CAREER:Integrated Research and Education on Delta-Sigma Based Digital Signal Processing Circuits for Low-Power Intelligent Sensors

近年来，低功耗集成智能传感器的应用已经非常丰富，包括例如环境观测、安全监视、基础设施监测和通信以及生物医学健康护理监测。尽管这种传感器通常具有无线数据通信能力，但是由于有限的电池功率，发送原始感测数据通常不是一种选择，因为在无线传感器中，无线电通信功率通常比信号处理功率高得多。因此，智能传感器需要能够基于信号处理算法提供原始数据的预处理，并仅发送处理后的结果。生物医学应用中另一个不可避免的挑战是过采样。这是因为目标生物医学信号通常在频谱中具有宽的分数带宽，并且需要时频分析。因此，在这些应用中，模数转换器（ADC）必须应用过采样，以避免由于快速滚降和平坦群延迟之间的权衡而由抗混叠滤波器引入的信号失真。过采样和更高分辨率的要求也加剧了功耗问题。上述挑战的一个有前途的解决方案是Delta-Sigma技术。本CAREER提案的研究目标是在数字信号处理（DSP）中应用基于Delta-Sigma的集成电路（IC）设计，以解决下一代低功耗智能传感器的电路功耗和面积问题。这项研究将对下一代普适计算和普适传感应用产生广泛的影响。该职业建议的教育目标是促进积极的，基于探究的学习，向学生介绍跨学科的学习和研究，并通过为他们提供一套完整的技能，包括系统设计，组装，验证和优化，为他们做好准备，以满足学术界和工业界的未来期望。提出的集成线性信号处理电路基于Delta-Sigma重新设计，使用数字Delta-Sigma调制器进行调制。三角积分线性处理电路，如加法器，系数乘法器和滤波器将设计，制造，表征，并与传统的信号处理系统进行比较。特别是，基于初步的结果，该研究将针对生物医学信号处理器应用交叉频率耦合（CFC）算法，其中包括FIR和IIR带通滤波器，希尔伯特滤波器，和坐标旋转数字计算机（CORDIC）电路的设计。建议的系统与可重构的分辨率将进行调查，以评估引入自适应分辨率传感器，以节省传感和通信功率的好处。该研究通过采用Delta-Sigma调制（DSM），数字信号处理和集成电路设计的思想，以跨学科的方式显着推进了集成智能传感器的最新技术。 具体而言，拟议的研究将通过解决以下目标来实现高功率效率的电路和系统架构：1）研究基于Δ-Σ的加法器和系数乘法器的设计和特性，以便从理论上理解它们如何应用于各种电路结构，2）研究基于Delta-Sigma的数字滤波器和神经网络系统，并探索将所提出的电路应用于机器学习的可行性，3）利用Delta-Sigma数字信号处理电路设计了一种用于脑电信号处理的交叉频率耦合算法集成电路，并与传统结构进行了性能比较。

Applications of low-power integrated intelligent sensors have been prolific in recent years, including, for instance, in environmental observation, security surveillance, infrastructure monitoring and communication, and biomedical health care monitoring. Although such sensors usually have wireless data communication capability, transmitting raw sensed data is usually not an option because of limited battery power, since in a wireless sensor the radio communication power is usually much higher than the signal processing power. Therefore, intelligent sensors need to be capable of providing preprocessing of raw data based on signal processing algorithms and sending only the processed results. Another inevitable challenge in biomedical applications is oversampling. This is because the targeted biomedical signals usually have a wide fractional bandwidth in the frequency spectrum, and require time-frequency analysis. Therefore, in these applications, the Analog to Digital Convertor (ADC) have to apply oversampling in order to avoid the signal distortion introduced by anti-aliasing filters due to the trade-off between fast roll-off and flat group delay. The requirements of both oversampling and the higher resolution also exacerbate the power problem. One promising solution to the above challenges is Delta-Sigma technology. The research objective of this CAREER proposal is to apply Delta-Sigma based Integrated Circuit (IC) design in Digital Signal Processing (DSP) to solve circuit power and area problems for the next-generation of low-power intelligent sensors. The proposed research will have a broad impact on next-generation pervasive computing and ubiquitous sensing applications. The educational objectives of this CAREER proposal are promoting active, inquiry-based learning, introducing students to interdisciplinary study and research, and preparing them to meet the future expectations of both academia and industry by providing them a complete skill set including system design, assembly, verification, and optimization, and making cutting-edge research projects more accessible to minority and underrepresented groups of students.The proposed integrated linear signal processing circuits are based on Delta-Sigma re-modulation using digital Delta-Sigma modulators. Delta Sigma linear processing circuits such as adders, coefficient multipliers, and filters will be designed, fabricated, characterized, and compared to conventional signal processing systems. In particular, based on the preliminary results, the research will target a biomedical signal processor applying the Cross-Frequency-Coupling (CFC) algorithm, which includes designing of FIR and IIR Band-pass filters, Hilbert filters, and Coordinate Rotation Digital Computer (CORDIC) circuits. The proposed system with reconfigurable resolution will be investigated in order to evaluate the benefit of introducing adaptive resolution sensors to save sensing and communication power. The proposed research significantly advances the state-of-the-art in integrated intelligent sensors by employing ideas from Delta-Sigma Modulation (DSM), Digital Signal Processing, and Integrated Circuits design in a cross-disciplinary fashion. Specifically, the proposed research will lead to a highly power-efficient circuit and system architecture by addressing the following objectives: 1) studying design and properties of Delta-Sigma based adders and coefficient multipliers for a theoretical understanding of how they can be applied to various circuit architectures, 2) investigating Delta-Sigma based digital filters and neural network systems and exploring the feasibility of applying the proposed circuits in machine learning, and 3) developing an integrated circuit of Cross-Frequency-Coupling algorithm for electroencephalograph (EEG) signal processing using Delta-Sigma digital signal processing circuits, and evaluating the system performance compared with conventional architecture.