A novel efficient and low-complexity congestion management method in a differentiated service (DiffServ) domain, so-called edge-based congestion management (ECM), to improve network resource utilization, is proposed. Based on a pro-active congestion control and a dynamic rate regulation, the core nodes and the edge nodes cooperate to respond early to the onset of congestion and to quickly recover from the congestion. This cooperative ECM is implemented using a modRED and a token bucket filter (TBF) rate regulator. Based on early local congestion notification, the ingress nodes adjust their traffic rates via a state-dependent token consumption of the outgoing packets at the TBF. It regulates traffic, avoiding loss during congestion period, and takes advantage of transient available bandwidth quickly by exploiting the lower round trip times between a given pair of ingress and egress nodes. The ECM scheme does not require a per-flow state maintenance hence it is scalable, in addition to its low complexity that is comparable to that of the RED and the token bucket. The simulation study shows that packet loss in the domain can be substantially improved with ECM by at least 30% over a non-feedback scheme especially at high utilization factors. It is also able to maintain the DiffServ domain in a congestion-free state thereby improving overall resource utilization. Finally, the ECM is robust under the presence of unresponsive, aggressive applications in that the packet loss is localized at the ingress TBF with more traffic than that can be accommodated at the core of a DiffServ domain.
[1]
J. Turner,et al.
New directions in communications (or which way to the information age?)
,
1986,
IEEE Communications Magazine.
[2]
Sally Floyd,et al.
Promoting the use of end-to-end congestion control in the Internet
,
1999,
TNET.
[3]
V. Jacobson,et al.
Congestion avoidance and control
,
1988,
CCRV.
[4]
QUTdN QeO,et al.
Random early detection gateways for congestion avoidance
,
1993,
TNET.
[5]
Ariel Orda,et al.
QoS routing in networks with inaccurate information: theory and algorithms
,
1999,
TNET.