|
|
||
|
OPNET Technologies OPNET Technologies,
Inc. |
TEACHING WITH OPNET A number of final year undergraduate and taught masters projects on the BEng in Information and Communications Engineering and taught MEng in Electronic Systems (Network Implementation Major) in the School of Electronic Engineering, Dublin City University use OPNET as the basis of the project implementation work. Examples of recently completed student projects (August 2011) are: 1. Application of Percolation Theory to the Analysis of Loss Networks Percolation theory is a branch of probability theory dealing with the structural and transport properties of porous media. The theory has classically been applied to modelling physical systems, for instance the flow of fluid through a porous mass, a mass containing holes at random sites which may form pathways through the material when they join. Recently, percolation theory has been proposed as a method of analysis of communications networks, by considering network nodes as conductive holes and traffic as a fluid. This project will investigate the applicability of the theory to analysing complex networks of communicating nodes where loss may occur when a node is fully utilised. An example of such a network is a wavelength division multiplex optical transport network. The experimentation will involve constructing a network model in the OPNET simulation environment (C/C++ language based) and then assessing the usefulness of percolation theory results in predicting the network’s performance. 2. Congestion Control Mechanisms for Optical Burst Switched Networks This project will evaluate congestion control algorithms for optical burst switched (OBS) networks by way of simulation studies, using an industry-standard discrete-event simulation tool OPNET. Implementation of the project will involve extending an existing network simulation, written in the C language, to include the communications mechanisms and algorithms required to improve network performance by preventing traffic congestion and burst loss occurring within the network. Possible approaches for investigation are traffic thinning/reshaping/buffering at ingress nodes and/or re-transmission of bursts. 3. Producing Empirical Queuing Formulae by Simulation Optimisation Techniques Queuing theory is a quite well-developed field with many applications in the design and analysis of communications networks. Although many types of queues have known solutions, there are many interesting systems which so far have no known exact closed-form analytic solutions, due to the intractability of the underlying mathematics, and the best approximate attempts at analysis of such systems can often yield quite inaccurate results. The objective of this project is to investigate how/if accurate empirical formulae may be produced by combining curve-fitting (e.g. Levenberg–Marquardt algorithm) and simulation optimisation techniques (e.g. ranking and selection). The project implementation will use the OPNET simulation environment, a C/C++-language simulation tool. A strong interest in engineering mathematics is advisable for this project. 4.Variance Reduction Methods for Rare-Event Analysis in Loss Systems In discrete-event network simulation, it is often difficult to estimate the likelihood of occurrence of events that happen only very occasionally. Simulation run-times can become infeasible long when such events are very rare and/or the required confidence in the estimate is high. None the less, there is often the requirement for such estimates, for example when estimating the probably of loss in a high-performance optical switching network. This project will investigate the efficacy of variance reduction methods to reduce the simulation run-time whilst maintaining accuracy of the results. The implementation of the project will involve first constructing a simple loss network to perform basic tests to investigate the fundamental properties of the methods and then to extend an existing simulation of an optical burst switching network to include the methods. This will involve use of the OPNET simulation tool. 5. Cost/Performance Study of Sparse/Partial Wavelength-Conversion in Optical Burst Switches The overall objective of this project is to determine the best cost/performance trade-offs for an optical burst switch (OBS) design by evaluating different possible configurations of wavelength converters in the switch. Implementation of the project will involve extending an existing network simulation, written in the C language and using the OPNET simulation environment. The goal is to derive a cost function from empirical data from simulation, which can then be used to inform switch configuration decisions.
PREVIOUS RESEARCH USING OPNET Optimal Design and Resource Dimensioning Methods Enabling Cost-Effective Optical Burst Switched Network Deployments Optical Burst Switching has been promoted as a low-cost solution for transporting very high bandwidth Internet traffic. The properties of burst switching give rise to a trade-off between potentially very efficient use of available optical fibre capacity against potentially high packet loss, if the network is not carefully optimised. This project aims to deliver methods for guaranteeing cost/performance levels through network resource analytic modelling and multi-objective optimisation techniques. The overall goal of these methods is to support technical and business decisions on migration to optical burst switching as a future high bit-rate Internet transport. Validation of analytic models developed during the project will involve building a detailed discrete-event simulation model of a large-scale OBS network using OPNET. This project runs from 2008-2011 and is sponsored under the Science Foundation of Ireland (SFI) Research Frontiers Programme. Research papers based on OPNET simulations from this, and earlier, projects are: D. Tafani, C. McArdle, L. P. Barry, "Analytical Model of Optical Burst Switched Networks with Share-per-Node Buffers," The Sixteenth IEEE Symposium on Computers and Communications (IEEE ISCC'11), Corfu, Greece, June-July 2011. C. McArdle, D. Tafani, T. Curran, A. Holohan, L. P. Barry, "Renewal Model of a Buffered Optical Burst Switch," IEEE Communications Letters, vol. 15, no. 1, pp. 91-93, January 2011. C. McArdle, D. Tafani, L. P. Barry, "Analysis of a Buffered Optical Switch with General Interarrival Times," Journal of Networks (JNW), Selected Papers of The International Symposium on Performance Evaluation of Computer and Telecommunication Systems (SPECTS 2009), Academy Publisher, accepted for publication, to appear Q1, 2011. C. McArdle, D. Tafani, L. P. Barry, A. Holohan, T. Curran, "Simplified Overflow Analysis of an Optical Burst Switch with Fibre Delay Lines," Proc. Sixth International Conference on Broadband Communications Networks and Systems (Broadnets 2009), September 2009, pub. IEEE. C. McArdle, D. Tafani, L. P. Barry, "Equivalent Random Analysis of a Buffered Optical Switch with General Interarrival Times," Proc. 2009 International Symposium on Performance Evaluation of Computer and Telecommunication Systems (SPECTS 2009), vol. 41, pp. 238-245, July 2009, pub. IEEE. C. McArdle and T. Curran, “Optimal Object Placement, Load Distribution and Load Control for Distributed Telecommunication Service Applications”, in Proceedings of 17th International Teletraffic Congress: Teletraffic Engineering in the Internet Era (ITC 17), pub. Elsevier, pp. 317-382, Salvador, Brazil, December 2001.
|
