Information Technology Reference
In-Depth Information
schemes for mobile sensors which react to events such that the distribution of the
group of sensors tends toward the distribution of the sensed events. The algorithms
rely on an initial uniform distribution of sensors and can then generate good results
without requiring the sensors to communicate and exchange the information about
their positions to each other. One technique uses a very simple update rule to move
each sensor based only on the position of the event and the sensor itself (similar to
the potential field's approach described in the previous section). This movement is
easily computed and the amount of motion can be easily bounded. The second
technique uses more computation and keeps a compressed history of the events.
Here, to determine its correct position, each sensor maintains a discrete version of
the CDF (cumulative distribution) which is updated after each event. The CDF is
scaled based on the number of events and the length l of the interval of interest, so
that CDF( l ) = l. Then each segment of the CDF is associated with a proportional
number of sensors, and thus the sensor density will truck the event density. Or in
other words, each sensor chooses its new position so that the CDF at this position
returns its initial position.
Another method for stochastic event capturing (SEC) is presented in [ 32 ]. The
authors have derived algorithms for motion planning based on defined quality of
coverage (QoC) metrics, such as the fraction of events captured and the probability
that an event is lost. The analytical results provide guidelines for choosing the
velocity and the number of sensors to be deployed for satisfying constraints on
fraction of captured events. Only mathematical analyses and proofs derived on
approximation factor are presented, while neither simulations nor emulations are
made to test the feasibility of the proposed solutions.
Another event-driven sensor self-deployment algorithm (EDSSA) is proposed in
[ 33 ], based on the virtual force algorithm, where the potential force of the detected
event is added to the calculation of the vector sum. This force pulls distant nodes
toward the event location and pushes nearby nodes away. By adjusting the event inten-
sity threshold, different node density can be achieved around the event location. The
approach considers large-scale relocation of nodes, compared with relocation toward
coverage holes in some of the previous approaches. Starting from the initial topology,
the nodes will tend to a regular triangular pattern position, achieving the optimal cover-
age when the length of the grid parts is
D = , where R s is the sensing range.
These examples for sensor mobility are representative examples of the models
found in literature. They form the basics allowing for more comprehensive study
later on, since familiarity with the state-of-the-art approaches and their pros & cons
is helpful for a research in the considered area. Moreover, the following text pres-
ents the main mobility challenges and possible ways to handle them.
Mobility Challenges: Data Dissemination
Mobility in WSNs is a permanent and challenging issue. It is often followed by
network topology modification, necessity of new protocols, and consequently,
requirement of appropriate mobility models. Above all, challenges lie in designing
Search WWH ::

Custom Search