P
Peter Key
Researcher at Microsoft
Publications - 142
Citations - 5171
Peter Key is an academic researcher from Microsoft. The author has contributed to research in topics: Network congestion & Network packet. The author has an hindex of 39, co-authored 142 publications receiving 5042 citations. Previous affiliations of Peter Key include University of Cambridge & BT Group.
Papers
More filters
End-User Policies for Predicting Congestion Patterns in Data Networks
TL;DR: Simulations show that, when a fraction of end users modify their congestion control based on estimations, oscillations in network buffer occupancy otherwise induced by TCP users are significantly reduced, which benefit both TCP users and the users applying the proposed scheme.
Cell Delay Variation and Burst Expansion in ATM Networks: Results from a Practical Study Using Fairisle
TL;DR: This study uses the Fairisle ATM LAN to measure the CDV experienced by CBR and periodic, bursty sources when multiplexed with Bernoulli, VBR video and LAN interconnection traac, in a network of non-blocking ATM switches, and concludes that signiicant performance beneets can be achieved using non-FCFS queueing policies.
Patent
Active probing for sustainable capacity estimation of networked dataflows
TL;DR: In this paper, the authors use actively probing to estimate the sustainable capacity of the network and calculate a delay trend for each probe packet, which leads to estimate of the maximum network capacity and the background load on the network.
Journal Article
Stochastic-Models of Computer-Communication Systems - Response-Time Problems in Communication-Networks - Scheduling and Characterization Problems for Stochastic Networks - Discussion
J. F. C. Kingman,JD Biggins,A Hordijk,MH Ackroyd,Philip K. Pollett,RJ Gibbens,JA Bather,Peter Key,IM Macphee,PJ Donnelly,C Bernerslee,S Stidham,Stan Zachary +12 more
In-call Probing and End-to-End Congestion Control: Theory and Performance
Alan Bain,Peter Key +1 more
TL;DR: This paper derives analytic models, based on diffusion limits under a natural scaling, to quantify the benefits of in-call probing, and shows that this simple theory is remarkably accurate in predicting large-scale behaviour.