Developing and managing distributed systems is a complex task that requires expertise in various domains such as programming models, hardware, networking, operating systems, and distributed protocols.
TaRDIS aims to develop cutting-edge technology to simplify the process of building efficient and reliable
heterogeneous swarms– swarm systems that are heterogeneous, intelligent, dynamic, and decentralised – by leveraging machine learning (ML), programming models, and data management.
TaRDIS aims to develop:
1. A
novel programming model for heterogeneous swarms that is language-independent and event-driven.
2. A
development environment for correct-by-design heterogeneous swarms through novel verification techniques and tools to analyse decentralised systems.
3.
Decentralised intelligence for heterogeneous swarms through methods such as Federated Learning.
4.
Runtime support for distributed heterogeneous swarms that concerns communication, swarm membership, and data management.
5. An
interoperable execution environment across different devices and programming languages.
The combination of standalone tools, such as application programming interfaces (APIs), that are developed with the vision described above composes the TaRDIS toolbox. The TaRDIS approach is guided and validated by four highly distinct, and compelling, use cases provided by high impact industrial partners that range from swarms of satellites, decentralised dynamic marketplaces in the energy sector, decentralised machine learning solutions for privacy-preserving applications in the setting of intelligent homes, to distributed control processes of smart factories.
Telefónica’s primary role in the TaRDIS project concerns the use case on intelligent homes. In a nutshell, the intelligent home paradigm usually includes a range of highly heterogeneous devices designed to work together as a swarm, through artificial intelligence (AI) algorithms, to assist us and make our lives more comfortable. Common concerns in this setting include the privacy of personal information and the heterogeneity of computational resources. The TaRDIS toolkit will be used to abstract the infrastructure, data distribution, and learning algorithms from the developer.
This will lead to:
1)
Collaborative intelligence irrespective of heterogeneity in local data, resources, learning goals, etc...
2)
A correct-by-design development environment implementing privacy-preserving solutions.
Further, Telefónica plays an important role in the efforts towards defining the functional requirements for the TaRDIS development environment, and the programming abstractions required for building intelligent heterogeneous swarms.
Finally, Telefónica also contributes to the development of AI/ML methods for distributed systems, and to activities related to standardisation, open-source, and policies.