The “Swarmchestrate: Application-level Swarm-based Orchestration Across the Cloud-to-Edge Continuum” project aims to develop a decentralized orchestration framework inspired by swarm intelligence to manage complex, hyper-distributed applications across the Cloud-to-Edge continuum. This approach addresses challenges associated with the exponential growth of intelligent devices generating vast amounts of data at network edges, which traditional centralized cloud processing struggles to handle due to latency, bandwidth, and security concerns. Key objectives include creating decentralized AI-driven matchmaking algorithms for optimized resource allocation, establishing secure knowledge management using blockchain technologies, and implementing real-world demonstrators in areas such as flood prevention and urban noise classification.
Our team is the responsible partner for Working Package 1, and is mainly involved in the following tasks:
T1.1: Definition of logical proximity for Swarms: TUB contributes to defining the concept of logical proximity, which extends the current understanding of Swarms from being based on geographical location to a multidimensional space. This involves formalizing the semantics of logical proximity to include various contextual information like resource specification, energy constraints, and network characteristics.
T1.2: Ad-hoc formation of self-organising Swarms: TUB leads this task, focusing on implementing the ad-hoc formation and management of resources in an orchestration space to create interconnected Swarms. This involves enabling the deployment of container-based microservices and managing the dynamic nature of resources joining and leaving Swarms.
T1.3: Collaboration and interaction between Swarms: Again, TUB leads this task, which is about integrating mechanisms for Swarm-level interaction and collaboration. This includes developing an auto-discover functionality for Swarms to be aware of other Swarms and providing interfaces for inter-Swarm collaboration.
Partners of Swarmchestrate Project: