A routing algorithm for connectionless networking in B-ISDN
To achieve an efficient routing scheme in a B-ISDN connectionless overlay network, the
existing hierarchical, distributed and link-state routing protocols need to be improved, in
order to minimise the end-to-end delay and prevent congestion situations. This paper
presents a new routing protocol, based on the existing hierarchical and link-state protocols,
which uses dynamic metrics, but is improved with two additional mechanisms to avoid the
frequent updates of the routing tables. The first mechanism consists in dividing the traffic into …
existing hierarchical, distributed and link-state routing protocols need to be improved, in
order to minimise the end-to-end delay and prevent congestion situations. This paper
presents a new routing protocol, based on the existing hierarchical and link-state protocols,
which uses dynamic metrics, but is improved with two additional mechanisms to avoid the
frequent updates of the routing tables. The first mechanism consists in dividing the traffic into …
Abstract
To achieve an efficient routing scheme in a B-ISDN connectionless overlay network, the existing hierarchical, distributed and link-state routing protocols need to be improved, in order to minimise the end-to-end delay and prevent congestion situations. This paper presents a new routing protocol, based on the existing hierarchical and link-state protocols, which uses dynamic metrics, but is improved with two additional mechanisms to avoid the frequent updates of the routing tables. The first mechanism consists in dividing the traffic into a finite set of load classes, considering that a metric is associated with a load class. It is possible to reduce the bandwidth used by the routing protocol, as the metric changes only if a new load class is reached. The second mechanism consists in finding an order of path selection, within the shortest path set, according to the traffic conditions. The advantage of this second mechanism lies in two facts: first, the time used to order the set of shortest paths is smaller than the time needed to calculate an entire new set, and second, the additional information introduced provides a load splitting within the shortest path set, which might reduce the frequency of class variation.
Springer
Showing the best result for this search. See all results