Distribution of 50 dominating frequencies.
Frequency Range Lucent D-Link Linksys
0-10 54% 8% 2%
40-50 - 28% -
50-60 4% 26% -
80-90 - - 10%
90-100 12% 8% 34%
100-110 10% 4% 8%
120-130 - - 4%
140-150 - 2% 2%
150-160 - 6% -
190-200 20% 4% 16%
200-210 - 12% 24%
220-230 - - -
230-240 - 2% -
of 50 frequency points as a spectral profile proved to be adequate, because the
distribution of the set F coincides with the observations made using NCS.
9 Conclusion
In this chapter we presented an approach for identifying unauthorized sensor
nodes within a network. We identified the rate switching algorithm as an
attribute of an RI as an opportunity for distinguishing between RIs manufactured
by di?®erent vendors. We used signal processing to extract markedly
di?®erent tra?±c characteristics among di?®erent RIs during rate switching.
An important aspect to explore is the stability of the spectral profile. We
applied our technique and successfully tracked nodes in a real network. Other
factors like the setting of the RI configuration parameters can be considered
(i.e., RTS threshold, maximum retries, etc.). If the settings change the spectrum,
then they become an essential part of the spectral profile. This could
help distinguish between RIs manufactured by the same vendor. Heterogeneous
sensor systems should also be considered. During our experiments we
only used a single type of sensor. Sensors with di?®erent capacities may also
a?®ect the spectral profile.
Pages:
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680