End-to-End Learning of Communication Systems with Novel Data-Efficient IIR Channel Identification
Sprache des Vortragstitels:
Englisch
Original Tagungtitel:
Asilomar Conference on Signals, Systems and Computers (ACSSC 2023)
Sprache des Tagungstitel:
Englisch
Original Kurzfassung:
In this work, a novel end-to-end deep learning procedure for communication systems is introduced, which is data efficient and capable of dealing with infinite memory length of communication channels. Therefore, as opposed to recent works, a low-complexity algorithm is utilized to identify the communication channel. The channel model is obtained purely from data and its output is differentiable with respect to its input, which is a basic requirement for gradient-based optimization of an auto-encoder neural network implementing an end-to-end optimized communication system. The performance of the algorithm is studied for a variety of challenging channels from different domains of communication engineering showing the broad applicability of the proposed approach.
Forschungs-