Autors: Mihaylova, D. A., Iliev, G. L., Valkova-Jarvis, Z. V. Title: Methods for PCA Detection: An Efficiency Assessment Through Classification Metrics Keywords: classification metrics, pilot contamination attack, statistical evaluation Abstract: As a means to improve channel conditions for eavesdropping on the information exchange, an intruder can initiate a pilot contamination attack during the training phase for channel estimation. In this paper, the efficient operation of two methods for its detection - 2-N-PSK and Shifted 2-N-PSK - is assessed through classification metrics. When considering cybersecurity attacks, the contaminated and non-contaminated samples are usually not equally distributed, i.e. an imbalanced distribution of classes is the expected scenario. Hence, the methods' performance is evaluated by some of the most prominent classification metrics used for imbalanced distribution, namely the Precision-Recall Curve and its Area Under Curve. These metrics are studied in scenarios with different signal-to-noise-ratio and with different antenna array at the base station. The Receiver Operation Characteristic curve and Area Under Curve are also examined for varying numbers of incorporated antennae. References
Issue
|
Вид: публикация в международен форум, публикация в реферирано издание, индексирана в Scopus