Open AccessEnd‐to‐End Multi‐Domain and Multi‐Step Jamming Prediction in Wireless Communications
Author(s) -
Su Zhe,
Qi Nan,
Jia Luliang,
Chen Jiaxin,
Liu Yijia,
Sun Wen
Publication year - 2021
Publication title -
electronics letters
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.375
H-Index - 146
eISSN - 1350-911X
pISSN - 0013-5194
DOI - 10.1049/ell2.12128
Subject(s) - computer science , robustness (evolution) , jamming , wireless , artificial intelligence , convolutional neural network , data modeling , deep learning , end to end principle , machine learning , algorithm , data mining , telecommunications , biochemistry , chemistry , physics , database , gene , thermodynamics
In this letter, the problem of jamming data prediction in wireless communications is investigated. Both time and frequency domains are considered to construct the multi‐domain historical jamming data tensor. Besides, due to the perceiver's limited ability, false alarm data and missing detection data are considered. Two neural network prediction models are proposed to predict the jammers' future actions based on deep learning techniques. One is the multi‐variate long‐short‐term‐memory (multi‐variate LSTM) model, and the other is the 2‐D convolutional long‐short‐term‐memory model. Simulation results show that the proposed models have better prediction accuracy and robustness than the benchmark method.