PROJECT TITLE :

ICE Buckets: Improved Counter Estimation for Network Measurement - 2018

ABSTRACT:

Measurement capabilities are essential for a variety of network applications, such as load balancing, routing, fairness, and intrusion detection. These capabilities require massive counter arrays in order to watch the traffic of all network flows. While commodity SRAM recollections are capable of operating at line speed, they're too little to accommodate large counter arrays. Previous works recommended estimators, which trade precision for reduced area. But, so as to accurately estimate the largest counter, these strategies compromise the accuracy of the smaller counters. In this Project, we present a closed form illustration of the optimal estimation operate. We have a tendency to then introduce independent counter estimation buckets, a novel algorithm that improves estimation accuracy for all counters. This is achieved by separating the flows to buckets and configuring the optimal estimation perform according to each bucket's counter scale. We prove a tighter higher bound on the relative error and demonstrate an accuracy improvement of up to fifty seven times on real Net packet traces.