| Abstract: | Automatic Modulation Classification (AMC) is responsible for detecting the correct modulation types in the intelligent receivers. AMC performance degrades when the signal-to-noise ratio (SNR) decreases because of the overlapping among the digital modulation types’ features, and this performance worsens under fading channel conditions. This paper proposes two new algorithms that improve the AMC performance accuracy of the overlapped digital modulations in feature space by improving their discrimination. These algorithms are named temporal Fisher discriminant analysis (TFDA) and supervised Fisher discriminant analysis (SFDA). The simulation results show that TFDA improves AMC performance accuracy up to 19.01% compared with the reference paper (Ge et al., 2021) and up to 38.15% compared with the reference paper (Teng et al., 2018). In contrast, SFDA improves AMC performance accuracy up to 23.12 % compared with the reference paper (Ge et al., 2021) and up to 49.025% compared with the reference paper (Teng et al., 2018). |
