T.M. Lim, B.S. Lee, and C.K. Yeo (Singapore)
Information Infrastructure, Quality-of-Service, AggregatedScheduling, DiffServ
Today, we know that serving aggregated flows of common characteristic scales in the core network. However, this may not meet the fast growing needs of application network requirements. It has been shown that DiffServ is unable to fulfill the latency requirements of a collaborative virtual re ality applications when network becomes overloaded. In addition, core networks loose important traffic information when flows are aggregated. Such information can be im portant in the design of future schedulers. EVTRS pro vides core routers with aggregation and flow information. At the ingress, packets are classified into different queues and scheduled by their eligible time. EVTRS views each queue as a packet train. Using variables embedded in each packet, packet trains can be reconstructed at core nodes. With a relative virtual clock method, more information can be derived. Lastly, we illustrate a method of mapping com mon QoS flows onto our system with simulation.
Important Links:
Go Back
