Neural Network-Based Jamming Detection and SINR Estimation for Linearly Modulated Signals

Jamming of wireless communication signals is a widely considered problem in performance-critical applications. Jamming often refers to wanted (intentional) or unwanted (accidental) interference which arises systematically and reduces signal to noise ratio (SNR) significantly. Multiple approaches of jamming detection has been carried out in the last decades. Our contribution in this paper is the development of a new combined multi-feature detection approach which is realized by means of neural networks. We evaluated different neural network types concerning their jamming detection and SNR estimation performance. Moreover we show how to minimize the number of neurons in order to trade off computational complexity and detection performance.