Fig. 7From: Robust localization in wireless networks from corrupted signalsThe weights in the robust method can be used to classify samples \({\mathcal {C}}(\varvec{z}) \in \{0,1 \}\) as ‘normal’ or ‘corrupted’. This yields probabilities of correct detection \(\Pr \{{\widehat{{\mathcal {C}}}}(\varvec{z})=1|{\mathcal {C}}(\varvec{z}) = 1\}\) and false alarm \(\Pr \{{\widehat{{\mathcal {C}}}}(\varvec{z})=1|{\mathcal {C}}(\varvec{z}) = 0\}\) shown in the figure as the corruption fraction \(\epsilon\) is varied from 10 to 50% \(= {\widetilde{\epsilon }}\)Back to article page