基于GWO-SA-LSTM模型的调制识别算法
Modulation Recognition Algorithm Based on GWO-SA-LSTM Model
张逸凡 1雷斌 2苏晨 2刘光辉2
作者信息
- 1. 西安工业大学 兵器科学与技术学院,西安 710021
- 2. 西安工业大学 电子信息工程学院,西安 710021
- 折叠
摘要
针对传统自动调制识别方法对现代信号识别准确率较低的问题,该文提出了GWO-SA-LSTM模型,该模型通过自注意力机制增强长短期记忆网络(LSTM)关键特征捕获能力,并利用灰狼算法(GWO)优化超参数.在真实环境的调制识别数据集上的实验表明,该文模型在10 dB信噪比下达到了97%的最高准确率,整体表现优于单纯LSTM模型.
Abstract
Addressing the issue of traditional automatic modulation recognition methods'low accuracy for modern sig-nals,this paper introduces the GWO-SA-LSTM model.This model enhances the capability of long short-term memory(LSTM)networks to capture key features through a self-attention mechanism and utilizes the grey wolf optimizer(GWO)to optimize hyperparameters.Experiments conducted on a real-world modulation recognition dataset demon-strate that the proposed model achieves a highest accuracy of 97%at a 10 dB signal-to-noise ratio,outperforming the standalone LSTM model in overall performance.
关键词
无线电监测/调制识别/长短期记忆网络/灰狼算法/自注意力机制Key words
radio monitoring/modulation recognition/long short-term memory(LSTM)/grey wolf algorithm(GWO)/self-attention mechanism引用本文复制引用
出版年
2024
NETL
NSTL
维普
万方数据