%0 Conference Proceedings
%T Machine Learning Methods for Connection RTT and Loss Rate Estimation Using MPI Measurements Under Random Losses
%+ Oak Ridge National Laboratory [Oak Ridge] (ORNL)
%+ Argonne National Laboratory [Lemont] (ANL)
%A Rao, Nageswara
%A Imam, Neena
%A Liu, Zhengchun
%A Kettimuthu, Rajkumar
%A Foster, Ian
%< avec comité de lecture
%( Lecture Notes in Computer Science
%B 2nd International Conference on Machine Learning for Networking (MLN)
%C Paris, France
%Y Selma Boumerdassi
%Y Éric Renault
%Y Paul Mühlethaler
%I Springer International Publishing
%3 Machine Learning for Networking
%V LNCS-12081
%P 154-174
%8 2019-12-03
%D 2019
%R 10.1007/978-3-030-45778-5_11
%K Round Trip Time
%K Loss rate
%K Message Passing Interface
%K Machine Learning
%K Generalization bounds
%K Regression
%K Information fusion
%Z Computer Science [cs]
%Z Computer Science [cs]/Networking and Internet Architecture [cs.NI]Conference papers
%X Scientific computations are expected to be increasingly distributed across wide-area networks, and Message Passing Interface (MPI) has been shown to scale to support their communications over long distances. Application-level measurements of MPI operations reflect the connection Round-Trip Time (RTT) and loss rate, and machine learning methods have been previously developed to estimate them under deterministic periodic losses. In this paper, we consider more complex, random losses with uniform, Poisson and Gaussian distributions. We study five disparate machine leaning methods, with linear and non-linear, and smooth and non-smooth properties, to estimate RTT and loss rate over 10 Gbps connections with 0–366 ms RTT. The diversity and complexity of these estimators combined with the randomness of losses and TCP’s non-linear response together rule out the selection of a single best among them; instead, we fuse them to retain their design diversity. Overall, the results show that accurate estimates can be generated at low loss rates but become inaccurate at loss rates 10% and higher, thereby illustrating both their strengths and limitations.
%G English
%Z TC 6
%2 https://inria.hal.science/hal-03266451/document
%2 https://inria.hal.science/hal-03266451/file/487577_1_En_11_Chapter.pdf
%L hal-03266451
%U https://inria.hal.science/hal-03266451
%~ IFIP-LNCS
%~ IFIP
%~ IFIP-TC
%~ IFIP-TC6
%~ IFIP-LNCS-12081
%~ IFIP-MLN