Technion

1
Technion Israel Institute of Technology
Computer Networks Laboratory Digital laboratory
Real Time Ethernet
Semester Winter 2001
       Students Shay Auster Hagit Chen
      Supervisor Vitali Sokhin
2
RTE - Preview

• An Ethernet protocol for Real-Time.
• Analytic analisys of estimated performance.
• Design an adiqute simulation.
• Run various scenarios in simulation.
• Conclusions.

3
Abstract

• Real Time Streaming requires a bound on the time
  of which a packet is created until it reaches its
  destination.
• IEEE 802.3u protocol does not support this
  requirment.
• Hence, a Real Time Ethernet protocol needs to be
  defined.

4
RTE Protocol - Overview

• Combine Ethernet and RTE transmisions on the same
  network.
• On the same Lan All RTE stations support the
  same application.
• In order to coordinate transmisions between RTE
  stations A mechanism to serializes
  transmisions.

5

• Serializations of RTE transmitsions

• Head and Tail are required for the Handshaking -
  a mechanism which serealizes RTE transmisions.

6
RTE frame
Ethernet frame bounded between a head and a tail

• Head
• Clean channel for RT transmisions.
• Notify all other RT stations on RTE transmision
  status.

• Tail
• Notify all other RT stations on RTE transmisions
  status.

7
Overview cont.

• Two possible situations in channel
• RTE transmision in channel A new RTE station
  join the end of the chain.
• No RTE transmision The RTE station generates a
  new chain.
• A RTE chain transmision in channel
• RTE station interupt at the end of the chain no
  handshaking at 1st time.
• Part of chain - handshaking from next time.

8
Final ResultsAnalysis
9
Ethernet always transmits

• Basic Ethernet simulation.
• Stations always have packets to transmit.

10
Ethernet always transmits

• Ethernet simulation results are used as a
  reference in analysing RTE simulation results.

11
RTE Always transmits

• Ethernet always transmit.
• RTE According to protocol.

12
RTE always transmits

• A Single RTE Station
• Various number of Ethernet stations

13
RTE always transmits

• Three RTE Station
• Various number of Ethernet stations

14
RTE always transmits

• Five RTE Station
• Various number of Ethernet stations

15
Ethernet The poissonic case

• Poissonic arrival of packets to stations.
• The interval between arrival of packets is
  exponential distributed ? poissonic arrival of
  packets.
• For exponential probability function we used an
  inverse distribution function.

16
Ethernet poissonic case

• Ethernet packets arrival rate is poissonic.
• t 1000uSec mue 1

17
Ethernet poissonic case

• Ethernet packets arrival rate is poissonic.
• t 500uSec differnet mue (0.5/1/2)

18
Ethernet poissonic case

• Ethernet packets arrival rate is poissonic.
• Different t (500/1000/2000uSec) mue 1

19
RTE The poissonic case

• Ethernet Poissonic arrival of packets to
  stations.
• RTE According to protocol.

20
RTE poissonic case

• Ethernet packets arrival rate is poissonic.
• A single RTE station.
• t 1000uSec mue 1

21
RTE poissonic case

• Ethernet packets arrival rate is poissonic.
• Three RTE stations.
• t 1000uSec mue 1

22
RTE poissonic case

• Ethernet packets arrival rate is poissonic.
• Five RTE stations.
• t 1000uSec mue 1

23
RTE poissonic case

• Ethernet packets arrival rate is poissonic.
• Different RTE stations.
• t 1000uSec mue 1

24
Ethernet The On/Off case

• On Always transmits.
• Off Never transmits.
• The on/off intervals are exponentily distributed.

25
Ethernet On/Off case

• 64 Bytes packet.
• Different On/Off data.

26
Ethernet On/Off case

• 256 Bytes packet.
• Different On/Off data.

27
Ethernet On/Off case

• 1024 Bytes packet.
• Different On/Off data.

28
RTE The On/Off case

• Ethernet -
• On Always transmits.
• Off Never transmits.
• RTE According to protocol.

29
RTE On/Off case

• 1024 bytes Ethernet packets.
• A Single RTE station.
• Different On/Off data.

30
RTE On/Off case

• 1024 bytes Ethernet packets.
• Three RTE stations.
• Different On/Off data.

31
RTE On/Off case

• 1024 bytes Ethernet packets.
• Five RTE stations.
• Different On/Off data.

32
Ethernet Stations Wait Time

• Ethernet Allways transmit.
• No RTE.
• Wait time increases with packet size.

33
RTE Stations Wait Time

• Ethernet Allways transmit.
• One RTE station.
• Wait time increases with packet size.
• Wait time increases with number of RTE stations.

34
RTE Stations Wait Time

• Ethernet Allways transmit.
• Three RTE stations.
• Wait time increases with packet size.
• Wait time increases with number of RTE stations.

35
RTE Stations Wait Time

• Ethernet Allways transmit.
• Five RTE stations.
• Wait time increases with packet size.
• Wait time increases with number of RTE stations.

36
RTE - Jitter

• Ethernet Allways transmit.
• Various number of RTE stations.
• Jitter increases with packet size number of RTE
  stations.

37
Time to genrate RTE chain

• Ethernet Allways transmit.
• Various number of RTE stations.
• Chain time increases with number of RTE stations.

38
Application example

• Ethernet Allways transmit.
• Various number of RTE stations.
• Application sampeling rate 1.5Mbps.

39
Conclusions

• RTE stations uses a part of the Ethernet channel
  ? Ethernet stations Efficiency decreases.
• The total chanel efficiency increases.
• For Ethernet allways transmit on/off arrival
  times we get an immediate reduce of efficiency.
• For poisonic arrival of packets we dont get an
  immediate reduce of efficiency.

40
Conclusions

• For each arrival pattern channel efficiency
  converges to the allways transmits results (for
  sufficient number of stations).
• More stations (regular/RTE) ? Larger wait time.
• Bigger packets ? Larger wait time.? Larger
  Jitter.