A Novel Combination of PCA and LSTM for Multi-Carrier Modulation Signals Classification

Fast communication is essential in today's world, and signal demodulation is a key factor affecting latency. Traditional demodulation methods are complex and time-consuming, making Automatic Modulation Classification (AMC) crucial for modern communication systems, particularly in wireless communication and signal processing. This paper presents a novel approach to classify five types of Multicarrier Modulation (MCM) signals. It also analyzes two subcarrier waveforms for each signal type, resulting in 10 unique MCM signals. This paper is divided into 2 parts -feature extraction from signals and classification of signals using those extracted features. To extract relevant features, Principal Component Analysis (PCA) and Recursive Feature Elimination (RFE) are applied. A Long Short-Term Memory (LSTM) network is then used to classify the MCM signals. Our results show that the combination of PCA and RFE. LSTM achieves superior performance in categorizing the MCM signals, making it a promising approach for efficient and accurate AMC. This paper is classifying MCM signals of range - 16dB to 16dB with the accuracy of 91% to 97%.