Abstract:
Network congestion and unfair network resource
allocation are the main causes of the difficulty in providing QoS
to a network application. When a network becomes congested, it
is unable to provide adequate resources required by the
application. Even in the absence of congestion conditions, the
network may not be able to deliver the intended QoS since it is
not capable of allocating resources fairly among competing
traffic streams due to the varied attributes of IP traffic such as
packet size, round trip time, and traffic type (e.g., flow-controlled
TCP, uncontrolled UDP). This paper introduces a new congestion
control scheme, the Fair Intelligent Congestion Control (FlCC)
to tackle these problems. FlCC employs a feedback loop to
maintain the operation of the controlled network around a
desired target point, hence avoiding congestion. The key design
features of FICC include fairness, scalability, and efficiency.
Simulation results demonstrate that FICC handles congestion
effectively and allocates bandwidth fairly among competitive
sources. FlCC is scalable in that it only requires simple perqueue
rather than per-aggregate accounting.
