“Minimizing Noise in HyperLogLog-Based Spread Estimation of Multiple Flows”

Congratulations to all of the authors, Nguyen, Jiyoo, Changhun, David!

Our paper on estimating the spreads of multiple flows is accepted at the 52nd Annual IEEE/IFIP International Conference on Dependable Systems and Networks(acceptance ratio=49/262). We proposed RRSE (Rank Recovery-based Spread Estimator), an algorithm on counting multiple spreaders on a programmable router, which is a multi-tenant version of the famous Hyper-LogLog. The importance of this paper is in the way to eliminate noise from HLL’s estimation when sharing a memory space with multiple flows.  RRSE shows greater noise reduction performance compared to vHLL, MCSE, state of the art algorithms.

Check out this presentation clip!