On the self-similar nature of Ethernet traffic (extended version)
IEEE/ACM Transactions on Networking (TON)
Traffic matrix estimation on a large IP backbone: a comparison on real data
Proceedings of the 4th ACM SIGCOMM conference on Internet measurement
On guaranteed smooth scheduling for input-queued switches
IEEE/ACM Transactions on Networking (TON)
Designing packet buffers for router linecards
IEEE/ACM Transactions on Networking (TON)
IEEE Spectrum
A low-jitter guaranteed-rate scheduling algorithm for packet-switched IP routers
IEEE Transactions on Communications
SNDlib 1.0—Survivable Network Design Library
Networks - Network Optimization (INOC 2007)
IEEE Journal on Selected Areas in Communications
IEEE/ACM Transactions on Networking (TON)
Advanced QoS provisioning in IP networks: the European premium IP projects
IEEE Communications Magazine
Fundamental design issues for the future Internet
IEEE Journal on Selected Areas in Communications
Hi-index | 0.00 |
A framework for an Autonomic Future Internet which supports 2 services classes, the Essentially-Perfect QoS (QoS) class and the Best-Effort (BE) class, is proposed. All provisioned traffic flows in the QoS class can achieve 100% throughput and essentially-perfect QoS guarantees. Bandwidth is provisioned for the QoS class periodically using measured and projected traffic demand matrices. Each router uses an efficient stochastic matrix decomposition scheduling algorithm to to achieve essentially-perfect QoS guarantees, for every competing QoS traffic flow, i.e., all admitted competing QoS traffic flows never experience congestion. (This QoS scheduling problem is a long-standing unsolved theoretical problem.) The multicasting of aggregated self-similar video streams over the proposed network is explored. It is shown that thousands of bursty self-similar video streams can be multicast across the proposed network with essentially-perfect throughput and QoS guarantees. The technology can be incorporated into Internet routers with minimal hardware cost, it is compatible with the existing IETF DiffServ and MPLS service models, it can achieve link efficiencies as high as 100%, and it can reduce Internet router buffer requirements and power requirements significantly.