Real-time signal processing systems for advanced and emerging technologies

我的研究是基于探索采用先进和新兴电子技术构建的处理器的计算特性，以及它们在信息、通信技术和生命科学中的应用。我们使用的计算机中包含的微处理器，以及嵌入我们的手机、汽车、家用电器、娱乐系统、医疗设备等中的微处理器，都作为其设计中的基本功能来执行计算。对于必须连续处理数据流的实时系统，“嵌入式”微处理器每秒通常需要执行数十亿次计算。这些计算反映了我们可能用纸和铅笔进行的各种计算。在现代集成电路技术中，制造所需的设计程序是如此复杂，以至于许多设计者将简单地使用设计工具中提供的可用功能。然而，为了以最有效的方式使用这些技术，有必要研究执行基本算术运算的替代方法，甚至是寻找表示数据的替代方法。随着集成电路技术成倍地提高晶体管的堆积密度(被称为摩尔定律)，这些基本的计算模块的速度大大提高；然而，过去被忽视的问题特征现在变得至关重要，因此我们需要不断寻找这些新问题的解决方案。一个例子是现代高密度集成电路的过度功耗，这是限制我们台式机和笔记本电脑中使用的微处理器的最大时钟频率的原因。在我的研究中，我们探索在集成电路上构建处理器的替代方法，并在选定的应用领域检查这些替代方案。我最近的工作可以在生命科学中找到应用，例如：助听器处理器和植入式葡萄糖传感器，在这些领域，极低的功耗是一个主要限制因素，以及在非常高带宽的无线基站中，需要非常可靠的计算，涉及每秒万亿次基本算术运算。

My research is based on exploring computational properties of processors built with advanced and emerging electronic technologies, and their application to information, communication technology, and the life sciences. The microprocessors that are contained in the computers we use, and that are embedded in our cell phones, automobiles, appliances, entertainment systems, medical devices etc., all perform computations as fundamental functions in their design. For real-time systems, where streams of data have to be continuously processed, "embedded" microprocessors often need to perform many billions of computations every second. The computations mirror the sorts of calculations we might do longhand with paper and a pencil. The design procedures required for fabrication in modern integrated circuit technologies are so complex, that many designers will simply use the available functions that are provided within the design tools. However, in order to use the technologies in the most efficient way possible, it is necessary to examine alternative ways to perform basic arithmetic operations, even down to looking at alternative ways to represent the data. As integrated circuit technology has exponentially improved the packing density of transistors (known as "Moore's Law"), these basic computational blocks have dramatically increased in speed; however problem features, that used to be neglected, have now become critical and so we need to continually search for solutions to these new problems. One example is the excessive power dissipation of modern high density integrated circuits, that has been responsible for limiting the maximum clock frequency of the microprocessors used in our desktop and laptop computers. In my research we explore alternative ways of building processors on integrated circuits and examine these alternatives in selected application areas. Recent applications of my work can be found in the life sciences, for example: hearing-aid processors and implantable glucose sensors, where extremely low power dissipation is a major constraint, and in very high bandwidth wireless base stations, where very reliable computation involving a trillion basic arithmetic operations per second is required.