Approximate Causal Observer
Sandeep S. Kulkarni and Mahesh Arumugam
Abstract
In this paper, we focus on the problem of approximate causal delivery. This
problem identifies the tradeoff between causal delivery and timely delivery of
messages. Causal delivery requires that delivery of a message, say $m$, be
delayed until all messages on whom $m$ is causally dependent are delivered. By
contrast, timely delivery requires that messages be delivered as soon as
possible. However, the requirements of causal delivery and timely delivery are
conflicting. We show how a simple logical timestamp program can be used to
obtain a solution for approximate causal observer. This solution is intended
for sensor networks that provide simple guarantees about the clock drift among
sensors and about maximum delay of messages that are not lost. Our solution
lets the sensors to choose the level of causality violations it can tolerate
($0\%$ or more) and the time for which it will have to buffer the received
messages. We also show that our solution provides a continuum where the
application can choose the size of the timestamps it maintains by identifying
the level of causality violations it can tolerate.
Paper:
Return to the publication list
Return to the Sandeep's home page