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 the 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 28 nm HPC+ process technology, the estimated power is 77.8 µW ∼ 115.9µW, the energy efficiency can achieve 163 TOPS/W, which is over 1.8× better than the state-of-the-art architecture.
A long-gauge fiber Bragg grating (FBG) strain sensor with enhanced strain sensitivity is proposed, which is encapsulated with two T-shaped metal blocks. Its fabrication method is described briefly, and the strain sensitivity can be flexibly adjusted through changing its packaging method. A series of experiments are carried out to study the packaging and its sensing properties. The experimental results show that the strain and temperature sensitivity coefficient of the sensor are three times larger than the common FBG sensors. The linearity coefficients of the FBG sensor are larger than 0.999, and the relative error of the repeatability of all sensor samples is less than 1%. Through the stability test on the actual bridge, it is revealed that the long-term stability of the sensor is excellent, and the maximum error is less than 1.5%. In addition, the proposed FBG strain sensors are used to conduct a shear strengthening experiment on a reinforced concrete (RC) beam to verify its working performance. The experimental results show that the strain change and crack propagation of the RC beam are well monitored by the sensors during the loading process.
Many smart grid communications are delay sensitive and have very strict timing requirements for message deliveries. For example, trip protection messages must be delivered to the destination within 3 ms according to IEC 61850. Such timecritical communications are vulnerable to flooding attacks which attempt to increase message delivery delay through congesting the network channel and exhausting the computation resources of the communicating nodes. However, there is a lack of understanding on how much flooding attacks affect message delivery delays. In this paper, we conduct experimental studies to investigate how flooding attacks affect message delivery delays for time-critical communications in smart grid. Our experiments are based on both wireless networks in a lab and wired networks in a real, industry-standard electric power facility. Experimental results show that even low-rate flooding attacks can significantly increase the message delivery delays, especially when wireless networks are used.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.