Résumé:
measurements of local and wide-area network traffic have proven that the
widely used Markovian process models cannot be applied for today’s network traffic. If the traffic
were Markovian process, the traffic’s burst length would be smoothed by averaging over a long time
scale contradicting with the observations of today’s traffic characteristics. Measurements of real
traffic also prove that traffic burstiness is present on a wide range of time scales. Traffic that is
bursty on many or all time scales can be characterized statistically using the concept of long-range
dependency. Long-range dependent traffic has noticeable bursts, long periods with extremely high
values on all time creating traffic congestions. Several conventional methods have been implemented
to avoid congestion, but they are not responsive enough to the varying transmission capacity and
network delay in high-speed networks.
A new research direction, programmable networking, offers a more promising solution for congestion control. Traditional packet networks perform only the processing necessary to forward packets
toward their destination. As computing power becomes cheaper, more and more functionality can be
deployed inside the network. Programmable networks support dynamic modification of the network
software and hardware to manipulate the network’s behavior. A special way of network programmability is when special programs called mobile or software agents are carried in the packets to the
routers. Software agents are loaded, executed, migrated, and suspended in order to implement some
network functions. Software agents provide the highest possible degree of flexibility in congestion control. They can carry congestion-specific knowledge into the network at locations where it b needed,
rather than transferring information to the sending hosts as it happens in traditional flow conlrol
solutions.
In our paper, we propose a new programmable network architecture using software agents that
can reduce the harmful consequences of congestions due to aggregated bursty traffic, such as packet
losses. extremely long response times, etc. @ 2003 Elsevier Ltd. All rights reserved.
