Wireless Sensor Networks : Technology, Protocols, and Applications

220 ROUTING PROTOCOLS FOR WIRELESS SENSOR NETWORKS
may be required for a router to compute the shortest path to each destination [6.28].
Furthermore, to maintain correct paths to all destinations, routers are called upon to
update the state describing the current topology in a periodic fashion and when link
failure occurs. The need to update the topology state constantly may lead to substantial
overhead, proportional to the product of the number of routers and the rate
of topological changes in the network.
Geographical routing, on the other hand, does not require maintaining a ‘‘heavy’’
state at the routers to keep track of the current state of the topology. It requires only
the propagation of single-hop topology information, such as the position of the
‘‘best’’ neighbor to make correct forwarding decisions. The self-describing nature
of geographical routing, combined with its localized approach to decision, obliterates
the need for maintaining internal data structures such as routing tables.
Consequently, the control overhead is reduced substantially, thereby enhancing its
scalability in large networks. These attributes make geographical routing a feasible
solution for routing in resource-constrained sensor networks.
Routing Strategies The objective of geographical routing is to use location information
to formulate a more efficient routing strategy that does not require flooding
request packets throughout a network. To achieve this goal, a data packet is sent to
nodes located within a designated forwarding region. In this scheme, also referred
to as geocasting, only nodes that lie within the designated forwarding zone are
allowed to forward the data packet [6.23,6.24]. The forwarding region can be statically
defined by the source node, or constructed dynamically by intermediate
nodes to exclude nodes that may cause a detour when forwarding the data packet.
If a node does not have information regarding the destination, the route search can
begin as a fully directed broadcast. Intermediate nodes, with better knowledge of
the destination, may limit the forwarding zone in order to direct traffic toward
the destination. The idea of limiting the scope of packet propagation to a designated
region is commensurate with the data-centric property of sensor networks, in which
the interest in the data content, rather than the sensor, provides the data. The efficacy
of the strategy depends largely on the way the designated forwarding is
defined and updated as data travel toward the destination. It also depends on the
connectivity of the nodes within a designated zone.
A second strategy used in geographical routing, referred to as position-based
routing, requires a node to know only the location information of its direct neighbors
[6.25,6.26]. A greedy forwarding mechanism is then used whereby each node
forwards a packet to the neighboring node that is ‘‘closest’’ to the destination.
Several metrics have been proposed to define the concept of closeness, including
the Euclidean distance to the destination, the projected distance to the destination
on the straight line joining the current node and the destination, and the deviation
from the straight direction toward the destination.
Position-based routing protocols have the potential to reduce control overhead
and reduce energy, as flooding for node discovery and state propagation are localized
to within a single hop. The efficiency of the scheme, however, depends on the
network density, the accurate localization of nodes, and more important, on the