• ETRI Journal
• /
• v.38 no.2
• /
• pp.217-226
• /
• 2016

We propose an architecture that reduces the power consumption and active area of such a modulator through a reduction in the number of active components and a simplification of the topology. The proposed architecture reduces the power consumption and active area by reducing the number of active components and simplifying the modulator topology. A novel second-order loop filter that uses a single operational amplifier resonator reduces the number of active elements and enhances the controllability of the transfer function. A trapezoidal-shape half-delayed return-to-zero feedback DAC eliminates the loop-delay compensation circuitry and improves pulse-delay sensitivity. These simple features of the modulator allow higher frequency operation and more design flexibility. Implemented in a 130 nm CMOS technology, the prototype modulator occupies an active area of $0.098mm^2$ and consumes 5.23 mW power from a 1.2 V supply. It achieves a dynamic range of 62 dB and a peak SNDR of 60.95 dB over a 15 MHz signal bandwidth with a sampling frequency of 780 MHz. The figure-of-merit of the modulator is 191 fJ/conversion-step.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.9 no.1
• /
• pp.17-24
• /
• 2014

In this paper, we implement and evaluate the performance of a vector-based rasterization algorithm for 3D graphics by using a SIMD (single instruction multiple data) many-core processor architecture. In addition, we evaluate the impact of a data-per-processing elements (DPE) ratio that is defined as the amount of data directly mapped to each processing element (PE) within many-core in terms of performance, energy efficiency, and area efficiency. For the experiment, we utilize seven different PE configurations by varying the DPE ratio (or the number PEs), which are implemented in the same 130 nm CMOS technology with a 500 MHz clock frequency. Experimental results indicate that the optimal PE configuration is achieved as the DPE ratio is in the range from 16,384 to 256 (or the number of PEs is in the range from 16 and 1,024), which meets the requirements of mobile devices in terms of the optimal performance and efficiency.

• JSTS:Journal of Semiconductor Technology and Science
• /
• v.14 no.4
• /
• pp.419-426
• /
• 2014

This paper proposes a high-throughput encoding process and encoder architecture for quasi-cyclic low-density parity-check codes in IEEE 802.11ac standard. In order to achieve the high throughput with low complexity, a partially parallel processing based encoding process and encoder architecture are proposed. Forward and backward accumulations are performed in one clock cycle to increase the encoding throughput. A low complexity cyclic shifter is also proposed to minimize the hardware overhead of combinational logic in the encoder architecture. In IEEE 802.11ac systems, the proposed encoder is rate compatible to support various code rates and codeword block lengths. The proposed encoder is implemented with 130-nm CMOS technology. For (1944, 1620) irregular code, 7.7 Gbps throughput is achieved at 100 MHz clock frequency. The gate count of the proposed encoder core is about 96 K.

• Journal of Semiconductor Engineering
• /
• v.2 no.2
• /
• pp.130-135
• /
• 2021

Recently, deep neural networks (DNNs) are actively used for action control so that an autonomous system, such as the robot, can perform human-like behaviors and operations. Unlike recognition tasks, the real-time operation is essential in action control, and it is too slow to use remote learning on a server communicating through a network. New learning techniques, such as reinforcement learning (RL), are needed to determine and select the correct robot behavior locally. In this paper, we propose an energy-efficient DNN processor with a LUT-based processing engine and near-zero skipper. A CNN-based facial emotion recognition and an RNN-based emotional dialogue generation model is integrated for natural HRI system and tested with the proposed processor. It supports 1b to 16b variable weight bit precision with and 57.6% and 28.5% lower energy consumption than conventional MAC arithmetic units for 1b and 16b weight precision. Also, the near-zero skipper reduces 36% of MAC operation and consumes 28% lower energy consumption for facial emotion recognition tasks. Implemented in 65nm CMOS process, the proposed processor occupies 1784×1784 um2 areas and dissipates 0.28 mW and 34.4 mW at 1fps and 30fps facial emotion recognition tasks.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2015.10a
• /
• pp.416-419
• /
• 2015

In this paper, we propose an effective design of transform hardware for real-time HEVC(High Efficiency Video Coding) encoder. HEVC encoder determines the transform mode($4{\times}4$, $8{\times}8$, $16{\times}16$, $32{\times}32$) by comparing RDCost. RDCost require a significant amount of computation and time because it is determined by bit-rate and distortion which is computated via transform, quantization, dequantization, and inverse transform. This paper therefore proposes a new method for transform mode determination using sum of transform coefficient. Also, proposed hardware architecture is implemented with multiplexer, recursive adder/subtracter, and shifter only to derive reduction of the computation. Proposed method for transform mode determination results in an increase of 0.096 in BD-PSNR, 0.057 in BD-Bitrate, and decrease of 9.3% in encoding time by comparing HM 10.0. The hardware which is proposed is implemented by 256K logic gates in TSMC 130nm process. Its maximum operation frequency is 200MHz. At 140MHz, the proposed hardware can support 4K Ultra HD video encoding at 60fps in real time.

• ETRI Journal
• /
• v.42 no.4
• /
• pp.596-607
• /
• 2020

Scaling is an important operation because of the iterative nature of arithmetic processes in digital signal processors (DSPs). In residue number system (RNS)-based DSPs, scaling represents a performance bottleneck based on the complexity of intermodulo operations. To design an efficient RNS scaler for special moduli sets, a body of literature has been dedicated to the study of the well-known moduli sets {2ⁿ - 1, 2ⁿ, 2ⁿ + 1} and {2ⁿ, 2ⁿ - 1, 2ⁿ⁺¹ - 1}, and their extension in vertical or horizontal forms. In this study, we propose an efficient programmable RNS scaler for the arithmetic-friendly moduli set {2^n+p, 2ⁿ - 1, 2ⁿ⁺¹ - 1}. The proposed algorithm yields high speed and energy-efficient realization of an RNS programmable scaler based on the effective exploitation of the mixed-radix representation, parallelism, and a hardware sharing technique. Experimental results obtained for a 130 nm CMOS ASIC technology demonstrate the superiority of the proposed programmable scaler compared to the only available and highly effective hybrid programmable scaler for an identical moduli set. The proposed scaler provides 43.28% less power consumption, 33.27% faster execution, and 28.55% more area saving on average compared to the hybrid programmable scaler.

• IEIE Transactions on Smart Processing and Computing
• /
• v.4 no.5
• /
• pp.354-365
• /
• 2015

This work presents an Internet of Things (IoT) system for home energy management based on a custom-designed Impulse Radio Ultra-Wideband (IR-UWB) transceiver that targets a generic and multi-standard control system. This control system enables the interoperability of heterogeneous devices: it integrates various sensor nodes based on ZigBee, EnOcean and UWB in the same middleware by utilizing an ad-hoc layer as an interface between the hardware and software. The paper presents as a first the design of the IR-UWB transceiver for a portable sensor node integrated with the middleware layer, and also describes the receiver connected to the control system. The custom-designed low-power transmitter on the sensor node is fabricated with 130 nm CMOS technology. It generates a signal with a 1.1 ns pulse width while consuming $39{\mu}W$ at 1 Mbps. The UWB sensor node with a temperature measurement capability consumes 5.31 mW, which is lower than the power level of state-of-the-art solutions for smart-home applications. The UWB hardware and software layers necessary to interface with the control system are verified in over-the-air measurements in an actual office environment. With the implementation of the presented sensor node and its integration in the energy management system, we demonstrate achievement of the broad flexibility demanded for IoT.