every cycle, and one SM4 grouping operation is completed in eight cycles. The implementation results of the FPGA show that the clock frequency of SM4 cipher operation is up to 100MHz and the throughput rate under ECB mode is up to 1.6gbps, which has the advantages of high performance and anti-side channel attack.
10 February 2025 in IEEE Transactions on Circuits and Systems I: Regular Papers
An LDO that utilizes Bipolar-Junction-Transistor (BJT) as input differential pair with base current compensation structure is proposed. The NPNs and PNPs complementary input structure is designed to suppress both the flicker and thermal noise of the LDO and ensure the rail-to-rail common mode input range. The base current compensation structure is designed to ensure the output accuracy of the LDO. Moreover, by designing the current-mode reference, the proposed LDO eliminates the thermal noise from the feedback resistor network and the amplification of the noise due to the feedback coefficient. The proposed LDO was fabricated with 180nm BCD double well process with a die area of about 2.884mm $^{\mathbf{2}}$. The measured output noise of the LDO is only $\mathbf{248nV/}\sqrt {\mathbf{Hz}} $ at 10Hz, and the integrated RMS noise is 1.44$\mu $V from 10Hz to 100kHz. The LDO supports a wide supply rail range from 2.3V to 20V and a wide output range from 0.1V to 19.7V. The measured Line and Load regulations are 0.035mV/V and 0.0008mV/mA, respectively.
This paper presents an adaptive biased NMOS LDO which contains an adaptively biased pulse-frequency-modulated (PFM) charge pump to improve efficiency. Compared with conventional architecture, the ground current of the proposed LDO is halved under 3mA load current by regulating the charge pump frequency based on the sensed output current. In addition, the power supply rejection is also improved to 66dB @ 10 kHz due to the relatively stable charge pumped. The proposed LDO achieved 99.995% current efficiency under 1A load. The proposed LDO was built with 0.18μm BCD technology. The simulated line and load regulations were 0.199mV/V and 53nV/mA, respectively.
19 May 2025 in IEEE Transactions on Power Electronics
This article proposes a symmetrical inductor current based spread-spectrum frequency modulation (SIC-SSFM) scheme for buck converters, aiming at optimizing both the electromagnetic interference (EMI) and the output voltage ripple. Compared with conventional frequency modulation approach, the proposed symmetrical structure helps minimize the averaged inductor current ripple and thus the output ripple. By utilizing the proposed method, the proposed buck converter eliminates the need for oscillators or any slope compensation, effectively optimizing the quiescent power consumption. The proposed converter supports input and output ranges of 5–40 V and 3.3–5 V, respectively. It is capable of delivering 2A maximum load while consumes only 8.4 μA quiescent current. The measured output ripple is 9–12.5 mV, achieving a peak efficiency of 93.36%, while the measured EMI is reduced by 18.5 dB due to the proposed SIC-SSFM. The active SIC-SSFM circuit occupies an area of 0.06 mm2, with a total buck area of 2.32 mm×1.88 mm.
6 November 2023 in IEEE Transactions on Nuclear Science
A high amount of hits on the vertex detector of the International Linear Collider are generated by the beam background, which leads to an increase in the data flow of the detector system. Charged particles coming from the beam background have low momentum, resulting in the generation of elongated clusters. The CMOS pixel sensor, which integrates pre-processing functions and on-chip artificial neural networks, could remove these elongated clusters. Clustering is the first step for data pre-processing and is used to collect clusters from raw data. In this paper, a pixel-level clustering algorithm with a 5×5 window executed in real-time is proposed. The algorithm is tested using 4500 frames (500 frames for each angle of incidence) of raw data (12bits/pixel) from a MIMOSA-18 sensor and compared to conventional clustering algorithms. The clustering implementation for an example array of 5×5 pixels is synthesized for different frequencies (100 and 200MHz) and ADC resolutions (4 and 8bits). The power dissipation and occupied area of the different implementations are analyzed. The hardware implementation of the algorithm provides the possibility to integrate the clustering function into the CMOS pixel sensor.
10 February 2026 in IEEE Transactions on Circuits and Systems II: Express Briefs
This brief presents a 60 V chopper-stabilized operational amplifier (opamp) with an 8 MHz gain-bandwidth product (GBW) and precision calibration. A novel high-voltage adaptive clock-bootstrap technique is proposed, compensates for temperature variations of the threshold voltage, and maintains a constant overdrive voltage. The opamp integrates a $500\times $ gain module to amplify and calibrate the non-chopped offset voltage, thereby reducing output ripple. It adopts a calibration scheme with temperature-compensated bit weights which are proportional to the transconductance of the chopper-path input transistor pair. Fabricated in a 0.18-$\mu $ m Bipolar-CMOS-DMOS process, the amplifier employs input-referred offset and ripple minimization techniques while ensuring consistent performance over a wide input common-mode voltage range and temperature span, achieving a 3-$\sigma $ offset voltage of $2~\mu $ V across 100 samples (-$40~^{\circ }$ C to $125~^{\circ }$ C), an offset drift of 7 nV/°C, and a maximum input-referred ripple of $0.12~\mu $ VRMS(Root Mean Square).