Compute-in-Read-Only-Memory (CiROM) accelerators offer outstanding energy efficiency for CNNs by eliminating runtime weight updates. However, their scalability to Large Language Models (LLMs) is fundamentally constrained by their vast parameter sizes. Notably, LLaMA-7B—the smallest model in LLaMA series—demands more than 1,000 cm<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> of silicon area even in advanced CMOS nodes. This paper presents BitROM, the first CiROM-based accelerator that overcomes this limitation through co-design with BitNet’s 1.58-bit quantization model, enabling practical and efficient LLM inference at the edge. BitROM introduces three key innovations: 1) a novel Bidirectional ROM Array that stores two ternary weights per transistor; 2) a Tri-Mode Local Accumulator optimized for ternary-weight computations; and 3) an integrated Decode-Refresh (DR) eDRAM that supports on-die KV-cache management, significantly reducing external memory access during decoding. In addition, BitROM integrates LoRA-based adapters to enable efficient transfer learning across various downstream tasks. Evaluated in 65 nm CMOS, BitROM achieves 20.8 TOPS/W and a bit density of 4,967 kB/mm<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup>—offering a $10 \times$ improvement in area efficiency over prior digital CiROM designs. Moreover, the DR eDRAM contributes to a 43.6% reduction in external DRAM access, further enhancing deployment efficiency for LLMs in edge applications. Code is available at https://github.com/Wenlun-Zhang/BitROM

Analog compute-in-memory (ACIM) architectures offer significant throughput and energy benefits by performing multiplication-and-accumulation (MAC) operations directly within memory arrays. However, their overall efficiency is fundamentally constrained by the high power consumption of the per-column high-resolution analog-to-digital converters (ADCs) required to support modern DNNs (e.g., transformers), many of which demand both high computational precision and large-column throughput. In conventional ADC designs, energy in the noise-limited regime scales near-exponentially, typically by$4\times $per additional bit, making high-resolution ADCs on every column power-prohibitive. This article proposes a high-precision and power-efficient multicolumn residue accumulation (MCRA) ACIM architecture to efficiently support precision-demanding modern DNNs. Each column uses a low-resolution coarse ADC (cADC), while the per-column residuals are accumulated and further quantized by an energy-efficient time-domain multi-input incremental sigma–delta (Mi-$\Sigma \Delta $) fine ADC (fADC). This approach amortizes the near-exponential energy growth across columns, while exploiting the more favorable power-resolution scaling of the time-domain Mi-$\Sigma \Delta $quantization. Postlayout simulations demonstrate a 66.2-dB signal-to-noise-and-distortion ratio (SNDR) per column at only 1/21 the energy of the baseline with per-column high-resolution ADCs, and$0.405\times $(1/2.47) the energy of an energy-saving ADC per column, which achieves a similar SNDR. Circuit- and system-level simulations demonstrate that our MCRA CIM architecture achieves negligible accuracy degradation in precision-demanding ViT tasks while delivering high energy and area efficiency.

Accurate localization is essential for enabling modern full self-driving services. These services heavily rely on map-based traffic information to reduce uncertainties in recognizing lane shapes, traffic light locations, and traffic signs. Achieving this level of reliance on map information requires centimeter-level localization accuracy, which is currently only achievable with LiDAR sensors. However, LiDAR is known to be vulnerable to spoofing attacks that emit malicious lasers against LiDAR to overwrite its measurements. Once localization is compromised, the attack could lead the victim off roads or make them ignore traffic lights. Motivated by these serious safety implications, we design SLAMSpoof, the first practical LiDAR spoofing attack on localization systems for self-driving to assess the actual attack significance on autonomous vehicles. SLAMSpoof can effectively find the effective attack location based on our scan matching vulnerability score (SMVS), a point-wise metric representing the potential vulnerability to spoofing attacks. To evaluate the effectiveness of the attack, we conduct real-world experiments on ground vehicles and confirm its high capability in real-world scenarios, inducing position errors of$\geq 4.2$meters (more than typical lane width) for all 3 popular LiDAR-based localization algorithms. We finally discuss the potential countermeasures of this attack. Code is available at https://github.com/Keio-CSG/slamspoof.

Static random-access memory (SRAM)-based analog compute-in-memory (ACiM) demonstrates promising energy efficiency for deep neural network (DNN) processing. Nevertheless, efforts to optimize efficiency frequently compromise accuracy, and this trade-off remains insufficiently studied due to the difficulty of performing full-system validation. Specifically, existing simulation tools rarely target SRAM-based ACiM and exhibit inconsistent accuracy predictions, highlighting the need for a standardized, SRAM compute-in-memory (CiM) circuit-aware evaluation methodology. This article presents ASiM, a simulation framework for evaluating inference accuracy in SRAM-based ACiM systems. ASiM captures critical effects in SRAM-based analog compute in memory systems, such as analog-to-digital converter (ADC) quantization, bit-parallel encoding, and analog noise, which must be modeled with high fidelity due to their distinct behavior in charge-domain architectures compared to other memory technologies. ASiM supports a wide range of modern DNN workloads, including CNN and Transformer-based models such as ViT, and scales to large-scale tasks like ImageNet classification. Our results indicate that bit-parallel encoding can improve energy efficiency with only modest accuracy degradation; however, even 1 LSB of analog noise can significantly impair inference performance, particularly in complex tasks such as ImageNet. To address this, we explore hybrid analog-digital execution and majority voting schemes, both of which enhance robustness without negating energy savings. ASiM bridges the gap between hardware design and inference performance, offering actionable insights for energy-efficient, high-accuracy ACiM deployment. The code is available athttps://github.com/Keio-CSG/ASiM

Computing-in-Memory (CIM) has shown great potential for enhancing efficiency and performance for deep neural networks (DNNs). However, the lack of flexibility in CIM leads to an unnecessary expenditure of computational resources on less critical operations, and a diminished Signal-to-Noise Ratio (SNR) when handling more complex tasks, significantly hindering the overall performance. Hence, we focus on the integration of CIM with Saliency-Aware Computing—a paradigm that dynamically tailors computing precision based on the importance of each input. We propose On-the-fly Saliency-Aware Hybrid CIM (OSA-HCIM) offering three primary contributions: (1) On-the-fly Saliency-Aware (OSA) precision configuration scheme, which dynamically sets the precision of each multiply-and-accumulate (MAC) operation based on its saliency, (2) Hybrid CIM Array (HCIMA), which enables simultaneous operation of digital-domain CIM (DCIM) and analog-domain CIM (ACIM) via split-port 6T SRAM, and (3) an integrated framework combining OSA and HCIMA to fulfill diverse accuracy and power demands.Implemented on a 65nm CMOS process, OSA-HCIM demon-strates an exceptional balance between accuracy and resource utilization. Notably, it is the first CIM design to incorporate a dynamic digital-to-analog boundary, providing unprecedented flexibility for saliency-aware computing. OSA-HCIM achieves a 1. 95x enhancement in energy efficiency, while maintaining minimal accuracy loss compared to DCIM when tested on CIFAR100 dataset.

Approximate computing emerges as a promising approach to enhance the efficiency of compute-in-memory (CiM) systems in deep neural network processing. However, traditional approximate techniques often significantly trade off accuracy for power efficiency, and fail to reduce data transfer between main memory and CiM banks, which dominates power consumption. This paper introduces a novel probabilistic approximate computation (PAC) method that leverages statistical techniques to approximate multiply-and-accumulation (MAC) operations, reducing approximation error by 4× compared to existing approaches. PAC enables efficient sparsity-based computation in CiM systems by simplifying complex MAC vector computations into scalar calculations. Moreover, PAC enables sparsity encoding and eliminates the LSB activations transmission, significantly reducing data reads and writes. This sets PAC apart from traditional approximate computing techniques, minimizing not only computation power but also memory accesses by 50%, thereby boosting system-level efficiency. We developed PACiM, a sparsity-centric architecture that fully exploits sparsity to reduce bit-serial cycles by 81% and achieves a peak 8b/8b efficiency of 14.63 TOPS/W in 65 nm CMOS while maintaining high accuracy of 93.85/72.36/66.02% on CIFAR-10/CIFAR-100/ImageNet benchmarks using a ResNet-18 model, demonstrating the effectiveness of our PAC methodology. Software simulation framework is available at GitHub.

LiDAR (Light Detection And Ranging) is an indispensable sensor for precise long- and wide-range 3D sensing of the surrounding environment. The recent rapid deployment of autonomous driving (AD) has highly benefited from the advancement of LiDARs. At the same time, the safety-critical application strongly motivates its security research. Recent studies demonstrate that they can manipulate the LiDAR point cloud and fool object detection by shooting malicious lasers against LiDAR scanning. However, prior efforts focus on limited types of LiDARs and object detection models, and their threat models are not clearly validated in the real world. To fill the critical research gap, we plan to conduct the first large-scale measurement study on LiDAR spoofing attacks against a wide variety of LiDARs with major object detectors. To perform this measurement, we first significantly improved the LiDAR spoofing capability (30x more spoofing points than the prior attack) with more careful optics and functional electronics, which allows us to be the first to clearly demonstrate and quantify key attack capabilities assumed in prior works. In this poster, we present our preliminary results on VLP-16 and our research plan.

The quality of robot vision greatly affects the performance of automation systems, where occlusions stand as one of the biggest challenges. If the target is occluded from the sensor, detecting and grasping such objects become very challenging. For example, when multiple robot arms cooperate in a single workplace, occlusions will be created under the robot arm itself and hide objects underneath. While occlusions can be greatly reduced by installing multiple sensors, the increase in sensor costs cannot be ignored. Moreover, the sensor placements must be rearranged every time the robot operation routine and layout change.To diminish occlusions, we propose the first robot vision system with tilt-type mirror reflection sensing. By instantly tilting the sensor itself, we obtain two sensing results with different views: conventional direct line-of-sight sensing and non-line-of-sight sensing via mirror reflections. Our proposed system removes occlusions adaptively by detecting the occlusions in the scene and dynamically configuring the sensor tilt angle to sense the detected occluded area. Thus, sensor rearrangements are not required even after changes in robot operation or layout. Since the required hardware is the tilt-unit and a commercially available mirror, the cost increase is marginal. Through experiments, we show that our system can achieve a similar detection accuracy as systems with multiple sensors, regardless of the single-sensor implementation.

A voltage-controlled oscillator (VCO)-based comparator that automatically adapts its noise performance reflecting the input voltage difference ( $\Delta V_{\text {in}}$ ) is presented. Such adaptive operation significantly reduces the power of high-precision comparators in successive-approximation-register (SAR) ADCs. $\Delta V_{\text {in}}$ is integrated as a time difference via the VCO, where the integration continues as long as the time difference is below a certain threshold, defined by the phase detector deadzone. Thus, when $\Delta V_{\text {in}}$ is large, the comparator operates as a low-power delay line-based comparator, and with small $\Delta V_{\text {in}}$ , the VCO oscillates to integrate the input signal and suppresses the comparator noise. The required oscillations to complete the comparison are inversely proportional to $\Delta V_{\text {in}}$ , realizing fully adaptive noise and power scaling. This article provides a detailed analysis and specific design guidelines of the VCO comparator. Moreover, the PVT drift tolerance and detailed circuit implementations are deeply discussed as well. For proof-of-concept, a 13-bit SAR ADC with the proposed VCO-based comparator was fabricated in 65-nm CMOS. By off-chip LMS calibration, the ADC achieves peak SNDR 66 dB at 1 MS/s with a peak FoM of 29 fJ/conv.-step.

Real-time CNN-based object detection models for applications like surveillance can achieve high accuracy but are computationally expensive. Recent works have shown 10 to 100× reduction in computation cost for inference by using domain-specific networks. However, prior works have focused on inference only. If the domain model requires frequent retraining, training costs can pose a significant bottleneck. To address this, we propose Dataset Culling: a pipeline to reduce the size of the dataset for training, based on the prediction difficulty. Images that are easy to classify are filtered out since they contribute little to improving the accuracy. The difficulty is measured using our proposed confidence loss metric with little computational overhead. Dataset Culling is extended to optimize the image resolution to further improve training and inference costs. We develop fixed-angle, long-duration video datasets across several domains, and we show that the dataset size can be culled by a factor of 300× to reduce the total training time by 47× with no accuracy loss or even with slight improvement.

To realize high-resolution pipelined and pipelinedSAR analog-to-digital-converters (ADCs), an accurate residue amplifier is necessary. However, realizing such an amplifier in scaled CMOS is challenging due to the worsened transistor characteristics. Prior works focused on gain calibration techniques to mitigate the use of low-gain amplifiers, in return of system complexity and prolonged startups. In this paper, we introduce a digital amplifier (DA) technique to realize powerefficient and accurate amplification in scaled CMOS. DA cancels out all errors (i.e., gain error, nonlinearity, incomplete settling, power supply noise, and thermal noise) of the low-gain amplifier by feedback based on successive approximation. The DA accuracy can be arbitrary set by configuring the number of bits in the DA capacitor digital-to-analog-converter; the amplifier gain is decoupled from the transistor intrinsic gain which is suitable for scaled CMOS integration. We also show that the power efficiency can be enhanced over conventional opamp-based designs with relaxed settling error requirements of DA-based multiplying digitalto-analog-converters (MDACs). Moreover, the circuit design of DA-based MDACs is further discussed. Measurement results of the calibration-free 0.7-V 12-bit 160-MS/s pipelined-SAR ADC implemented in 28-nm CMOS are reported. Without calibration, the ADC achieves signal-to-noise-and-distortion-ratio = 61.1 dB, figure-of-merit = 12.8 fJ/conv., which is over 3× improvement compared with conventional calibration-free high-speed pipelined ADCs. In addition, we evaluate the DA's process scalability by comparing the measured results of the DA-based MDAC prototyped in 65and 28-nm CMOS. We observe 2×-3× improvement in speed, power, and area mainly resulting from the DA's process scalability.

A 0.3-0.8 V low-power 2-bit/step asynchronous successive approximation register analog-to-digital converter (ADC) is presented. A low-power 2-bit/step operation technique is proposed which uses dynamic threshold configuring comparator instead of multiple digital-to-analog converters (DACs). Power and area overhead is minimized by successively activated comparators. The comparator threshold is configured by simple Vcm biased current source, which keep the ADC free from power supply variations over 10%. Simple digital calibration is enabled by generating the reference internally. The prototype ADC fabricated in a 40 nm CMOS achieved a 44.3 dB signal-to-noise-plus-distortion ratio (SNDR) with 6.14 MS/s at a single supply voltage of 0.5 V. The ADC achieved a peak FoM of 4.8 fJ/conv-step at 0.4 V and operates down to 0.3 V.

An extremely low power and area efficient threshold configuring ADC (TC-ADC) for time interleaved ADC is proposed. The threshold configuring comparator (TCC) performs a binary search. 5b conversion is carried out by TCC with source voltage shifting technique. Additional 2b resolution is achieved by the proposed threshold interpolation (TI) technique with only 15% power overhead. Prototype ADC in 40nm CMOS occupies a core area of only 0.0038mm2and when calibration circuit included, 0.0058 mm2. With a supply voltage of 0.7V, the ADC achieves 7.0 ENOB with 24MS/s. Peak FoM of 9.8fJ/conv. is obtained at 0.5V supply, which is over 15x improvement compared with conventional TC-ADC.

An extremely low-voltage operating high speed and low power 2bit/step asynchronous SAR ADC is presented. Wide range dynamic threshold configuring comparator is proposed to enable power and area efficient 2bit/step operation. By configuring the comparator threshold by simple Vcmbiased current sources, the ADC holds immunity against 10% power supply variation. The prototype ADC fabricated in 40nm CMOS achieved 44.3 dB SNDR with 6.14 MS/s at a single supply voltage of 0.5 V. The ADC achieved a peak FoM of 5.9fJ/conv-step at 0.4V and operates down to 0.35V.

A 2nd order discrete time (DT) ΔΣ modulator with very wide supply voltage range of 0.25–1.8 V is presented. An inverter-opamp self-configuring amplifier (IOSA) is proposed with sufficient gain through sub-threshold to strong inversion region. The amplifier senses the supply voltage and automatically configures either to operate as an inverter or an opamp. The prototype ADC fabricated in 180nm CMOS achieved SNDR of 53dB at a supply voltage of 0.3V and a best FoM of 0.09 pJ/step. At 1.8 V supply the modulator achieves SNDR 61dB and operates down to 0.25V.