An Empirical Study of Flooding in Mesh Networks Thomas Zahn, Greg OâShea and Antony Rowstron Microsoft Research, Cambridge, UK 1. INTRODUCTION Flooding in wireless mesh networks is a fundamental operation to many routing (e.g. AODV, OLSR), networklevel (e.g. server-less DNS) and application-level (e.g. WSDiscovery) protocols. Thus, e cient ooding is important. It has been shown that naive ooding can generate broadcast storms [2]. This has inspired much research on optimized ooding, most of which has been based on analysis and simulation. However, it has been shown that broadcasts can behave very di erently in real-world wireless networks [1]. The goal of this paper is to quantitatively evaluate the performance of ooding protocols on a large-scale o ce 802. 11a mesh network of 110 nodes distributed across four oors. We compare three protocols: naive ooding and two optimized ooding protocols. The two optimized protocols are inspired by the multi-point relay algorithm [3], which selects subsets of the nodes to rebroadcast. We have found that there is a clear trade-o for ood-based protocols between e ciency and reliability. This implies that such protocols need to explicitly manage this trade-o . A full version of this paper is available as
/lp/association-for-computing-machinery/an-empirical-study-of-flooding-in-mesh-networks-yrV008GOE8