PTAS: A High-Performance TAS in TSN Networks

This repository contains a module of PTAS with inet and nesting in OMNeT++.

Introduction

PIFO-based Time-Aware Shaper (PTAS) is a novel TAS scheme for massive concurrent flows. In PTAS, a PIFO queue is responsible for sorting and buffering all time-triggered flows.

Through frame sorting rather than conventional isolation, the PTAS efficiently provides a deterministic transmission guarantee for concurrent flows

Compatibility

The current version of the PTAS module has been tested with OMNeT++ version 5.5.1, INET version 4.1.2 and NeSTiNg under Ubuntu16.04.

Getting Started

• Download and install OMNeT++ version 5.5.1 at https://omnetpp.org/.
• Clone and unzip nesting follow the instructions at https://gitlab.com/ipvs/nesting/-/blob/master/README.md.

  $ git clone https://gitlab.com/ipvs/nesting.git

• Clone and unzip INET version 4.1.2 framework.

  $ git clone https://github.com/inet-framework/inet.git
  $ git checkout -b v4.1.2 v4.1.2

• Clone and unzip PTAS module.

  $ git clone https://github.com/lxacc/PTAS.git

• Make sure the directory is like

  <workspace>
  ├── nesting
  ├── inet
  └── ptas

Create a Project in OMNeT++

• Launch OMNet++ IDE and set the workspace to the directory where packages have been downloaded in the previous step
• Import INET,NeSTiNg,and PTAS packages into your workspace.
  • File -> Import
  • select General>Existing Projects into Workspace option and Next>
  • under Select root directory add your previously chosen workspace directory
  • select inet, nesting and ptas projects and Finish
• The three projects should appear in the Project Explorer

Build and run

Build

• First, right click ptas project and click on Properties
• Under Project References, select inet and nesting, Apply and Close (Important)
• Each project can be set to release or debug build by right click and Build Configurations -> Set Active
• Right click each project (in the order inet, nesting, ptas) and Build Project
• You can now run the simulation

Run simulation

• Navigate into ptas > simulations in the project explorer
• Right-click paper_test_ptas_7hop.ini and choose Run As (or Debug As depending on the build configuration) OMNeT++ Simulation
• After finishing the simulation, find the statistics file in result-dir and export the end-to-end delay data. These data are the source for the figures in the paper

Change parameters

• Flow24.ned is the topology used for simulation, which is shown below
• In the topology, RobotController is the source of time-triggered flows, and RoboticArm is the destination. 24 RobotControllers are set to simulate a multi-access scenario with massive concurrent flows
• If a new topology is needed, please modify Flow24.ned
• paper_test_ptas_7hop.ini, paper_test_spq_7hop.ini, and paper_test_tas_7hop.ini are the configurations of the proposed PTAS and the benchmarks
• In simulations/xml, there are flow and gate configurations in PTAS, TAS, and SPQ with 4, 8, 12, 16, 20, and 24 flows. Each flow is set with the following parameters

  <schedules>
    <defaultcycle>1000us</defaultcycle>
    <host name="robotController1">
      <cycle>1000us</cycle>
      <entry>
        <start>0us</start>
        <queue>7</queue>
        <dest>00:00:00:00:01:01</dest>
        <size>95B</size>
        <burst>2</burst>
        <flowId>1</flowId>
      </entry>
    </host>
  </schedules>

The load of one flow can be calculated according to the <size> (frame length), <burst> (number of frames in a cycle), and <cycle>

• When running with different numbers of flows, please modify the xml's name in the corresponding .ini, such as

  **.filteringDatabase.database = xmldoc("**xml/PTASRouting24.xml**", "/filteringDatabases/")

  **.switchA.eth[24].queue.gateController.initialSchedule = xmldoc("xml/Flow24_ptas.xml", "/schedules/switch[@name='switchA']/port[@id='24']/schedule")
  ...

  **.robotController*.trafGenSchedApp.initialSchedule = xmldoc("xml/Flow24_ptas.xml")