The SmartEdge project aims to enable decentralized edge intelligence for smart IoT applications, ensuring reliability, security, privacy, and scalability. This is achieved through the innovative SmartEdge tool-chain for autonomous intelligent swarms, featuring real-time semantic integration, discoverability, and composability.
To illustrate, envision a group of individuals with diverse roles collaborating on shared tasks. Each person independently decides on actions based on their senses and knowledge, communicating with others through messages. People can join or leave the group and switch roles as needed. Knowledge fusion ensures everyone has up-to-date situational awareness.
Now, envision replacing some or all of these individuals with software systems—cognitive agents on devices with various capabilities. Device sensors and actuators mimic human senses and muscles, necessitating handling of diverse communication technologies and data formats. Knowledge graphs simplify application development by representing this complexity.
Agents are modeled in terms of perception, cognition, and action, building on the concept of digital twins. Perception involves interpreting sensor data and messages, while cognition entails decision-making. Actions include messaging other agents and real-time control over actuators for low-latency control loops.
This approach views the swarm as a collection of communicating agents, with emergent behavior resembling a resilient “hive mind” capable of adapting to agents joining or leaving, hardware and network issues, demand spikes, and cyber-attacks.