An Efficient Distributed Network-Wide Broadcast Algorithm for Mobile Ad Hoc Networks John Sucec and Ivan Marsic CAIP Center, Rutgers University Piscataway, NJ 08854 Abstract--In this paper, an algorithm for efficient network-wide broadcast (NWB) in mobile ad hoc networkds (MANETs) is proposed. The algorithm is performed in an asynchronous and distributed manner by each network node. The algorithm requires only limited topology knowledge, and therefore, is suitable for reactive MANET routing protocols. Simulations show that the proposed algorithm is on average 3-4 times as efficient as brute force flooding. Further, simulations show that the proposed algorithm compares favorably over a wide range of network sizes, with a greedy algorithm using global topology knowledge, in terms of minimizing packet transmissions. The application of the algorithm to route discovery in on-demand routing protocols is discussed in detail. Proofs of the algorithm's reliability and of the intracatability of solving for a minimum sized transmitter set to perform NWB are also given. I. INTRODUCTION Network-wide broadcast (NWB) is an essential feature of some of the emerging on-demand routing protocols in mobile ad hoc networks (MANETs). MANETs are multiple-hop, packet-forwarding networks consisting of potentially mobile nodes interconnected by wireless links. Since for on-demand routing protocols nodes do not proactively maintain forwarding tables for destinations lying beyond some small radius that is less than the network diameter, nodes must routinely perform a reactive route discovery procedure. As implemented in the Ad hoc On-demand Distance Vector (AODV) [1] and Dynamic Source Routing (DSR) [2] protocols, route discovery involves disseminating a route request (RREQ) packet throughout the network. Without an efficient NWB algorithm, such a route discovery procedure results in considerable packet overhead. Both AODV [1] and DSR [2] have timeout mechanisms that allow the protocol to reinitiate a route query if the previous attempt fails. Thus, the route discovery process for these protocols assumes a potentially unreliable NWB procedure. A given NWB may fail for a number of reasons including partitioning of the network and packet collisions due to the shared media nature of the network interfaces. (Since network partitioning can cause NWB to fail for any protocol, this paper will consider only networks that are connected.) In any event, if a particular RREQ packet fails to reach the target node (T), or a node that has a path to T in its route cache, then the route discovery attempt will be declared a failure and reinitialized by the source node (S). Due to the retry measure of the on-demand route discovery process, it is evident that although an unconditionally reliable NWB implementation is desirable, it is not essential. This is a useful property of the AODV and DSR protocols. Instead of implementing a NWB procedure that is proven to be 100% reliable in that it guarantees packet delivery to all network nodes within some bounded time, one need only to implement a procedure that in the absence of collisions or network partitions can be shown to be reliable. Thus, with the requirement of unconditionally reliable NWB relaxed for on-demand route discovery, a NWB procedure that requires modest topology knowledge and yet achieves high operational efficiency can be considered. Minimizing the topology information required by the NWB procedure is crucial for AODV and DSR because neither protocol proactively acquires topology information beyond a single hop radius of a given node (DSR actually specifies zero proactive acquisition of topology data). This means the NWB procedure must be able to function with considerably less than complete topology knowledge. Further, due to the fact that AODV and DSR are routing protocols for flat (non- hierarchical) network topologies, topology structure built into networks with hierarchical or clustering properties, as in [8,10], cannot be exploited for NWB. Lastly, both protocols have minimal periodic routing message overhead and it is desirable not to substantially increase the existing volume of periodic messaging. In terms of performance, a new NWB procedure should minimize the number of transmissions needed to disseminate a given packet (e.g., a RREQ packet) to every network node. The number of transmissions