Predictable and Robust Data Distribution in Systems of Embedded Systems

1
Predictable and Robust Data Distribution in
Systems of Embedded Systems

• Roberto Baldoni
• Middleware Laboratory Joint-Lab on Security
  Research
• Dipartimento di Informatica e Sapienza
  Innovazione
• Sistemistica Antonio Ruberti
• Università di Roma La Sapienza
• Long Terms Challenges in High Confidence Embedded
  Systems
• Helsinki, June 21-23, 2006
• Email baldoni_at_dis.uniroma1.it
• Url http//www.dis.uniroma1.it/baldoni/
• MIDLAB Url http//www.dis.uniroma1.it/midlab/
  

2
MIDLAB research lines

• Interoperability in Large and Dynamic Distributed
  Systems
• Objectives
• Shared operational pictures
• Global Data Management
• Robust Service provisioning
• Distributed monitoring

Cooperative Information Systems
NetCentric Computing
Critical Infrastructure protection
3
Sapienza Innovazione

• Incubator of the University of Rome La Sapienza
• Partners
• Joint-Labs
• Aerospace
• Biotech
• Technology for Cultural assets
• Genomics
• Security
• Nanotech

Biotech against terrorism Critical Infrastructure
Protection Interoperability of complex
systems Situation awareness Robotics and
Vision Intelligent Sensors and sensors networks
4
Cybersecurity in Complex Networked Embedded
Systems Current Status

• Ubiquitous connectivity
• Usage of IP technology for private networks
• Private networks intersect the Internet
• SCADA systems are becoming built with COTS
• Increase of system vulnerabilities!

5
Cybersecurity in Complex Networked Embedded
Systems Current Status
Priority of Security
6
Cybersecurity in Complex Networked Embedded
Systems Future
App.s Power Systems Nowcasting Operating Room
Archeology excavation Security check of event
places
TRUST
Traffic confinement
Instant deployment
Multi-systems application
Multi-systems application
7
Cybersecurity in Complex Networked Embedded
Systems Future

• Long term goal
• exchange of information among trusted parties,
  on a peer-to-peer base, wherever it is needed,
  whenever it is needed
• Priority System Monitoring
• Need Predictable and robust large scale data
  distribution

8
Where are we?

• Data distribution
• bringing data information to the right people at
  the right time
• Data distribution is (more or less) predictable
  and robust in small scale environments (see
  RTI-DDS and OpenSlice products)
• Data distibution is best effort on large scale.
  Problems in
• scalability
• timeliness
• ordering
• dynamic discovery of peers
• persistence
• How to enable predictable and robust data
  distribution on large scale?

9
Important Results from distributed systems in the
last 10 years

• Peer-to-peer systems
• Sensors Networks
• Mobile ad-hoc networks
• Autonomic computing and communication
• Event based systems (Publish/subscribe systems)

10
Problem Arena
MIDDLEWARE
Security Topology Management Dependability Real-Ti
me
11
Problems to be solved to enable this vision

• Network level From client-server to peer-to-peer
  (autonomous) network management and trust
• Overlay network level security, robustness,
  predictability, topology control
• Data Dissemination level QoS tradeoffs, QoS
  graceful degradation issues

12
Industry

• Moving from static to dynamic network management
• Instrumenting overlay network tools
• Moving from small scale data distribution to
  large scale one
• Implements basic functions of data distribution
  on the ground

13
Timescale for research progress

• Parallel and intersecting process among basic
  research, RD for getting products asap
• Peer-to-peer network management
• Basic research yr 6
• RD yr 8
• Products in 7 years
• Rapid deployable and customizable overlay
  networks
• basic research yr 5
• RD yr 7
• Products in 5 years
• Large scale Data Distribution services
• basic research yr 7
• RD yr 7
• Products in 10 years

14
Conclusion

• Data distribution for large scale networked
  embedded systems is a priority
• Many research issues inside for getting the
  required QoS
• Network level
• Overlay network level