Simulation and performance analysis of time-critical real-time LANs

• Main Author: Amin, Issam
• Published: University of Sussex 2004
• Subjects: 621.3981
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.407752

Simulation and performance analysis of wired and wireless Time-Critical Real-Time(TCRT) LANs is the subject of this research work. Special emphasis has been placed on deterministic Medium Access Control (MAC) method as defined by the Institute of Electrical and Electronics Engineers (IEEE) 802.4 Token Passing Bus. Another very popular MAC method based on the IEEE 802.11 wireless LAN standard has also been investigated. The demands imposed by TCRT LANs are of a time-bounded nature which require messages to be delivered on time or within allowable delay defined by the applications they serve. A number of metrics for measuring network performance have been proposed for use throughout the thesis, specific important parameters, such as response time, waiting time, access time, and delay which greatly influence the performance of TCRT LANs have been analysed and examined in great detail. Fine tuning of these parameters was carried out to observe the influence they have on networks performance. Distinct approaches for measuring network performance are proposed and analysed. An analytical approach using mathematical models to determine network performance for different real-time process control applications is analysed and tested. The advantages and limitations of this approach are identified and evaluated for real-time applications. The second approach is modelling by simulation employing industry standards simulation tools, namely Network II.S and COMNET III. These simulation tools provide an effective platform for studying time-critical applications. Simulation models representing process control applications were created. A number of practical simulation models characterising real-time manufacturing cells have been modelled, analysed, and tested. Both simulation tools are used to model different network scenanos utilising the strengths and advantages of each. Simulation results and comparison of specific models were carried out. Network II.S is used to simulate IEEE 802.4 and COMNET III is used to simulate IEEE 802.11. A third approach based on an empirical network is investigated. Real data were collected and fed into a simulationmodel representing this practical network. Results from the simulation models were analysed and compared to evaluate the performance of the practical network and verify the simulation model. This cross-approach concept is found to be a very important way of studying performance of real-time LANs. A number of real-time network applications and scenarios representing process control applications were modelled using the various techniques. A generic network application was modelled to permit a comparison of the three methods. Most of the analyses are modelled using the simulation approach alone. This is due to the complexity and limitations involved in mathematically modelling dynamically changing situations in real-time applications. However, this approach was adopted only after having verified the correctness of the simulation models by cross referencing the results obtained from the mathematical and simulation approaches as applied to the generic (base) model. The simulation models enabled the analysis of the performance of IEEE 802.4 and IEEE 802.11 media access networks protocols used in real-time environments. Hypothetical and actual network scenarios were considered to fully investigate the effects of varying the various parameters on network performance. This research has clearly demonstrated that real-time networks impose different timing restrictions based on the applications they serve. Type of carried traffic by the real-time network plays a major role in influencing the choice of network protocols. Use of wireless network in real-time environment based on IEEE 802.11 under heavy load is ruled out under the current available proposals, however, they could be used in situation under low loading conditions serving small process control and manufacturing cells with limited number of processing elements. On the other hand, Real-time deterministic networks using access protocol based on IEEE 802.4 are found to be suitable for the most demanding network loading conditions and configurations. Simplifying the management functions of the IEEE 802.4 protocol reduces its complexity and costs of deployment without undermining its performance. This in tum will encourage more vendors to adopt the IEEE 802.4 standard for implementation in TCRT applications