Low-latency Synchronization of Loosely-coupled Sensornet Republishing. Park, U. & Heidemann, J. Technical Report ISI-TR-2009-660b, USC/Information Sciences Institute, April, 2009. updated June 2010 (original at ˘rlhttp://www.isi.edu/%7ejohnh/PAPERS/Park09a_200904.pdf)Paper abstract bibtex Today many individual deployments of sensornets are successful, but they will have much greater impact when, rather than standing alone, they share data across deployments so each can build upon the others. We expect data to be shared over the Internet, and as the number of processing and reprocessing steps grows, \emphtimely data synchronization is increasingly important. Today, such sharing is often hard-coded or driven by fixed-interval polling. Fixed-interval polling can provide poor worst-case performance (mean latency approaching the data generation period), and best performance requires careful manual configuration of both poll period and phase. We instead propose \emphData Generation Tracking (DGT), a new family of adaptive polling algorithms that \emphlearn and predict good times to pull data to minimize both latency and unfruitful queries. Our approach avoids manual configuration and automatically adapts to outages and changes in data generation rate. To evaluate our work, we examine four sensornet deployments and develop a rough \emphmodel of sensornet data generation. We then use this model and replay of real traces to evaluate DGT, finding that, depending on application, its median latency is only 10–30% of that of fixed-interval polling, with a configurable rate of network load that is the same or slightly higher.
@TechReport{Park09a,
author = "Unkyu Park and John Heidemann",
title = "Low-latency Synchronization of Loosely-coupled Sensornet Republishing",
institution = "USC/Information Sciences Institute",
month = apr,
year = 2009,
sortdate = "2009-04-01",
project = "ilense, siss",
jsubject = "sensornet_sharing",
note = "updated June 2010 (original at \url{http://www.isi.edu/%7ejohnh/PAPERS/Park09a_200904.pdf})",
number = "ISI-TR-2009-660b",
location = "johnh: pafile",
keywords = "sensornet, sensor-internet, republishing,
synchronization, RSS",
url = "http://www.isi.edu/%7ejohnh/PAPERS/Park09a.html",
pdfurl = "http://www.isi.edu/%7ejohnh/PAPERS/Park09a.pdf",
myorganization = "USC/Information Sciences Institute",
copyrightholder = "authors",
abstract = "Today many individual deployments of sensornets
are successful, but they will have much greater
impact when, rather than standing alone, they share
data across deployments so each can build upon the
others. We expect data to be shared over the
Internet, and as the number of processing and
reprocessing steps grows, \emph{timely data
synchronization} is increasingly important. Today,
such sharing is often hard-coded or driven by
fixed-interval polling. Fixed-interval polling can
provide poor worst-case performance (mean latency
approaching the data generation period), and best
performance requires careful manual configuration of
both poll period and phase. We instead propose
\emph{Data Generation Tracking} (DGT), a new family
of adaptive polling algorithms that \emph{learn and
predict} good times to pull data to minimize both
latency and unfruitful queries. Our approach avoids
manual configuration and automatically adapts to
outages and changes in data generation rate. To
evaluate our work, we examine four sensornet
deployments and develop a rough \emph{model of
sensornet data generation}. We then use this model
and replay of real traces to evaluate DGT, finding
that, depending on application, its median latency
is only 10--30\% of that of fixed-interval polling,
with a configurable rate of network load that is the
same or slightly higher.",
}
Downloads: 0
{"_id":"hAj678YSezu5CX8j5","bibbaseid":"park-heidemann-lowlatencysynchronizationoflooselycoupledsensornetrepublishing-2009","author_short":["Park, U.","Heidemann, J."],"bibdata":{"bibtype":"techreport","type":"techreport","author":[{"firstnames":["Unkyu"],"propositions":[],"lastnames":["Park"],"suffixes":[]},{"firstnames":["John"],"propositions":[],"lastnames":["Heidemann"],"suffixes":[]}],"title":"Low-latency Synchronization of Loosely-coupled Sensornet Republishing","institution":"USC/Information Sciences Institute","month":"April","year":"2009","sortdate":"2009-04-01","project":"ilense, siss","jsubject":"sensornet_sharing","note":"updated June 2010 (original at ˘rlhttp://www.isi.edu/%7ejohnh/PAPERS/Park09a_200904.pdf)","number":"ISI-TR-2009-660b","location":"johnh: pafile","keywords":"sensornet, sensor-internet, republishing, synchronization, RSS","url":"http://www.isi.edu/%7ejohnh/PAPERS/Park09a.html","pdfurl":"http://www.isi.edu/%7ejohnh/PAPERS/Park09a.pdf","myorganization":"USC/Information Sciences Institute","copyrightholder":"authors","abstract":"Today many individual deployments of sensornets are successful, but they will have much greater impact when, rather than standing alone, they share data across deployments so each can build upon the others. We expect data to be shared over the Internet, and as the number of processing and reprocessing steps grows, \\emphtimely data synchronization is increasingly important. Today, such sharing is often hard-coded or driven by fixed-interval polling. Fixed-interval polling can provide poor worst-case performance (mean latency approaching the data generation period), and best performance requires careful manual configuration of both poll period and phase. We instead propose \\emphData Generation Tracking (DGT), a new family of adaptive polling algorithms that \\emphlearn and predict good times to pull data to minimize both latency and unfruitful queries. Our approach avoids manual configuration and automatically adapts to outages and changes in data generation rate. To evaluate our work, we examine four sensornet deployments and develop a rough \\emphmodel of sensornet data generation. We then use this model and replay of real traces to evaluate DGT, finding that, depending on application, its median latency is only 10–30% of that of fixed-interval polling, with a configurable rate of network load that is the same or slightly higher.","bibtex":"@TechReport{Park09a,\n\tauthor = \t\"Unkyu Park and John Heidemann\",\n\ttitle = \t\"Low-latency Synchronization of Loosely-coupled Sensornet Republishing\",\n\tinstitution = \t\"USC/Information Sciences Institute\",\n\tmonth = \tapr,\n\tyear = \t\t2009,\n\tsortdate = \"2009-04-01\",\n\tproject = \"ilense, siss\",\n\tjsubject = \"sensornet_sharing\",\n\tnote = \"updated June 2010 (original at \\url{http://www.isi.edu/%7ejohnh/PAPERS/Park09a_200904.pdf})\",\n\tnumber =\t\"ISI-TR-2009-660b\",\n\tlocation = \t\"johnh: pafile\",\n\tkeywords = \t\"sensornet, sensor-internet, republishing,\n synchronization, RSS\",\n\turl =\t\t\"http://www.isi.edu/%7ejohnh/PAPERS/Park09a.html\",\n\tpdfurl =\t\"http://www.isi.edu/%7ejohnh/PAPERS/Park09a.pdf\",\n\tmyorganization =\t\"USC/Information Sciences Institute\",\n\tcopyrightholder = \"authors\",\n\tabstract = \"Today many individual deployments of sensornets\n are successful, but they will have much greater\n impact when, rather than standing alone, they share\n data across deployments so each can build upon the\n others. We expect data to be shared over the\n Internet, and as the number of processing and\n reprocessing steps grows, \\emph{timely data\n synchronization} is increasingly important. Today,\n such sharing is often hard-coded or driven by\n fixed-interval polling. Fixed-interval polling can\n provide poor worst-case performance (mean latency\n approaching the data generation period), and best\n performance requires careful manual configuration of\n both poll period and phase. We instead propose\n \\emph{Data Generation Tracking} (DGT), a new family\n of adaptive polling algorithms that \\emph{learn and\n predict} good times to pull data to minimize both\n latency and unfruitful queries. Our approach avoids\n manual configuration and automatically adapts to\n outages and changes in data generation rate. To\n evaluate our work, we examine four sensornet\n deployments and develop a rough \\emph{model of\n sensornet data generation}. We then use this model\n and replay of real traces to evaluate DGT, finding\n that, depending on application, its median latency\n is only 10--30\\% of that of fixed-interval polling,\n with a configurable rate of network load that is the\n same or slightly higher.\",\n}\n\n","author_short":["Park, U.","Heidemann, J."],"bibbaseid":"park-heidemann-lowlatencysynchronizationoflooselycoupledsensornetrepublishing-2009","role":"author","urls":{"Paper":"http://www.isi.edu/%7ejohnh/PAPERS/Park09a.html"},"keyword":["sensornet","sensor-internet","republishing","synchronization","RSS"],"metadata":{"authorlinks":{}}},"bibtype":"techreport","biburl":"https://bibbase.org/f/dHevizJoWEhWowz8q/johnh-2023-2.bib","dataSources":["YLyu3mj3xsBeoqiHK","fLZcDgNSoSuatv6aX","fxEParwu2ZfurScPY","7nuQvtHTqKrLmgu99"],"keywords":["sensornet","sensor-internet","republishing","synchronization","rss"],"search_terms":["low","latency","synchronization","loosely","coupled","sensornet","republishing","park","heidemann"],"title":"Low-latency Synchronization of Loosely-coupled Sensornet Republishing","year":2009}