Position-Based Routing Protocols for Ad Hoc Networks - Application and Multidisciplinary Aspects of Wireless Sensor Networks - page 73

of PAGs is medium since the failure of an individual node may cause the loss of a
packet in transit, but it does not require setting up a new route. Their implementa-
tion complexity is also considered as medium due to the usage of both greedy and
CFace algorithms.
Pure greedy forwarding algorithms have low processing overhead; however,
PAGs' processing overhead is considered to be medium due to the adjustment of
transmission range and the multiple switching among different algorithms, espe-
cially with the use of CFace(3) that requires projection of nodes into the three
planes. CFace detects the loops, i.e., PAGs are loop-free algorithms. PAGs' probability
of finding an optimal path is low due to using greedy forwarding and discovering
the nodes' neighbors, using low-transmission ranges. Finally, PAGs can be imple-
mented in both sparse and dense networks (considering the delivery rate and net-
work lifetime). Their simulations support especially the second routing algorithm
PAG:CFace(3), which increased the delivery rate to around 100% for both sparse
and dense networks. The advantage that can be obtained when the PAGs are imple-
mented in dense networks is the high probability of finding a path even with nodes
working at low transmission level; no need to change among different algorithms.
However, this will lead to longer paths. On the other hand, implementing PAGs in
sparse networks will increase the probability of finding a path compared to pure
greedy algorithms, but time will be lost in switching to high transmission range and
other CFace algorithms.
In SPAAR, each node maintains a neighbor table that contains the identity and
position information of each verified neighbor; the used location service type is
all-for-some. The source node can calculate the approximate geographic location of
the destination from the most recent location and most recent velocity information
stored in the source node's destination table. On the first attempt at communication
with a particular destination, the source may use a location service or a selective flooding
algorithm to reach the destination and receive its position information. The general
robustness of this approach is medium, since the position of a node will become
unavailable if a significant number of nodes fail.
SPAAR uses the restricted directional flooding, so it exhibits some properties
such as the high probability of using the optimal path. Moreover, it is loop-free
since it depends on forwarding the packet to the nodes toward the destination and
uses sequence number.
SPAAR tolerates position inaccuracy by the expected region; each node forwards
the RREQ only if it, or any of its neighbors, is closer to the destination. Its robust-
ness is low since the failure of an individual node might result in packet loss and
the setting up of a new route. SPAAR has high implementation complexity since
messages must be verified, signed with the private key and encrypted with the public
key of a neighbor. But it is still less than the complexity of SGF since there is no
reputation system.
SPAAR assumes the existence of one certificate server, which may be the
operation bottleneck especially in large area networks. Moreover, increasing
the number of nodes in the network with using the restricted directional flooding
will increase the packet overhead. Finally, in large area networks the probability