Database Reference
In-Depth Information
Figure 7.1.
Value and time tolerance, assuming a linear model (depicted by the thick
dashed line) for the sensed data [79].
T
such that
all
the values reported in this interval are incorrect, i.e.,
V
i
∈ R
V
,
∀
V
i
∈
V
T
.
Similarly to the model-driven approach, each node (or group of nodes)
in the WSN generates a model for the sensed data. This model is then
sent to the sink, along with the last reading. From that point on, the
sink can predict the readings of the node based on this shared model.
The node is also checking whether its model can accurately describe its
own readings (within the error tolerance agreed with the sink), and if
this is not true then it computes a new model and transmits it to the
sink. Evidently, the sink always records accurate data (i.e., within
ε
VT
),
regardless of the quality of the model. The model quality affects only
the effectiveness of the proposed scheme in terms of energy savings.
We can now formally define the problem of data-driven data acquisi-
tion.
Problem 2.2 (Data-Driven Data Acquisition)
Given a sensor
network, and a sink that needs to collect all the sensed values within
ε
VT
, design a data collection protocol such that the energy used by the
sensor network is minimized.
This problem statement is deliberately vague on the specificities of
the design of such a protocol. In the following paragraphs we review
several techniques that solve this problem, each one focusing on different
aspects of the problem. Some studies focus on the selection of the sensed
data model (shared among sensors and sink), others concentrate on the
effective identification of temporal and/or spatial correlations among the
