In this paper, we present a deep learning based wireless transceiver. We describe in detail the corresponding artificial neural network architecture, the training process, and report on excessive over-the-air measurement results. We employ the end-to-end training approach with an autoencoder model that includes a channel model in the middle layers as previously proposed in the literature. In contrast to other state-of-the-art results, our architecture supports learning time synchronization without any manually designed signal processing operations. Moreover, the neural transceiver has been tested over the air with an implementation in software defined radio. Our experimental results for the implemented single antenna system demonstrate a raw bit-rate of 0.5 million bits per second. This exceeds results from comparable systems presented in the literature and suggests the feasibility of high throughput deep learning transceivers.

No abstract

No abstract

No abstract

Orthogonal Frequency Division Multiplexing (OFDM) while being an efficient scheme for high data rate wireless communications has drawbacks such as higher Peak-to-Average Power Ratio (PAPR). To reduce PAPR, use of multiple signal representation technique such as Partial Transmit Sequence (PTS) is one of the favored techniques. However, the use of conventional PTS technique need excessive number of complex calculations in order to search for all permissible combinations of phase sequences causing steep increase in complexity in terms of complex computations. Paper aims to reduce the cumbersome process of phase selection by making use of the similarity of the phase vectors. The phase vectors are obtained sequentially and thus minimize the number of changes from one phase vector to another. Theoretical analysis shows that computational complexity is significantly reduced with the help of this proposed novel technique. We have also demonstrated that PAPR values are similar i.e. PAPR reduction capability remains similar but at reduced complexity.