Pilot Pattern Design for Deep Learning-Based Channel Estimation in OFDM\n Systems

2020/03/19 by Mehran Soltani, Soltani, Mehran, Vahid Pourahmadi +3 · 1 citation
Computer Science · Engineering · #Wireless Signal Modulation Classification #Error Correcting Code Techniques #Wireless Communication Security Techniques

paper · pdf · doi:10.48550/arxiv.2003.08980

Abstract

In this paper, we present a downlink pilot design scheme for Deep Learning\n(DL) based channel estimation (ChannelNet) in orthogonal frequency-division\nmultiplexing (OFDM) systems. Specifically, in the proposed scheme, a feature\nselection method named Concrete Autoencoder (ConcreteAE) is used to find the\nmost informative locations for pilot transmission. This autoencoder consists of\na concrete layer as the encoder and a multilayer perceptron (MLP) as the\ndecoder. During the training, the concrete layer selects the most informative\npilot locations, and the decoder reconstructs an approximate estimation of the\nchannel. Eventually, the ChannelNet is trained on the output of the ConcreteAE\naiming to reconstruct the ideal channel response. The estimation error results\nshow that this approach outperforms the previously presented ChannelNet with a\nuniformly distributed pilot pattern, and its performance is comparable to the\nminimum mean square error (MMSE).\n

Cited by

• Uncoordinated Cooperative OFDM Multi-Hop UAV Relay Networks Using Virtual Channels Based on All-Pass Filters

Related

• Power of Deep Learning for Channel Estimation and Signal Detection in OFDM Systems
• Pruning the Pilots: Deep Learning-Based Pilot Design and Channel\n Estimation for MIMO-OFDM Systems
• ComNet: Combination of Deep Learning and Expert Knowledge in OFDM Receivers
• Low Complexity Channel estimation with Neural Network Solutions
• Deep Learning Based OFDM Channel Estimation Using Frequency-Time Division and Attention Mechanism