2019 | OriginalPaper | Buchkapitel
Deep Learning-Based Space Shift Keying Systems
Autoren: Yue Zhang, Xuesi Wang, Jintao Wang, Yonglin Xue, Jian Song
Erschienen in:
Artificial Intelligence for Communications and Networks
Abstract
To handle the performance degradation of space shift keying (SSK) systems under practical non-Gaussian channels, we propose a deep neural network model in which an auto-encoder (AE) is developed to design proper constellations and corresponding demodulation. With full knowledge of channel statistics, the transmitter and receiver are jointly optimized in our scheme. By representing the SSK system as an AE, we consider the cross-entropy loss function for antenna index and formulate the overall pipeline using deep learning techniques. Moreover, our implementation can be adopted in several noise conditions successfully. Results confirm that our model outperforms the maximum likelihood (ML) detection scheme in terms of block error rates (BLER).
