FeEdge - Innovative compute-in-memory modules for energy-efficient edge AI

More and more data is being transmitted worldwide. In order to deal with this more efficiently, more and more research is being carried out on edge nodes. Here, data is processed directly at the point of origin. For example, devices such as wearables, sensors, smartphones and cars can analyze data locally and make autonomous decisions using AI.

In addition to lower energy requirements, edge AI devices therefore also have great potential to enable new applications with higher performance and support local embedded intelligence, real-time learning and autonomy.

The main goal of this project is to design and implement a state-of-the-art, low-power ferroelectric field-effect transistor (FeFET)-based compute-in-memory (CIM) accelerator for artificial intelligence (AI). In this project, we aim to realize the first large-scale accelerator based on CIM functionality with FeFET technology.

The project is a cooperation with Cheng Kung National University (NKCU) in Taiwan. In addition to the technological collaboration, exchange visits for students and researchers are also planned. This planned exchange is intended to maximize the synergies between the expertise of NKCU and Fraunhofer IPMS and offer all participants valuable insights and skills enhancement. This promotes a global perspective and enriches the project results through different perspectives and approaches.