Our aim is to research and develop circuit technology that executes AI processing with high efficiency and low power consumption, cryo-CMOS circuit technology for quantum computers, analog and digital integrated circuit technology for physical and chemical sensors, and to explore an LSI design and development environment using open source EDA.
Analog mixed-signal IC design for MEMS sensor devices
Research and development of edge AI chips that can perform ultra-highly efficient AI processing
Exploring an LSI development environment using OpenSource EDA
Analog mixed-signal integrated circuits for MEMS sensors include various essential components such as low-noise amplifiers, AD converters, and digital circuits. In this study, after modeling sensor device as an equivalent circuit, its appropriate circuit architecture will be created to satisfy given specifications and/or target cost. We also adopt digitally-assisted techniques to enhance its performance and functionalities. This sensor circuit co-optimization scheme can be applicable to other devices such as actuators and energy harvesting as well.
AI (deep-learning) tasks are usually processed in data centers because of the huge amount of computation. However, it is expected that AI tasks are processed at user side, due to large power and latency in communication between users and data centers and security concerns. In user side (edge) devices, the power allowed for AI tasks is often limited. Therefore, circuits dedicated to process AI tasks efficiently, called “edge AI chips,” are required. We are conducting research and development of ultra-efficient AI chips.
OpenSource EDA has become a global trend, and cases of its use in Japan are gradually increasing. In this research, we are accumulating knowledge related to the development and use of a usage environment for the widespread use of OpenSource EDA/PDK in Japan.
