This paper proposes a high power-performance-area efficient background noise aware keyword-spotting (KWS) processor based on an optimized binarized weight network (BWN). To reduce the power consumption while maintaining the system recognition accuracy for different background noise, the KWS processor with a SNR prediction module can be adaptively configured to use dual computing modes (standard computing mode and approximate computing mode) for both high recognition accuracy under high background noise and ultra-low power consumption under low background noise. The mel-scale frequency cepstral coefficients (MFCC) module is optimized with approximate computing technologies, which can reduce the power consumption by up to 3.1x and 5.7x for high/low background noise, respectively. Based on the evaluation of the architecture design space exploration, an ultra-low power BWN accelerator with low voltage, area and leakage power and using precision self-adaptive approximate computing units was proposed. Evaluated under 22nm process technology, this work can support up to 10 keywords real time recognition with power consumption of 15.1μW for high background noise and 10.8μW for low background noise. Compared to the state-of-the-art KWS architectures, our work can achieve ultra-low power consumption (about 1.7x reduced), while maintaining high system capability and adaptability.
Utilizing the randomness caused by process variations in chip manufacturing, PUF can provide identification and verification by generating unique challenge-response pair. The output response of Arbiter PUF is due to path delay differences from different input challenge. However, due to the strong linear correlation between the response and challenge of the Arbiter PUF, the attacker can model the APUF through a machine learning algorithm. This paper proposes a challenge pre-processing structure arbiter PUF (CPP-APUF), which increases the unknowingness of the input challenge, and improves the APUF's ability to resist machine learning attacks. The 64-stage CPP-APUF is implemented based on FPGA, the machine learning algorithm is used to attack the CPP-APUF. The output response prediction accuracy is lower than 61.33%, which is effective against the modeling attack of machine learning. Finally, the challenge-response pair obtained from experimentally verifies the PUF characteristics.
This paper proposed an energy-efficient reconfigurable accelerator for keyword spotting (EERA-KWS) based on binary weight network (BWN) and fabricated in 28-nm CMOS technology. This keyword spotting system consists of two parts: the feature extraction based on Melscale Frequency Cepstral Coefficients (MFCC) and the keywords classification based on a BWN model, which is trained through the Google’s Speech Commands database and deployed on our custom. To reduce the power consumption while maintaining the system recognition accuracy, we first optimize the MFCC implementation with approximate computing techniques, including Pre-emphasis coefficient transformation, rectangular Mel filtering, Framing and FFT optimization. Then, we propose a precision self-adaptive reconfigurable accelerator with digital-analog mixed approximate computing units to process the BWN efficiently. Based on SNR prediction of background noise and post-detection of network output confidence, the BWN accelerator data path can be dynamically and adaptively reconfigured as 4, 8 or 16 bits. For the BWN accelerator, we proposed a time-delay based addition unit to process bit-wise approximate computing for the convolution layers and fully connected layers, and a LUT based unit for the activation layers. Implemented under TSMC 28nm HPC+ process technology, the estimated power is 77.8 μW ∼ 115.9μW, the energy efficiency can achieve 163 TOPS/W, which is over 1.8X better than the state-of-the-art architecture.
Spintronic based embedded magnetic random access memory (eMRAM) is becoming a foundry validated solution for the next-generation nonvolatile memory applications. The hybrid complementary metal-oxide-semiconductor (CMOS)/magnetic tunnel junction (MTJ) integration has been selected as a proper candidate for energy harvesting, area-constraint and energy-efficiency Internet of Things (IoT) systems-on-chips. Multi-VDD (low supply voltage) techniques were adopted to minimize energy dissipation in MRAM, at the cost of reduced writing/sensing speed and margin. Meanwhile, yield can be severely affected due to variations in process parameters. In this work, we conduct a thorough analysis of MRAM sensing margin and yield. We propose a current-mode sensing amplifier (CSA) named 1D high-sensing 1D margin, high 1D speed and 1D stability (HMSS-SA) with reconfigured reference path and pre-charge transistor. Process-voltage-temperature (PVT) aware analysis is performed based on an MTJ compact model and an industrial 28 nm CMOS technology, explicitly considering low-voltage (0.7 V), low tunneling magnetoresistance (TMR) (50%) and high temperature (85 °C) scenario as the worst sensing case. A case study takes a brief look at sensing circuits, which is applied to in-memory bit-wise computing. Simulation results indicate that the proposed high-sensing margin, high speed and stability sensing-sensing amplifier (HMSS-SA) achieves remarkable performance up to 2.5 GHz sensing frequency. At 0.65 V supply voltage, it can achieve 1 GHz operation frequency with only 0.3% failure rate.
The keyword spotting (KWS) system is one of the most important interfaces between humans and machines since it is usually the start of automatic speech recognition and natural language processing techniques. However, for KWS hardware, it is still a problem to make one specified chip both low power and high performed under multiple scenarios, such as in meeting rooms, on different traffic or in parks and so on, for different scenarios own wide range signal-noise-ratios (SNRs). The problem leads to the requirements of balanced design between KWS system accuracy and the hardware cost under various noise types and levels. To overcome the balanced design and tradeoff problems, a complete KWS processor including an Mel-Frequency Cepstrum Coefficients (MFCC) feature extractor and a quantized Convolutional Neural Network (QCNN) accelerator is proposed for wide SNR range and low-power KWS in this paper. Firstly, the approach to quantize CNNs into QCNNs with high accuracy is proposed with considerations of hardware-software tradeoff. With the tradeoff of KWS system accuracy and hardware cost, the 4bit/8bit dual-working-mode strategy is proposed to keep low hardware cost and high accuracy under different scenarios. To be specific, the training, tuning and validating of the CNNs and QCNNs are taken with the dataset of 10 keywords chosen from the Google Command Speech Dataset (GCSD). Secondly, a serial FFT based MFCC extractor is implemented with low power and small footprint. Finally, with a novel hybrid reuse strategy of input data and network weight, a reconfigurable and approximate computing based QCNN accelerator is designed. Implemented and verified under TSMC 22nm ULL technology, with the area of $1.42mm^{2}$ , the QCNN accelerator can achieve $5.26~\mu \text{W}/9.08~\mu \text{W}$ power consumption in 4bit/8bit work mode with accuracy of 88% and 93% respectively, which is superior to the state-of-the-art processors.