Tutorials

Nature inspired computation using neural networks has given rise to new problem-solving approaches using examples instead of formal reasoning. In particular, the deep learning paradigm can lead to efficient solutions of complex classification and prediction problems when large numbers of training examples are available. On the other hand, there exist also design problems, which are typically undetermined and for which the training data are limited. The tutorial explores these challenges, both in software and hardware, when applying AI to wearable devices. Using the example of a surface electromyography-driven hand prosthesis control system, the sensing, signal processing, and machine learning pipelines of a full wearable system are detailed.

Memory architecture and design have been critical for digital systems to achieve ample storage, low latency, fast access time, and energy efficiency, especially for battery-operated devices. The increase of data generated by many devices such as mobile phones, sensors, communications, and security increased the requirements for memory capacity and the challenges of memory access and energy. The memory interface has limited throughput and high latency, which has not been scaling at the same rate as data size or processing speed; this limits the performance of accessing the data, which is referred to as the memory wall. In addition to the negative impact on latency and performance, large data movement results in high energy consumption. Research has focused on elevating the memory wall issue by engineering more memory hierarchy and increasing local on-chip memory. This has partially reduced the timing issue but did not address the high leakage and active energy consumption. It is estimated that over 60% of energy spent on most computing platforms is spent on data movements and memory access.

The new era of big data and artificial intelligence-based applications has increased the urgency to solve memory capacity, data movement energy, and memory wall issues. Some solutions have brought processing into centralized cloud computing, with high performance and large memory hardware capacity available. However, this brought a new challenge to communications, privacy, security, and latency, especially for real-time applications. This tutorial highlights the challenges above and presents a new computing paradigm beyond von Neuman's architecture to enable processing as close to the data source as possible. This includes in-memory computing and near-memory computing architecture. Both existing and emerging memory technologies will be explored. Since the new computing paradigm is more data-centric than processing-centric, the traditional single architecture for all applications is not feasible. Hence, domain-specific architecture and hardware solutions need to be adopted. Popular high computing functions such as Query, MAC, hamming distance, and image compression will be presented as examples of in-memory hardware accelerators.

Implantable sensing is a promising solution to provide continuous monitoring of human body with a bunch of merits, such as direct measurement of vital signals, improved system robustness, anti-interference capability, etc. The highly-integrated and energy-efficient application specific integrated circuits (ASICs) are the key building components to build the miniature implantable sensors. Such ASICs will provide the functions of signal acquisition, processing and transmission. In this tutorial, we will first review the signal acquisition/processing/transmission requirements in such applications, foll0wed by the major considerations of these ASICs. We will then take two practical applications as the examples, namely, the implantable electrocardiogram (ECG) sensor and the intracranial pressure (ICP) sensor, to exhibit the state-of-the-arts ASIC techniques for ultra-low power bio-signal acquisition, near sensor processing, and short-range through-body data transmission. The design principles of energy-efficient ASICs will be illustrated through these two design examples. The technique trends in this specific area will also be briefly discussed in this tutorial.

Emerging Electromagnetic Acoustic technique is to combine the merits of traditional electromagnetic sensing technique (e.g. Radar and Lidar) with acoustic imaging (e.g. microphone and ultrasound), and go beyond. In this seminar, we would introduce three topics: (1) Low power phase arrayed Radar chip for SAR imaging, (2) Photoacoustics sensing and imaging system, (3) Thermoacoustics sensing system. On the phase array radar transceiver, the design challenge is to generate and transmit multiphased wideband chirp signal and to do stretch processing based high resolution beamforming. The detailed circuits and system to implement the radar sensor will be presented. Photoacoutics sensor transmit focused laser light deep penetrated to the tissue/blood vessel, inducing high frequency ultrasound signal, and then high resolution acoustic imaging can be formed. To miniaturize the whole sensor and achieve high sensitivity, the fibre coupled pulsed laser, beamforming ultrasound transducer, and low power low noise signal acquisition circuits are designed and implemented. Furthermore, there appears increasing interests to use microwave induced thermoacoustic and/or mageneto-acosutic signal for NDT applications. We will brief introduced some coil based EM transmitter for wireless power transfer, and EMAT based non-contact sensing.

Introduction

Electrochemical Impedance Spectroscopy (EIS) is one of the most important techniques employed in electrochemical analysis. It finds applications in a wide range of fields such as corrosion detection, biomedical sensors, battery and fuel cell development, surface characterization and physical electrochemistry, to name a few. The mean reason for its widespread adoption is that fact that it provides more information content than any other electrochemical techniques. Fully integrated on-chip EIS systems have contributed to the popularity of this technique and opened the door for new use cases. Portable and fully implantable biomedical devices for biomarker monitoring, smart Battery Management Systems (BMS) with Cell Monitoring Circuits (CMCs), distributed gas sensors and sensor array microsystems are some of the applications scenarios that rely on on-chip EIS.

Why this topic

The design and implementation of on-chip EIS presents interesting challenges from a circuit design perspective since it requires analog, mixed-signal and digital design techniques. This can be seen when reviewing the two techniques for on-chip EIS, which fall under two main categories, namely:

• Fast Fourier Transform (FFT) techniques: here, a broadband signal such as a Dirac pulse or a multi frequency wavelet is used to stimulate the System Under Test (SUT). The amplitude of the input must be small enough so as to induce a linear response from the system. The time domain response is measured, digitized and spectrally analyzed using an FFT algorithm.

• Frequency Response Analysis (FRA) techniques: here, the SUT is excited by a pure single frequency sinusoidal signal, where the amplitude is also kept low enough to produce a linear response. The frequency of the input sine signal is swept over the range of interest and the impedance is measured one frequency at a time.

Another aspect which makes the design of on-chip EIS interesting is the higher order of complexity of such systems. Typical integrated EIS blocks require sinusoidal signal generators, active filters, high precision amplifiers, analog multipliers, ADCs, digital control and post processing and other components. When considering that all of these circuits need to be tunable in order to operate on frequencies ranging from the milli-Hertz to several mega-Hertz, depending on the application, it becomes evident how carful system level planning is needed to fulfill the many conflicting requirements of the design space.

Aim

The aim of this tutorial is for the audience to gain a comprehensive overview of the topic of on-chip EIS with a specific focus on circuit design. In the first part relevant background theory and fundamentals are presented starting with a review on the concept of electrical impedance, the response of electrical systems to dynamic stimulus and the frequency analysis techniques. This is followed by a discussion of the foundations of electrochemical analysis where the importance of interfaces and the modeling of chemical processes as electrical analogs are presented, followed by presentation of the role of EIS and equivalent circuit models.The second part is dedicated to the design of fully integrated on-chip EIS systems, presenting FFT, FRA and other techniques. Finally biomedical applications and BMS systems are presented including design considerations and implementation of EIS for such systems.

Keywords:

electrochemical impedance spectroscopy, on-chip EIS, fully integrated signal generators, integrated electrochemical analysis. electronics, flexible electronics

Recently, the principle of intelligent sensors has become increasingly popular in the automotive market. Car sensors play a crucial role in ensuring that a vehicle operates safely and efficiently. The time now to take advantage of bus communication, digital signal processing and fault diagnosis. ASIC permits the level of integration needed to control the production costs while allowing libraries of component to be assembled cheaply and easily on a workstation to provide the required flexibility.

Sensors are now an essential part of any modern automotive design and serve many different purposes. They play a key role in helping car manufacturers to bring models to the market that are safer, more economical and more comfortable-to-drive. Over time, sensors will also enable greater degrees of vehicle automation, which is benefit to the industry. The next-generation sensors being developed will ultimately determine the desired autonomous driving experience. Through innovating in the areas mentioned in this insight, the cars of tomorrow will provide a clear and constantly updated picture of what is happening, both in relation to the external environment and in relation to what their occupants are doing. Therefore, sensing technologies hold key to the future of the automotive industry.

This tutorial will throw light on different types of sensors in a car, which monitor different aspects of a vehicle and send information to the ECU. With the focusing on their functions and their implementation in the system.

Keywords:

Smart sensors in cars, functions of automotive car sensor, sensors in a complete system, communication between sensors and ECU, and functional safety.