Network-Wide Broadcasts for Wireless Mesh Networks with regard to reliability

Network-Wide Broadcasts (NWBs) are a fundamental operation for Wireless Mesh Networks (WMNs) used by routing and for information distribution in general. Integrating reliability into this protocol primitive was rarely tackled because acknowledgements and retransmissions would amplify the so-called broadcast storm problem. It is also no strict requirement for the common use case. Some still require a high reliability, e.g. an alarming service for a WMN-based warning system. Traditional optimizations for NWBs try to remove most or all redundancy which arises from trivial flooding to avoid the broadcast storm problem. We show experimentally that this idea can hurt reliability. In this paper we develop a model for reliable NWBs and propose strategies to calculate a forwarder set taking into account link qualities. Further we use acknowledgements and retransmissions for reliability. These strategies are evaluated in simulations and testbeds together with well-known approaches from literature. Our strategies outperform existing NWB approaches with regard to reliability and experiments give further quite surprising results. It is not hard to make NWBs reliable, however most classic approaches but the simplest fail in this regard.