The Information Theoretic Approach to Signal Anomaly Detection for Cognitive Radio
Figure 1
(a) Spectrogram of a wireless local area network (WLAN) signal experiencing interference from a Bluetooth device. The regular pattern is a single WLAN frame repeatedly transmitted by a signal generator. The frequency hopping nature of the Bluetooth transmission is clearly visible in the plot. (b) Impact of the number of bins utilised, , on a histogram of the instantaneous power density of the WLAN signal. When is too small, the histogram is insensitive to small changes and does not effectively capture the subtleties of the process. At the other extreme, it is too sensitive and therefore susceptible to noise. The optimal is with respect to some minimum error criterion [1] and provides a good balance between resolution and sensitivity.
