|
Search our Site
ON THIS PAGE:
Course Objectives
Prerequisites
Approach
Typical Text
Course Outline
|
|
COMP 831 [241]: Internet
Architecture and Performance
(3 hours)
Course
Objectives
The phenomenal growth of the Internet in sheer size and diversity of
applications has created important new problems in the engineering of
large-scale networks. This course focuses on contemporary research issues
related to network scalability, quality of service for specific applications,
and differentiated levels of service based on economic considerations.
Various approaches to solving these problems are presented and analyzed
with particular emphases on understanding the characteristics of traffic
using the network and the performance of services that are provided
by the network to its users.
Prerequisites
COMP 631 or permission of the instructor (Note: COMP 431 or equivalent
is sufficient for permission).
Approach
This course emphasizes empirical approaches to networking research and
engineering. Students, working in teams, complete two projects: (1)
an experiment using a laboratory network to investigate performance
tradeoffs among different approaches to implementing some network service,
and (2) an analysis of network traffic using data collected from production
networks. Students also individually conduct an experiment using the
laboratory network to gain experience dealing with issues in measurement
methodology for networks.
The lectures and readings from the research literature provide a foundation
for understanding the major architectural and performance issues in
large-scale networks like the Internet. They will also provide the necessary
background information to support the empirical components of the course.
Typical Text
Readings from the research literature are used instead of a text.
Course Outline
Numbers in parentheses indicate approximate number of weeks
- Internet Structure and Architecture Review (IP/UDP/TCP) (3)
- Traffic Analysis (3)
- Essentials from statistics
- Probability distributions, parameter estimation, QQ plots,
autocorrelation
- Modeling traffic sources
- Network-independent characterizations
- Modeling traffic aggregations
- Arrival processes, long-range dependence, self-similarity
- Errors and Error Recovery (1)
- Congestion and Congestion Control (3)
- End-to-end strategies (TCP and variants)
- Router-based strategies (RED, ECN, DECBIT)
- Non-responsive traffic and congestion collapse
- Service Models and Implementations (3)
- Integrated services
- Differentiated services
- Active queue management vs scheduling
- Addressing and Unicast Routing (1)
- Multicast Routing (1)
For more information, see the home page for the most recent offering
at http://www.cs.unc.edu/Courses/comp241-s02/
|
