Dataverse Infrastructure

Presentation at BDVA Valencia conference in the Metaverse track

Transcript

1. Metaverse Telecommunication infrastructure for the virtual worlds Joaquín Salvachúa, Andres Muñoz, Gabriel Huecas, Alvaro Alonso @ UPM
2. Background
3. Origin Terminology
4. Philip K dick : Perky Pats multiverse (and minority report …)
5. Avoid Been under the control of a single Company Need to have common protocols
6. Avoid previous errors : Center into distributed infrastructure, not rendering
8. Other names • Cyber-Physical-Social Systems (CPSS), the abstract and scientific name for metaverses. (IEEE Intelligent Systems (IS) ) • Cyberspace • Multiverse • Networked digital twins • Surreality
9. Mission • Provide a sound architecture as much distributed as posible. • Evolve from actual GAIA-X/data Spaces ideas about trust and distribution. • Provide sound distributed API semantics to build upon them different applications. • Start with data engineering field.
10. Proposal
11. Evolution :: Network Digital twin ( initial steps ) digital twin (DT) paradigm to model complex dynamic systems. network digital twin (NDT) goal is to build accurate data-driven network models that can operate in real-time. NDTs aim to achieve accurate data-driven network models operating in real-time Importance of failures models (key on distributed system design) ML-based NDT for routing optimization in a QoS-aware optimization use case. Webrtc as transport for real time data
12. Data spaces as enabler for data metaverse • Data spaces should be the layer where to build data metaverses over it. • Each metaverse client instance should be able to access data via dataspaces interfaces. • Failures models are key in this definition to succed (NFS vs ZFS). • Distributed data gobernance must be enforced over all the system : • We belive on formal methods for this (formal verification and validation in real time).
13. Data
14. Trust :: adding the web 3 capabilities • Using Blockchain to gain trust in distributed scenarios : • Depends heavily on the use case. • Adds extra complexity and performance problems. • Smart contracts as base for distributed trust aspects. • Maybe to slow in most cases; need to use just when needed. (DLT are not always just blockchain).
15. Distributed Data engeneering Colaborative data engineering Great to make sense over different dataspace (remember dataspaces maybe distributed ones). Full integration of AI and ML capabilities Needs a lot of work right now
16. OPEN CHALLENGES Data collection and storage in distributed way : linked to ML-OPS infrastructure Generalization and scalability to real networks : need to learn from P2P and distributed multimedia collaboration systems (choose a semantic for failure and recover) Fine-grained control and management : linked to distributed data gobernance on DataSpaces Dealing with uncertainty
17. Architecture Dataverse-X Meta-X Collaboration space Collab-X AI-Dapps Space Gaia-X Trust-X Reality-X
18. Thank you!