An iterative statistical method for congestion prevention in transit networks

With the expanding amount of data transferred over communication links it is necessary to improve the links and appropriate network devices to match the traffic requests. The most common way of increasing network throughput and performance in general is usually the replacement of the network devices and links. This way is reliable but usually expensive. Different way of improving network performance is the change of way how the traffic is handled and distributed over network. In this paper we propose an algorithm for dynamic traffic rerouting in IP networks based on statistical probability and load experienced on a network link. This algorithm accomplishes congestion prevention and even distribution of traffic on available network resources.