Optimizing the Trickle Algorithm

The Trickle Algorithm has enjoyed much popularity and widespread use as a basic network primitive ensuring low-cost data consistency in lossy networks. Trickle is shaped by the so-called short-listen problem, hence the imposition of a listen-only period. Such a period allows Trickle to robustly address the short-listen problem at the expense of increased latency. In this letter, we introduce a simple yet powerful optimization to Trickle that can dramatically decrease Trickle's latency with virtually no additional overhead to its scalability and robustness. Extensive simulation and testbed experiments are reported here, yielding greater than a factor of 10 decrease in propagation time.