Digital Cybernetics = Post-Shannon Communication + LLM-based AI Agents
JST CRONOS Project JPMJCS25N5: Digital Cybernetics: Towards Next-Generation Communication Architecture
CRONOS - Challenge for Revolutionary Research Network and Open Science
- 2025/09/11 - Our project "Digital Cybernetics: Towards Next-Generation Communication Architecture" has been selected for the CRONOS program by JST (Japan Science and Technology Agency).
- 📖 Project Overview - Goals, period, and principal investigators
- 🔬 Research Themes - Five foundational technologies
- 🎯 Our Goal and Methodology - Post-Shannon framework and biological analogies
- 👥 Research Groups - Wadayama and Watanabe groups
- 📞 Contact - Get in touch with the team
AI agents are rapidly transforming communication technology. This research realizes "Post-Shannon Communication" supporting autonomous, adaptive, and robust large-scale AI systems. We develop foundational technologies—goal-oriented communication, ultra-large-scale coding, digital homeostasis, physics-aware signal processing, and dual-process learning—to create next-generation communication architecture.
📄 Project Summary Slide (PDF) | 📖 Detailed Project Concept
October 2025 - March 2031 (5.5 years)
Tadashi Wadayama Principal Investigator Nagoya Institute of Technology [Homepage] |
Shun Watanabe Co-Principal Investigator Tokyo University of Agriculture and Technology |
Design communication methods specifically tailored to concrete goals and tasks. In digital cybernetics context, we focus on achieving task objectives while maintaining system autonomy.
Develop coding schemes for systems with massive numbers of sensors, actuators, and effectors, focusing on low-latency and ultra-large-scale connectivity.
Enable real-time autonomous optimization of communication parameters (data rate, coding schemes) and available resources (frequency bands, power consumption) to maintain stable operation and rapid self-recovery from failures.
Incorporate governing partial differential equations (PDEs) of physical media into signal processing:
- Maxwell's equations for wireless communication
- Nonlinear Schrödinger equation for optical fiber communication
Inspired by cognitive psychology's dual-process theory:
- System 1: Fast, reactive signal processing using online learning
- System 2: Deliberative meta-learning using LLM orchestrators
Utilizing the analogy with biological nervous systems as a conceptual framework for inspiration
Focus: Digital Cybernetics and Communication
- Physics-aware signal processing implementation
- Dual-process learning system development
- Digital homeostasis mechanism design
- System integration
PI: Tadashi Wadayama (Nagoya Institute of Technology)
Members:
- Ayano Nakai-Kasai (Nagoya Institute of Technology)
- Kazunori Hayashi (Kyoto University)
- Masaki Ogura (Hiroshima University)
- Satoshi Takabe (Institute of Science Tokyo)
- Takanori Hara (Tokyo University of Science)
Advisor: Takahiro Uchiya (Nagoya Institute of Technology)
Focus: Post-Shannon Information Theory
- Distributed hypothesis testing
- Distributed coding for ultra-large-scale systems
- Information-theoretic security
- Identification codes
PI: Shun Watanabe (Tokyo University of Agriculture and Technology)
Members:
- Shigeaki Kuzuoka (Wakayama University)
- Akira Kamatsuka (Shonan Institute of Technology)
- Tetsunao Matsuta (Saitama University)
For more information about this CRONOS project, please contact the principal investigators via email (wadayama@nitech.ac.jp, shunwata@cc.tuat.ac.jp).