Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Scribd
Upload
40 views

Week 1-Introduction

Why performance evaluation? Gives a measure of the service delivered by a system Performance is one of the most important non-functional aspects of any (hardware or software) system Performance evaluation comprises of certain techniques Direct measurements using testbeds Analytical or simulation modeling applied to existing or envisioned systems like computer systems, communication networks, algorithms and protocols The basic and final goal of the performance evaluation study is to come to a con

Uploaded by

MI Brand
Copyright
© © All Rights Reserved
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
40 views

Week 1-Introduction

Why performance evaluation? Gives a measure of the service delivered by a system Performance is one of the most important non-functional aspects of any (hardware or software) system Performance evaluation comprises of certain techniques Direct measurements using testbeds Analytical or simulation modeling applied to existing or envisioned systems like computer systems, communication networks, algorithms and protocols The basic and final goal of the performance evaluation study is to come to a con

Uploaded by

MI Brand
Copyright
© © All Rights Reserved
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 17

MCN7212: Network Simulation

and Performance Evaluation


Course Outline
Week 1 Course Introduction
Week 2 Introduction to Statistics
Week 3 Random number generation
Week 4 Benefits and limitations of simulations
Week 5 Workload Characterization
Week 6 Evaluation and improvement of models
Week 7 Simulation technique
Week 8 Computer network simulation tools
Week 9 Introduction and Fundamentals of Queuing Theory
Week 10 Simulation of queuing models in communication networks
Week 11 Simulation of functions and performance of protocols
Week 12 Experimental Design
Week 13 Empirical Methods and Measurements
Week 14 Network Management
Week 15 Stochastic Processes
Week 16 Discrete-Time Markov Chains
Course Assessment
• Assignments -20%
• Course Work (Individual): Paper review, summary write-up, and
presentation
• Test-20%
• Final examination -60%
Reading Materials
Journals and Papers (2014-now)
Journal papers on network performance and evaluation

Text books:
• Performance Evaluation by Simulation and Analysis with Applications to
Computer Networks,
2015, John Wiley & Sons,by Ken Chen
• Modeling and Simulation of Computer Networks and Systems by
Mohammad S. Obaidat, Faouzi Zarai, Petros Nicopolitidis, Publisher: Morgan
Kaufmann, 2015
• Network Performance and Evaluation R. Jain, "The Art of Computer Systems
Performance Analysis: Techniques for Experimental Design, Measurement,
Simulation, and Modeling
Motivation
The Internet
• Defined by TCP/IP protocols stack
• Collaboration of a number of elements/ devices
• Based on connectionless communication
• Issues of congestion and routing

• Supports varying applications with varying requirements


• Random and dynamically changing traffic behaviors
• Performance issues depend on type of service and application
• Data transfer, Audio, Video, interactive
Communication Systems
Traditional telephone system

• Shared dedicated lines


• Supported only audio
• Blocking probability
Why performance evaluation?
• Gives a measure of the service delivered by a system
• Performance is one of the most important non-functional aspects of any
(hardware or software) system
Performance evaluation comprises of certain techniques
• Direct measurements
• using testbeds
• Analytical or simulation modeling
• applied to existing or envisioned systems like computer systems, communication
networks, algorithms and protocols
• The basic and final goal of the performance evaluation study is to come
to a conclusion, whether the system is working in a good condition or
satisfactorily.
Performance evaluation
• Increase in user demands
• Field of communication is continuously evolving and to reach these demands
• New standards are being developed
• Institute of Electrical and Electronics Engineers (IEEE)
• Third-Generation Partnership Project (3GPP)
• Designing and performance evaluation of these new emerging
technologies
• It is a major challenge
• Considering the fact that wireless medium is unpredictable.
Methods of Evaluation
Performance evaluation can be classified into two types:
(i) Performance measurement
• Possible only if the system of interest is available for measurement and only if
one has access to the parameters of interest. E.G Performance of a network
server using performance logs
(ii) Performance modeling
• Typically used when actual systems are not available for measurement or if
the actual systems do not have test points to measure every detail of interest.
Performance Modelling
Performance modeling may further be classified two:
1. Simulation modeling
• Simulation models may further be classified into numerous categories
depending on the mode/level of detail of simulation
2. Analytical modeling
• Analytical models use probabilistic models, queueing theory, Markov models
(Markov Chains) or Petri nets.
Measurement-based Evaluation
Desired metrics are obtained from measurements of the system
behavior, measured;
• either in an operational
• or under a controlled environment
• This technique inherently yields the most believable results
• Measurement based evaluation is a very expensive technique
• Since it requires an actual system in existence
• Evaluation is not applicable during system design.
Model-based evaluation
Always deliver results which are less accurate than the ones that can be
obtained by conducting actual measurements on the system
However, model-based evaluation overcomes the two primary
limitations of measurement-based evaluation
• it is less expensive
• it can be used during the design phase
In this case models can be of two types:
a) Analytical models
b) Simulation models
Model-based evaluation(2)
• Analytical Models
• Can be used to predict system behavior
• Analytical models are less expensive
Simulation Models
• The system behavior can be simulated and the desired performance and
dependability measures can be estimated
• Since the actual system is not needed, this technique can be applied during
the system design phase
• Note: For credible simulation results, the system has to be simulated for a
long time
The advantage of simulation over analytical modeling lies in the fact
that very detailed system behavior can be captured.
Science of Network Performance Evaluation
Performance Evaluation
• Requires intimate knowledge of system
• Careful selection of methodology
• Workload
• Tools
• No one correct answer as two performance analysts may choose
different metrics or workloads
• There are techniques to learn
• how to use them
• when to apply them
Common Network Performance Evaluation
Errors/mistakes
Undefined goals
• Describe goals (Objectives)
Biased goals
• Not always your system should be better than others
• Tradeoff
Unrepresentative workload
• Need variations Eg: large or small packet size, not only one
Wrong evaluation technique
• Choose most appropriate: {Model, simulation or measurement}
Inappropriate level of detail
• Too much or too little detail
Improper representation of results
• The number of graphs needed to make decisions
Omitting assumptions and limitations
Network Simulation ideas
• Adaptive Video Transmission Over the Wireless Internet
• User-Level Video Quality Estimation
• Video Quality Estimation for Mobile H.264/AVC
• Video Streaming
• User-Perceived Quality of Service
----END----

You might also like