Broadcasting in Ad Hoc and Sensor Networks

We consider dynamic wireless networks in which each node has the same and fi xed transmission range. Ad hoc wireless networks are considered in conference, rescue, and military scenarios, while small size battery powered sensor networks are envisioned for monitoring the environment, with individual sensors reporting observed events (movement, temperature, chemicals, etc.) to a base station. The physical layer is approximated by assum ing that a message sent by one of nodes is received correctly by all neighbors located within the transmission range ( one-to-all communication). Among the common problems found in ad hoc and sensor networks is broadcasting , where a message is to be sent from one node to all hosts of the network. Applications of this broadcasting communication include route discovery, information dissemination, and synchronization. Because of limited range of radio waves, a host is normally unable to directly communicate with all the recipient nodes. To fulfi ll a broadcasting task, many hosts thus have to act as routers by relaying the message to their physical neighbors. The simplest procedure is to have each node relay the message (exactly once) to its neighbors. This straightforward solution is known as blind fl ooding and is part of the IETF standardized dynamic source routing (DSR) protocol, described by Johnson et al. 2004 . However, in non-sparse networks, this solution generates a lot of redundancy and collisions that could possibly prevent the broadcasting from being correctly performed. This issue, studied by Ni et al. 1999 , is known as the broadcast storm problem . Moreover, signifi cant energy is consumed by the redundant messages, which is an important consideration in low power sensor nodes. A number of other broadcast protocols, more energyeffi cient, have been proposed to replace blind fl ooding. Some of them are centralized , where nodes need to know the global topology of the network. In ad hoc or sensor networks, this causes a huge and unacceptable overhead for information exchange. Moreover, this information generally has a low lifetime due to frequent topology modifi cations which occur because of mobility and changes in activity status from active to sleeping and vice versa. Ad hoc and sensor networks require protocols, called localized , that rely only on local information that can be easily and quickly acquired. Methods covered here require each node to know the position of all its neighbors (1-hop positional knowledge), or the list of all its direct neighbors and their neighbors (2-hop topological knowledge). These protocols also need to be reliable . The reliability of a broadcast protocol refers to its capability of reaching all nodes connected to the source, assuming an ideal MAC layer. A high or even perfect delivery ratio is required for most applications. Localized broadcast protocols covered in this chapter provide reliable broadcasting. They can be divided into two categories: