Abstract

Neuromorphic spiking sensors are inspired by the functionality of their biological counterparts (e.g. retina to inspire vision sensors, cochlea to inspire auditory sensors, mechanoreceptor to inspire touch sensors, etc.) and provide change detection mechanisms at the sensor to directly produce sparse outputs. The asynchronous outputs of these event-driven sensors can enable always-on sensing at lower latencies and data rates compared to conventional sampled sensors for Internet of Things (IoT) and Brain-Machine Interface (BMI) applications. Recent developments in deep networks, spiking networks, and in-memory computing have led to very low-power neuromorphic systems that combine these sensors and networks for several edge computing systems. In-memory computing can help to reduce power from off-chip memory access, the main bottleneck for implementing neural networks, and can additionally, support different learning algorithms useful for adaptive neuromorphic systems. Also, these methods reduce the energy required to implement matrix multiplies, nonlinearities, and other signal-processing operations over digital systems.

This tutorial will describe the advances in the design of neuromorphic sensors, bio-inspired network architectures and algorithms, and hardware implementations that can also be applied to the spiking sensor output. We will present examples of the use of neuromorphic systems in low-latency low-power application domains of ubiquitous visual sensing and brain-machine interfaces.

Abstract

Our goal is to realize an ion imaging system that can simultaneously visualize the two-dimensional distribution of chemical substances in real time. We believe that this technology will dramatically deepen our understanding of conventional medical and physiological fields. In the tutorial presentation, I will describe the fabrication and design of a chemical image sensor developed by our group.

The role of bioactive substances in the extracellular microenvironment is important for understanding the function of brain neural networks. However, there has been no effective method to visualize the interaction of bioactive substances in the extracellular microenvironment.　

I have realized a chemical image sensor using CMOS image sensor technology to visualize this extracellular microenvironment. CMOS image sensor technologies are evolving dramatically. Among the sensors that are in use, the CMOS image sensor is manufactured with the most refined semiconductor-integrated-circuit technology. In recent years, the number of pixels has exceeded 20 million, sometimes even reached 100 million. Meanwhile, the pixel size has been reduced to 1 µm × 1 µm or below. Considering that human cells are approximately 20 µm and synapses are 1–2 µm in size, it is now possible to manufacture bioimaging sensors that directly sense multiple ions released from cells and neurotransmitters by integrating CMOS image sensor and biosensor technologies.

I will then show examples of the development of sensors that can be implanted in the brain and real-time monitoring of pH changes in the brains of mice in their natural environment. I will also present examples of imaging of extracellular lactic acid and potassium ions based on this sensor and discuss the importance of measuring the extracellular environment.

Abstract

We are living in the age of human-machine augmentation and coexistence (e.g., smartphones, AI assistants, computers, earbuds, smart watches) and steadily marching towards a new age of human-machine seamless cooperation (HMC) or even symbiosis. Radiative communication using electromagnetic (EM) fields is the state-of-the-art for connecting wearable and implantable devices enabling prime applications in the fields of connected healthcare, electroceuticals, neuroscience, augmented and virtual reality (AR/VR) and human-computer interaction (HCI) and HMC, forming a subset of the Internet of Things called the Internet of Bodies (IoB). However, owing to such radiative nature of the traditional wireless communication, EM signals propagate in all directions, inadvertently allowing an eavesdropper to intercept the information. Moreover, since only a fraction of the energy is picked up by the intended device, and the need for high carrier frequency compared to information content, wireless communication tends to suffer from poor energy-efficiency (>nJ/bit). Noting that all IoB devices share a common medium, i.e. the human body, utilizing the conductivity of the human the body allows low-loss transmission, termed as human body communication (HBC) and improves energy-efficiency. Conventional HBC implementations still suffer from significant radiation compromising physical security and efficiency. Our recent work has developed Electro-Quasistatic Human Body Communication (EQS-HBC), a method for localizing signals within the body using low-frequency transmission, thereby making it extremely difficult for a nearby eavesdropper to intercept critical private data, thus producing a covert communication channel, i.e., the human body as a ‘wire’ along with reducing interference and providing 100x more efficient communication than Bluetooth.

In this talk, I will highlight recent advancements in the field of IoB enabled by the Body-as-a-Wire technology which has a strong promise to become the future of Body Area Network (BAN) along with it’s counterpart in the Brain leading to broadband communication. We will focus on the circuit model developed in recent literature explaining the fundamental behavior of Electro-Quasistatic Body and Brain Communication, leading to understanding of channel loss. We will finally show how such low-power communication is paving the way forward for Secure and Efficient IoB for seamless Human-Machine Co-operation.

Abstract

Biopotential signals acquisition plays a critical role in a human-computer interface system. Most of these biopotential signals distributed in the band of lower than kilo-hertz, or even lower than hundreds hertz. Precise acquisition of low amplitude signal at near DC frequency band is a challenge. In addition, the wearable/implantable scenario rises a high requirement in power efficiency, which causes a very limited power budget for the typically power hungry wireless data transmission. The existed commercial solutions, such as BLE and/or WiFi, are not suitable. This seminar first introduced the origins of various common vital signals, such as brain signals, heart signals, and etc. An acquisition with high precision and μV level low input- referred noise (IRN) is required from the scenario. In addition, the performance of low-power consumption is also expected to extend the battery life, as well as to reduce self-heating of the implanted device for safety. The chopper-stabilized amplifier is promising due to its good flicker noise and power consumption performance. More detailed design strategy will be presented in the seminar. The data transmission workload increases while the acquisition channel increases to several thousands. A wireless data link with Gb/s throughput is strongly required. In addition to the data-rate, the wireless telemetry module faces a strictly limited power budget of only a few milliwatts, to avoid tissue heating. Thirdly, a compact implant design with as few as possible off-chip components is expected to minimize the harm caused by the implant. mW power consumption transmitter solution with Gbps througoutput will be presented in the seminar as well, realizing a high energy efficiency SoC solution for wireless vital signal acquisition. Frontier applications will be shared by the end of the talk.