Smart trunk scheduling strategies for future integrated services packet-switched networks

摘要

Trunk scheduling is considered as a means to reallocate delay fairly among various traffic classes and thus reduce the overall network delivery delay of packets. Two service prioritization schemes to accomplish this in integrated services packet switching networks are discussed. These are based on either the total HOP distance (source-destination distance) between the packets' source and destination or that weighted by the packets' local delay at the switching node. Three types of servicing are considered, and the case of highest-priority servicing (head of the line) is studied in some detail. It is shown while the highest-priority servicing of total-distance prioritization can potentially lead to extensive end-to-end network delays for packets in the low-priority classes, the delay-weighted total-distance case results in a more fair and balanced treatment of all priority classes. The basic fast hardware architecture is presented for implementation of the delay-weighted total-distance-based service prioritization, which should make the discipline suitable for practical purposes.