Information Technology Reference
In-Depth Information
Assessment of Related Work
results in a significantly increased stress of re-
sources. All approaches poorly perform with
particular regard to energy consumption and cost-
efficiency for collaboration. There is no doubt
that a robust application requires an overhead but
the efficiency of an application significantly de-
pends on a proper balance between the application
requirements and the implementation. Further,
most approaches shift final decisions to central-
ized nodes and are vulnerable if these nodes are
faulty or fail completely. The existence of backup
nodes, which can substitute the task of these
centralized nodes if necessary, enhances the ro-
bustness against node failures but significantly
increases the effort. However, only a fully decen-
tralized approach will provide a proper autonomy
for the sensor nodes.
It is quite obvious that fulfilling all criteria up
to a level of 100 percent is almost impossible but
existing approaches usually tackle only a subset
of those. The event detection concept presented
here is inspired by some ideas of the discussed
approaches and combines these in a new suit-
able event detection scheme tackling all design
criteria. This chapter introduces a novel concept
for autonomous sensor network configuration
considering all mentioned criteria. Based on the
description of the phenomenon to be recognized,
the respective event detection is autonomously
For comparison, presented event detection
schemes are evaluated against the introduced
criteria in Table 1. First of all, there exists no
approach that addresses and fulfils all criteria.
Krishnamachari and Krasniewski provide the
best robustness and even enable to autonomously
cope with malicious nodes but increase the
required overhead by at least a factor of three.
Because detection reliability is a prerequisite,
all approaches allow for adaptation to changing
conditions but usually focus on certain changes
only, e.g., faults, malicious nodes, unavailable
resources, environmental changes, connectivity
etc. Except for CoDED, which unfortunately was
not implemented so far, all approaches carelessly
neglect the autonomy required in a sensor network.
Fully distributed concepts not depending on spe-
cial nodes are in great demand, which by design
enables all nodes to fulfill every task required for
event detection in prevention of SPoFs.
To summarize, robustness and transparency
are best provided, whereas energy efficiency,
autonomy and convenience are marginally taken
into account or are even completely missing. There
exists no approach that associates all introduced
criteria. As shown in existing work, providing
high robustness is possible indeed but partially
Table 1. Comparison of discussed event detection schemes in consideration of the introduced design
criteria. There exists no approach that addresses and fulfils all criteria. Best provided is robustness and
transparency, whereas energy efficiency, autonomy and convenience are marginally taken into account
or are even completely missing.
Robustness
Autonomy
Transparency
Energy efficiency
Convenience
Vu
(+)
(-)
(0)
(0)
(?)
Phani Kumar
(+)
(-)
(+)
(--)
(?)
Krishnamachari
(+)
(-)
(+)
(--)
(-)
Krasniewski
(++)
(0)
(0)
(--)
(?)
Kamiya
(0)
(--)
(+)
(--)
(0)
Schwiderski
(?)
(+)
(0)
(-)
(-)
Validation: (++) very good; (+) good; (0) regular; (-) bad; (--) very bad; (?) not mentioned
Search WWH ::
Custom Search