Abstract
Network-Wide Broadcasts (NWBs) are a fundamental operation for Wireless Mesh Networks (WMNs), which are required by routing and for distributing information to all nodes in the network. Many optimizations for NWBs exist which aim to counter the broadcast storm problem by reducing redundant transmission. However removing the redundancy can also hurt reliability in lossy networks like WMNs. We therefore look at the other side. What happens if we increase redundancy to improve the reliability by using retransmissions and acknowledgements, which are said to amplify the broadcast storm? In this paper, a model for calculating forwarder sets locally based on measured link qualities is developed together with an algorithm for reliable NWBs. We evaluate our approach empirically in different simulation and testbed scenarios and compare it with protocols from the literature. The results show that redundancy is needed for achieving a high reliability and should not be abandoned. Furthermore, the impact of different parameters on our approach is evaluated.