Automatic Modulation Classification for OFDM Systems Using Bi-Stream and Attention-Based CNN-LSTM Model

Existing deep learning (DL) models for automatic modulation classification (AMC) of orthogonal frequency division multiplexing (OFDM) signals consider only spatial or temporal attributes, ignoring feature interaction, which limits their ability to classify higher-order modulation formats. To address...

Full description

Saved in:
Bibliographic Details
Published in:IEEE communications letters 2024-03, Vol.28 (3), p.552-556
Main Authors: Kumar, Anand, Chaudhari, Mahesh Shamrao, Majhi, Sudhan
Format: Article
Language:English
Subjects:
Artificial neural networks
attention
Automatic modulation classification
bi-stream
Carrier frequencies
Classification
CNN
Convolution
Convolutional neural networks
deep learning
Feature extraction
LSTM
Machine learning
Mathematical models
Modulation
OFDM
Orthogonal Frequency Division Multiplexing
Phase shift keying
Quadrature amplitude modulation
Signal to noise ratio
Citations: Items that this one cites
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Existing deep learning (DL) models for automatic modulation classification (AMC) of orthogonal frequency division multiplexing (OFDM) signals consider only spatial or temporal attributes, ignoring feature interaction, which limits their ability to classify higher-order modulation formats. To address this, we propose a bi-stream and attention-based convolutional neural network and long short-term memory network (CNN-LSTM) for AMC. It can efficiently extract special and temporal features from the in-phase and quadrature (IQ) samples and the amplitude and phase (AP) of the received signal for AMC. Each stream consists of a CNN, LSTM, and attention module to extract features from signal patterns, which improves the classification performance. The proposed AMC can classify M-ary phase-shift keying (M-PSK) and higher-order M-ary quadrature amplitude modulation (M-QAM) formats in the presence of randomized carrier frequency offset, symbol timing offset, phase offset, and unknown channel state information. The proposed model outperforms existing models in terms of classification accuracy and complexity. Finally, the proposed model has been validated on the real-time dataset generated by the radio frequency testbed.
ISSN:1089-7798
1558-2558
DOI:10.1109/LCOMM.2023.3348512