- Title:
- Statistical Multiplexing and Connection Admission Control
- Presenter:
- Guoqiang Mao
- Date:
- November 08, 2002
- Abstract:
- Bandwidth hungry computer and communication applications are on the rise
with a variety of services, as a few examples, voice over IP, video
conference, video on demand, high definition television, etc. Different
from traditional data communication applications, these applications are
real-time applications. They have the so-called quality of service
requirements.
Connection Admission Control (CAC) is the most effective traffic control
mechanism, which is necessary in networks in order to avoid possible
congestion at each network node, and to achieve the Quality-of-Service
(QoS) requested by each connection. CAC determines whether or not the network
should accept a new connection. A new connection will only be accepted
if the network has sufficient resources to meet its QoS requirements
without affecting the QoS commitments already made by the network for existing
connections. The design of a high-performance CAC is based on an
in-depth understanding of the statistical characteristics of the traffic sources.
This presentation will discuss the statistical characteristics of the
network traffic, and present a computationally efficient and robust CAC
scheme based on a novel closed-loop architecture, which is capable of
achieving high network resources utilization while satisfying the QoS
requirements of all connections. Other aspects of network traffic, e.g.
self-similarity in network traffic and its impact, network traffic
measurement, will also be discussed.
Back to:
|
