Cross-layer Adaptive Rate/Resolution Design for Energy-Aware Acquisition of Spectrally Sparse Signals Leveraging Spin-based Devices

利用基于自旋的器件实现频谱稀疏信号能量感知采集的跨层自适应速率/分辨率设计

• 批准号: 1810256
• 负责人: Nazanin Rahnavard
• 金额: $ 44.17万
• 依托单位: The University of Central Florida Board of Trustees
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2018
• 资助国家: 美国
• 起止时间: 2018-09-01 至 2024-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1810256&HistoricalAwards=false
• 关键词: Cross; layer; Adaptive; Rate; Resolution

Cross-layer Adaptive Rate/Resolution Sampling Leveraging Spin-based Devices for Sensing and Communication Systems This project will devise an adaptive framework for efficient data acquisition of spectrally sparse signals to enhance the sampling accuracy while reducing the energy consumption demands of future sensing and communication systems that are ubiquitous and vital to our modern and connected society. Novel adaptive sampling and reconstruction techniques will be designed concomitantly with a spin-based hardware approach to minimize the overall cost of data acquisition and transmission for applicability in a wide-range of communication systems and a large and important class of spectrally sparse signals, which arise in many applications such as cognitive radio networks, radar, and emerging spectrum-aware communication systems. Thus, this project will serve national interests of advancing vital technologies of communication systems with improved energy-efficiency and increased circuit density. Educational materials for undergraduate and graduate course modules will be created and disseminated, as well as an interactive website to engage and attract high school students to studies and careers in the field, including a diverse cohort of underrepresented and women learners. Broad dissemination through nano-device library webpages will be used to increase the impact while supplementing the publication of research outcomes via high-quality scholarly journals and conferences, and websites.A multi-disciplinary effort will be used to develop a systematic approach that bridges the gap between advanced theoretical research in signal processing/compressive sensing and innovative circuit designs that leverage the signal processing, memory, and thresholding capabilities inherent in emerging spin-based devices. The first research thrust focuses on investigating the tradeoffs between Sampling Rate (SR) and Quantization Resolution (QR) in the context of quantized compressive sensing (CS), under power and bandwidth constraints using dynamic optimization of SR and QR in an online manner. The energy consumption, hardware limitations, and specifics of the underlying sampler and quantizer will be optimized. Computationally-efficient signal reconstruction algorithms are investigated to reconstruct the original signal back from its non-uniform (in terms of sampling rate and quantization depth) quantized CS measurements. In the second research thrust, the investigators will research and design an Intermittent Spin-based Adaptive Quantizer which utilizes Voltage-Controlled Magnetic Anisotropy Magnetic Tunnel Junction (VCMA-MTJ) devices to provide fast SR and adaptive QR in a novel energy-efficient fashion. Expected contributions include: 1) A novel framework for efficient and intelligent sensing through integration of resource allocation, quantized CS, and adaptable spin-based devices will be developed; 2) SR and QR trade-offs under resource constraints are utilized to attain an energy-aware adaptive SR/QR optimization framework which is integrated with VCMA-MTJ devices; 3) Novel sampling and reconstruction algorithms will be developed in context of adaptive quantized CS; 4) VCMA-MTJ circuits will be designed to realize faster and more energy-efficient sampling and signal processing; 5) Spin-based lookup table and encoder circuits using new switching strategies will be designed; 6) The energy consumption of VCMA-MTJs will be analyzed and the derived energy equation will be utilized for SR/QR optimization.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

跨层自适应速率/分辨率采样利用基于自旋的设备用于传感和通信系统该项目将设计一个自适应框架，用于频谱稀疏信号的有效数据采集，以提高采样精度，同时降低未来传感和通信系统的能耗需求，这些系统对我们现代和互联社会至关重要。新的自适应采样和重建技术将被设计伴随着一个基于自旋的硬件方法，以尽量减少数据采集和传输的整体成本，适用于广泛的通信系统和一个大的和重要的一类频谱稀疏信号，出现在许多应用中，如认知无线电网络，雷达，和新兴的频谱感知通信系统。因此，这一项目将有利于国家利益，促进通信系统的关键技术，提高能源效率，增加电路密度。将编制和分发本科生和研究生课程单元的教材，以及一个互动网站，以吸引高中生，包括代表性不足的各类学生和女学生，参与和从事这一领域的学习和职业。通过纳米器件库网页的广泛传播将被用来增加影响，同时补充通过高质量的学术期刊和会议，和网站的研究成果的出版。多学科的努力将被用来开发一个系统的方法，桥梁之间的差距差距先进的理论研究在信号处理/压缩传感和创新的电路设计，利用信号处理，存储器，以及新兴的基于自旋的器件中固有的阈值能力。第一个研究重点是在量化压缩感知（CS）的背景下，在功率和带宽约束下，使用动态优化的SR和QR的在线方式调查采样率（SR）和量化分辨率（QR）之间的权衡。 能量消耗、硬件限制以及底层采样器和量化器的细节将被优化。计算高效的信号重建算法进行了研究，以重建原始信号从其非均匀的（在采样率和量化深度方面）量化CS测量。在第二个研究重点中，研究人员将研究和设计一种基于间歇自旋的自适应量化器，该量化器利用电压控制磁各向异性磁隧道结（VCMA-MTJ）器件以一种新的节能方式提供快速SR和自适应QR。预期的贡献包括：1）通过资源分配、量化CS和可适应的基于自旋的器件的集成，将开发用于高效和智能感测的新框架; 2）利用资源约束下的SR和QR权衡来获得与VCMA-MTJ器件集成的能量感知的自适应SR/QR优化框架; 3）将在自适应量化CS的上下文中开发新的采样和重构算法; 4）VCMA-MTJ电路将被设计为实现更快和更节能的采样和信号处理; 5）基于自旋的查找表和编码器电路将被设计为使用新的开关策略; 6）将分析VCMA-MTJ的能量消耗，并将导出的能量方程用于SR/QR优化。该奖项反映了NSF的法定使命，并被认为是值得通过使用基金会的知识价值和更广泛的影响审查标准进行评估的支持。

项目成果

RL-NCS: Reinforcement Learning Based Data-driven Approach for Nonuniform Compressed Sensing

RL-NCS：基于强化学习的数据驱动的非均匀压缩感知方法

• DOI: 10.1109/mlsp.2019.8918768
• 发表时间: 2019
• 期刊: IEEE 29th International Workshop on Machine learning for signal processing (MLSP
• 作者: Karim, N;Zaeemzadeh, A;Rahnavard, N.
• 通讯作者: Rahnavard, N.

Two-Way Spectrum Pursuit for CUR Decomposition and its Application in Joint Column/Row Subset Selection

CUR分解的双向谱追踪及其在联合列/行子集选择中的应用

• DOI: 10.1109/mlsp52302.2021.9596233
• 发表时间: 2021
• 期刊: 2021 IEEE 31st International Workshop on Machine Learning for Signal Processing (MLSP
• 作者: Esmaeili, Ashkan;Joneidi, Mohsen;Salimitari, Mehrdad;Khalid, Umar;Rahnavard, Nazanin
• 通讯作者: Rahnavard, Nazanin

Out-of-Distribution Detection Using Union of 1-Dimensional Subspaces

• DOI: 10.1109/cvpr46437.2021.00933
• 发表时间: 2021-06
• 期刊: 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
• 作者: Alireza Zaeemzadeh;N. Bisagno;Zeno Sambugaro;N. Conci;Nazanin Rahnavard;M. Shah

Mixed-Signal Spin/Charge Reconfigurable Array for Energy-Aware Compressive Signal Processing

• DOI: 10.1109/reconfig48160.2019.8994799
• 发表时间: 2019-12
• 期刊: 2019 International Conference on ReConFigurable Computing and FPGAs (ReConFig)
• 作者: Adrian Tatulian;Soheil Salehi;R. Demara

A Spin-based Analog to Digital Converter Interactive Simulation Framework

• 发表时间: 2019
• 作者: Gustavo Camero;Soheil Salehi;R. Demara