Deep Learning Based Prediction of Channel Profile for LTE and 5G Systems

Abstract:
Channel profiles or power delay profiles (PDP), representing multipath fading propagation conditions, are used for linklevel performance evaluation and optimization. Channel state information (CSI), acquired via channel estimation techniques, lumps all multipath components and hence contains no delay profiles. This paper presents a Deep Learning (DL) based methodology to categorically predict LTE and 5G channel profiles (i.e., [EPA, EVA, ETU] and [TDL-A, TDL-B, TDL-C, TDL-D, TDL-E], respectively). MATLAB simulations of single-antenna LTE and 5G systems are conducted with the resultant prediction accuracy of '99% and '96%, respectively, at a latency of five milliseconds. Simulations are configured on random payloads (e.g., non data aided - NDA), time-domain and frequency-domain signals embedded with variable modulation types [QPSK, 16QAM, 64QAM], Doppler shifts [0, 50, ..., 550] Hz, and signal-to-noise ratios (SNRs) [-10, 20] dB. The methodology utilizes a hybrid model of a Convolutional Neural Network (CNN) and a Long Short Term Memory (LSTM) and two techniques to enhance prediction accuracy, namely, input diversity and binary prediction.