More Like this

1. Exploration of the use of a proportional-integral-derivative controller for mitigation of seismic base excitation in civil structures

   https://doi.org/10.1117/12.2612604  

   Ngoma, Mary ; Peckens, Courtney A. ( April 2022 , SPIE Smart Structures + Nondestructive Evaluation)

   Zonta, Daniele ; Su, Zhongqing ; Glisic, Branko (Ed.)

   Civil infrastructures are susceptible to damage due to external forces such as winds and earthquakes. These external forces cause damage to buildings and different civil structures. To prevent this, active control systems are executed. These systems use sensors to measure the displacement of the infrastructure, then actuators are utilized to provide a force that counteracts that displacement. In this study, a Proportional Integral Derivative (PID) controller was used to minimize the impact of an earthquake disturbance on multi-story structures. The proportional, integral, and derivative gains of the controller were obtained using Particle Swarm Optimization (PSO). This PID controller was validated on a simulated five-story structure based on the Kajima Shizuoka building with five ideal actuators. The effectiveness of the PID controller in reducing the seismic response of the structure with regards to inter-story displacement and acceleration was compared to the uncontrolled response of the structure. It is found that the PID controller with PID parameters obtained from the PSO algorithm offers effective control for the simulated five story structure. 

   more » « less
2. High-speed data acquisition and computing for real-time active control of civil structures subject to seismic base excitation

   https://doi.org/10.1117/12.2612570  

   Peckens, Courtney A. ( April 2022 , SPIE Smart Structures + Nondestructive Evaluation)

   Zonta, Daniele ; Su, Zhongqing ; Glisic, Branko (Ed.)

   Active structural control of civil infrastructure in response to large external loads, such as earthquake or wind, requires the rapid integration of information between sensing nodes, computational nodes, and actuating nodes. Because of this, it is still not widely employed due to several key issues, such as latency in the system and challenges with information exchange. In this study, the Martlet, a high-speed data acquisition and computing node that was designed based on a Texas Instruments Piccolo microcontroller and capable of peer-to-peer wireless communication, is used for all three steps in the active control process. For rapid sensing, the Martlet is equipped with an interface board that interfaces with a displacement transducer and has an on-board differentiating circuit to derive velocity. The sensing Martlet transmits its data (i.e., displacement and velocity) to the actuating Martlet. The actuating Martlet calculates the necessary control force using an optimal control law, the full-state linear quadratic regulator. The resulting control force is then conveyed to the actuator via a controller interface board. This complete process is experimentally validated on a partial-scale, four-story shear structure and it is demonstrated that due to the fast processing speeds of the Martlet, real-time control of the structure can be achieved. 

   more » « less
3. Tiny but Accurate: A Pruned, Quantized and Optimized Memristor Crossbar Framework for Ultra Efficient DNN Implementation

   https://doi.org/10.1109/ASP-DAC47756.2020.9045658  

   Ma, Xiaolong ; Yuan, Geng ; Lin, Sheng ; Ding, Caiwen ; Yu, Fuxun ; Liu, Tao ; Wen, Wujie ; Chen, Xiang ; Wang, Yanzhi ( January 2020 , 2020 25th Asia and South Pacific Design Automation Conference (ASP-DAC))

   The memristor crossbar array has emerged as an intrinsically suitable matrix computation and low-power acceleration framework for DNN applications. Many techniques such as memristor-based weight pruning and memristor-based quantization have been studied. However, the high accuracy solution for the above techniques is still waiting for unraveling. In this paper, we propose a memristor-based DNN framework which combines both structured weight pruning and quantization by incorporating ADMM algorithm for better pruning and quantization performance. We also discover the non-optimality of the ADMM solution in weight pruning and the unused data path in a structured pruned model. We design a software-hardware co-optimization framework which contains the first proposed Network Purification and Unused Path Removal algorithms targeting on post-processing a structured pruned model after ADMM steps. By taking memristor hardware constraints into our whole framework, we achieve extreme high compression rate with minimum accuracy loss. For quantizing structured pruned model, our framework achieves nearly no accuracy loss after quantizing weights to 8-bit memristor weight representation. We share our models at anonymous link https://bit.ly/2VnMUy0. 

   more » « less
4. Power control and beamforming design for SWIPT in AF two-way relay networks

   https://doi.org/10.1109/ICCS.2016.7833632  

   Wang, Wei ; Wang, Rui ; Mehrpouyan, Hani ; Zhang, Guoan ( December 2016 , IEEE International Conference on Communication Systems (ICCS), 2016)

   In this paper, we study the problem of joint power control and beamforming design for simultaneous wireless information and power transfer (SWIPT) in an amplify-and-forward (AF) based two-way relaying (TWR) network. The considered system model consists of two source nodes and a relay node. Two single-antenna source nodes receive information and energy simultaneously via power splitting (PS) from the signals sent by a multi-antenna relay node. Our objective is to maximize the weighted sum energy at the two source nodes subject to quality of service (QoS) constraints and the transmit power constraints. However, the joint optimization of the relay beamforming matrix, the source transmit power and PS ratio is intractable. To find a closed-form solution of the formulated problem, we decouple the primal problem into two subproblems. In the first problem, we intend to optimize the beamforming vectors for given transmit powers and PS ratio. In the second subproblem, we optimize the remaining parameters with obtained beamformers. It is worth noting that although the corresponding subproblem are nonconvex, the optimal solution of each subproblem can be found by using certain techniques. The iterative optimization algorithm finally converges. Simulation results verify the effectiveness of the proposed joint design. 

   more » « less
5. ADMM-NN: An Algorithm-Hardware Co-Design Framework of DNNs Using Alternating Direction Methods of Multipliers

   https://doi.org/10.1145/3297858.3304076  

   Ren, Ao ; Zhang, Tianyun ; Ye, Shaokai ; Li, Jiayu ; Xu, Wenyao ; Qian, Xuehai ; Lin, Xue ; Wang, Yanzhi ( April 2019 , the Twenty-Fourth International Conference on Architectural Support for Programming Languages and Operating Systems)

   Model compression is an important technique to facilitate efficient embedded and hardware implementations of deep neural networks (DNNs), a number of prior works are dedicated to model compression techniques. The target is to simultaneously reduce the model storage size and accelerate the computation, with minor effect on accuracy. Two important categories of DNN model compression techniques are weight pruning and weight quantization. The former leverages the redundancy in the number of weights, whereas the latter leverages the redundancy in bit representation of weights. These two sources of redundancy can be combined, thereby leading to a higher degree of DNN model compression. However, a systematic framework of joint weight pruning and quantization of DNNs is lacking, thereby limiting the available model compression ratio. Moreover, the computation reduction, energy efficiency improvement, and hardware performance overhead need to be accounted besides simply model size reduction, and the hardware performance overhead resulted from weight pruning method needs to be taken into consideration. To address these limitations, we present ADMM-NN, the first algorithm-hardware co-optimization framework of DNNs using Alternating Direction Method of Multipliers (ADMM), a powerful technique to solve non-convex optimization problems with possibly combinatorial constraints. The first part of ADMM-NN is a systematic, joint framework of DNN weight pruning and quantization using ADMM. It can be understood as a smart regularization technique with regularization target dynamically updated in each ADMM iteration, thereby resulting in higher performance in model compression than the state-of-the-art. The second part is hardware-aware DNN optimizations to facilitate hardware-level implementations. We perform ADMM-based weight pruning and quantization considering (i) the computation reduction and energy efficiency improvement, and (ii) the hardware performance overhead due to irregular sparsity. The first requirement prioritizes the convolutional layer compression over fully-connected layers, while the latter requires a concept of the break-even pruning ratio, defined as the minimum pruning ratio of a specific layer that results in no hardware performance degradation. Without accuracy loss, ADMM-NN achieves 85× and 24× pruning on LeNet-5 and AlexNet models, respectively, --- significantly higher than the state-of-the-art. The improvements become more significant when focusing on computation reduction. Combining weight pruning and quantization, we achieve 1,910× and 231× reductions in overall model size on these two benchmarks, when focusing on data storage. Highly promising results are also observed on other representative DNNs such as VGGNet and ResNet-50. We release codes and models at https://github.com/yeshaokai/admm-nn. 

   more » « less