Continuum Computing Architecture Research Group

The Continuum Computing Architecture Research Group is advancing research on foundational technologies that support the high performance and advanced functionality of applications within ultra-distributed computing cores, spanning from the cloud to the edge and end devices. The group also focuses on cloud technologies that collaborate with edge systems to enable large-scale AI computation. While high throughput is required in the cloud and low latency is essential at the edge, the group addresses challenges such as real-time processing, energy efficiency, and performance optimization through a co-design approach, integrating both hardware and software development. In addition, the group actively collaborates within and outside of AIST to drive the development of next-generation technologies based on emerging semiconductor and photonics technologies.

News

Tags : All

Date order : New / Old

There is no notice at the moment

Research Topics

✪ Scalable Cloud Technologies for Intensive AI Workloads and Big Data Processing

Data management and use in continuum computing
- Offloading and data reduction at edge for scalability and energy saving, Data pipeline optimization
- DataOps automation, balanced with effective human interaction
- Data traceability in sharing and integration
Management of continuum computing application services
- Develop and build methods of high-performance and robust continuum computing applications for geo-distributed, unstable and heterogeneous environments
- Zero-touch service deployment, Dynamic QoS control, Offloading decision, Client-mobility support
- Low-latency service-to-service interaction, Autonomous service orchestration
High performance cloud technologies tightly connecting with edge
- Large-scale computing with accelerators
- High performance AI, AI resource hub
- Low-latency and scalable connection services with edge, Efficient resource management

✪ Hardware and Software co-design

While the Cloud requires throughput-oriented technologies, the Edge requires latency-oriented technologies. We tackle research issues on throughput improvement, real-time assurance and energy saving, through an interdisciplinary approach to hardware and software co-design.
For hardware and software co-design, our research is based on emerging semiconductor technologies and photonics technologies, in collaboration with the Department of Electronics and Manufacturing at AIST, and different universities and research institutes.

High-throughput virtualization: Enabling high-performance processing by exploiting emerging semiconductor technologies and photonics technologies.
Low-power, low-latency virtualization: Achieving energy-efficient, real-time processing toward 5G/6G communication.
Error-permissive computing: Realizing drastic performance improvement and power reduction by controlling reliability in computer hardware.