This course reviews how to model computing resources in terms of quantity and analyzes the performance of the models. It covers basic stochastic modeling, the theory of queues and its application to real cases.
- The principal goal of this course is to make the students build their own ability to model and analyze any kind of computer-related mechanisms, which is essential for their graduate studies and professional careers.
- Students will be trained with the basic queueing theory.
- Students will have hands-on experiences with simulation tools such as SMPL, OMNet++, and NS3.
- Students will have chances of using a real-time simulator for automotive systems.
- Students will learn and analyze example algorithms including
- resource scheduling for embedded systems
- real-time tracking in ubiquitous computing environments
- wireless sensor network protocols
Computer Modeling Overview
- what to model and how to model
- analytic method
- simulation method
Concept of Simulation
- real-time vs. simulation time
- workload modeling
- level of abstraction (circuit level simulation, instruction level simulation, event-level simulation)
Quick Look at Simulation Tools
- Xilinx circuit simulator
- CPU simulator
- OMNet++ event driven simulator
- NS3 network simulator
Modeling a Single FIFO Queue
- analytic method (M/M/1 analysis)
- simulation (hand-made simulation with no simulation tool support)
SMPL
- simple event simulation
- exercise with a sinle FIFO queue
Queueing Network
- analysis
- simulation (hand-made simulation with no simulation tool support)
Queueing Network Simulation with SMPL
- how easy to extend a single FIFO queue
- why we need a more structured way of simulation?
OMNet++
- object-oriented event simulation
- concepts and terminology of OMNet++
- simple tutorial example
OMNet++ Exercise
- simulate EDF-scheduling with resource reclaiming in single processor
- simulate EDF-scheduling with resource reclaiming in multi-processors
NS3
- object-oriented network simulation
- concepts and terminology of NS3
- simple tutorial example
NS3 Exercise
- simulate ad-hoc wireless protocols (e.g., AODV, DSDV)
Data Collection and Analysis
- what data need to be collected
- how the collected data should be averaged
- how to analyze and understand the variations
- how to present the data
Real-Time Scheduling in Embedded Systems (with OMNet++)
- Scheduling of multiple concurrent tasks with data-dependency
- Scheduling of sensors, computers, and wireless mediums for real-time tracking
Real-Time Simulator for Automotive Systems - Part 1
Real-Time Simulator for Automotive Systems - Part 2