SHF: Medium: Power-Adaptive, Event-Driven Data Conversion and Signal Processing Using Asynchronous Digital Techniques

这项研究解决了对极低功耗和高质量微电子系统的日益增长的需求，这些系统可以在信息可用时立即连续获取和处理信息。这种数据采集和实时信号处理在广泛的应用中使用，例如环境传感器网络(由太阳能电池供电或通过使用振动能量的能量收集)，以及用于假肢或连续诊断和监测的可植入或可摄取的生物医学设备。在这些应用中，新信息经常不频繁地以不规则和不可预测的间隔生成。这种基于事件的性质要求对如何监控和处理这些信号进行彻底的重新思考。传统的同步(即计时)数字技术使用固定速率运算来评估数据，而不管数据是否已经改变，这种技术不太匹配，通常导致高功耗。相反，这项研究的目的是提供可行的“基于事件”的系统：不是由时钟控制，而是由每个事件的到来控制。这种连续时间的数据采集承诺显著降低功率和能量，灵活支持各种信号处理协议和编码，高质量的输出信号，以及对未来微电子技术的优雅扩展。非常适合于这项工作的异步(即无时钟)数字逻辑技术与连续时间数字信号处理相结合，使这项任务成为可能；这两个领域都已由主要研究人员在先前NSF的支持下进行了研究。一系列硅芯片将被设计和充分评估，最终形成一个完全可编程的、事件驱动的数据采集和信号处理系统，该系统可用作各种现实世界应用的试验台。这项工作预计将产生广泛的影响。由此产生的芯片和方法将在能源稀缺的广泛应用中提供显著的实际好处，例如在生物医学电子设备、传感器网络和便携式通信设备中，这些设备必须使用小电池长时间运行。芯片开发将通过与生物医学工程领域的专家同事密切磋商，并通过研究人员？与美国微电子公司保持密切联系。此外，这项研究有望在超低功率微电子系统方面取得根本性的新进展，这将对电力电子和用于机械振动控制的新型控制系统等不同领域产生影响。最后，将向研究界广泛传播这项工作的结果，以促进和推动这些技术的实际应用，并促进这一领域的进一步研究和教育，包括将其纳入哥伦比亚大学和其他一流大学的课程。

This research addresses an increasing demand and need for very low power and high-quality microelectronic systems that can continuously acquire and process information, as soon as it becomes available. Such data acquisition and real-time signal processing are used in a wide range of applications, such as environmental sensor networks (powered by solar cells or by energy harvesting using vibrational energy), and implantable or ingestible biomedical devices for prosthetics or for continuous diagnostics and monitoring. In these applications, new information is often generated infrequently, at irregular and unpredictable intervals. This event-based nature calls for a drastic re-thinking of how these signals are monitored and processed. Conventional synchronous (i.e. clocked) digital techniques, which use fixed-rate operation to evaluate data whether or not it has changed, are a poor match, often leading to high power consumption. This research aims instead to provide viable "event-based" systems: controlled not by a clock but rather by the arrival of each event. Such continuous-time data acquisition promises significant power and energy reduction, flexible support for a variety of signal processing protocols and encodings, high-quality output signals, and graceful scalability to future microelectronic technologies. Asynchronous (i.e. clockless) digital logic techniques, which are ideally suited for this work, are combined with continuous-time digital signal processing, to make this task possible; both of these areas have been researched by the principal investigators under prior NSF support. A series of silicon chips will be designed and fully evaluated, culminating in a fully programmable, event-driven data acquisition and signal processing system, which can be used as a testbed for a wide variety of real-world applications.This work is expected to have broad impact. The resulting chips and methodology will provide significant practical benefits in widely-used applications where energy resources are scarce, such as in biomedical electronics, sensor networks and portable communications devices, which must operate on a small battery for a long lifetime. The chip development will be directed towards these applications through close consultation with expert colleagues in biomedical engineering, and through the investigators? existing close ties with US microelectronics companies. In addition, the research is expected to provide fundamental new advances in ultra-low-power microelectronics systems, which can have an impact on such diverse areas as power electronics and new types of control systems for mechanical vibration control. Finally, the results of this work will be extensively disseminated to the research community, to promote and advance the practical use of these techniques, as well as to promote further research and education in this area, including through its incorporation into curricula at Columbia University and other leading universities.