However, the extent of in-network
data aggregation that is possible, depends on the spatio-temporal correlation
of the signal of interest, and the nature of application.
In-network data aggregation itself has overheads due to energy spent on
computations. For example, in [23], the authors consider a clustered sensor
network in which measurements of all the nodes in a cluster can be aggregated
into a single measurement at the clusterhead. Thus instead of sending
multiple measurements, a single aggregated measurement is sent to the sink.
However this aggregation process requires extensive signal processing computations
at the clusterhead. The larger the number of nodes in a cluster,
the higher the extent of aggregation, and consequently, the higher the energy
savings on communication between the sensor nodes and the sink. However,
11 Chapter 1 Design of Large-scale Sensor Networks
Aravind Iyer et al.
the larger the number of nodes in a cluster, the larger the energy spent on
data aggregation computations. Thus there is an inherent tradeo?® between
communication and computation, and this tradeo?® is a characteristic feature
of many sensor networks.
4.2 Many-to-one Routing
As discussed in the previous section, sensor networks are characterized by
the many-to-one communication paradigm where all the sensor nodes wish to
send their data to a single sink node. This provides a sense of directionality
to the flow of data in the network, and results in ease of routing.
Pages:
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53