Real-Time Communication - Industrial Embedded Systems

1
Real-Time Communication - Industrial Embedded
Systems
Hoai Hoang, Magnus Jonsson School of Information
Science, Computer and Electrical Engineering,
Halmstad University, Sweden
2
Problem statement

• Embedded systems become more complex
• Increase the demand on heterogeneity, QoS
• Low down the cost installation, maintenance

ERIEYE radar system (EMW)
3
Main research's goals

• Developing Real-time services over System Area
  Networks using Switch Ethernet
• Support special traffic pattern for embedded
  system
• Find the communication solution
• Meet real-time demand
• Guarantee networks performances

4
Project overview

• Using pure switched base network
• Increase heterogeneity, performance...
• Support time-critical applications
• Focus on real-time service in industrial and
  embedded system.
• Provide both real-time and non-real-time traffic

5
Related work
Real-time traffic handling
Coordination ofmany-to-manycommunication etc
Switched Full-Duplex Real-Time Ethernet Network
6
Previous result

• Proposed method for distribution end-to-end
  deadline for messages
• Improve network performances special master-slave
  communication

7
Previous result
75 more channels using asymmetric deadline
partitioning for a network with 10 master nodes
and 50 slave nodes. Each channel i is
characterized by Ci3, Pi100, Di40
8
Real-time over WANs

• Extend the network topology
• More switches in the network
• Multihop network
• Guarantee real-time communication over WANs

9
Real-time over WANs

• Sender, Receiver
• Deadline (relative)
• End-to-End (D)
• Node-to-Node (Dj)
• Minimum (minDj)
• The RTC feasible If

10
Real-time over WANs

• Slack can occur
• Question What the system do with the slack?

11
Real-time over WANs

• Ferrari and Verma Divide it equally between the
  Node-to-Node deadlines of the channel
• We divide it such that links with a history of
  demanding traffic gets more

12
Real-time over WANs

• We want to see if our approach is better, in the
  sense that more real-time traffic can be feasible
  schedule.
• For what kind of topologies?
• Under what kind of traffic patterns?

13
Real-time over SAN - Future works
Emerging SAN (and LAN) standards

• Real-time services
• Normal data exchange
• Synchronization of software processes
• Real-time debugging
• Clock synchronization
• Group communication (many-to-many etc)
• etc

Single real-timemulti-service SANin each
application

• Problem The SAN standards do not deliver this
  real-time multi-service functionality as is. At
  the same time the industry has to cope with
  complex designs, high engineer efforts and high
  product costs when forced to use several
  service-specific networks in the same application.

14
Real-time over SAN - Future works

• Heterogeneous communication
• Heterogeneous network
• Heterogeneous services
• Example Radar signal processing systems
• Control traffic, periodic with a master-slave
  pattern and delay-bound
• Data traffic (radar data), periodic with a very
  high bit-rate and a partially pipelined data
  flow. Probabilistically guaranteed delay-bound
• Other non-real-time traffic