Broadcast with variable data rates in mobile ad hoc networks

We assess the broadcast transmission cost in ad hoc networks with multipoint relay (MPR) flooding and a variable data rate. Mobile ad hoc networks consists of wireless nodes that build a robust radio network without any pre-existing infrastructure or centralized servers. Tactical mobile communication frequently generates multicast traffic (one-to-many) and broadcast (one-to-all) traffic. Some examples are distribution of status information, position information and voice group calls. Using variable and adaptive rates on the links in a network will potentially both improve capacity and allow nodes at difficult positions to access the rest of the network with minimal degrading of the network capacity. The routes in the network will also be more robust if its links can degrade gracefully with decreasing rates rater than disappearing at some SNR threshold. In this study, we assume that all nodes use the same single data rate for transmissions and that they can simultaneously vary this rate to improve the performance. A main reason for this approach is to find indications on good strategies for multi-rate broadcast algorithms where different rates are allowed at the same time in the network. For reference, we also evaluate the transmission cost for unicast traffic (one-to-one) with variable data rates. Here, it seems to be a good strategy to always use the largest possible data rate that does not partition the network. A similar general strategy for the broadcast case is difficult to formulate. However, for networks with low frequency bandwidths and high node densities, there is a transmission cost minimum close to the data rate for which centrally located nodes can reach all other nodes. For these networks it might be a good strategy to chose the data rate so that we have a twohop broadcast: one hop in to a central node and one hop out to all other nodes in the network. This resembles the transmission behaviour in a cellular base station radio system, which is perhaps not the way we are used to think about ad hoc networks.