INTRODUCTION

Uplink-Downlink Channel Prediction

Scaling the number of antennas up is a key characterstic of current and future wireless systems. Realizing the multiplexing and beamforming gains of large number of antennas requires channel knowledge. Usually channel feedback from users are used to get the channel knowledge. This results in signalling overhead and wastage of radio resources. Suppose we know the channels between a user and a certain set of antennas at one frequency band, can we map this knowledge to the channels at a different set of antennas and at different frequency band? Essentially this mapping means that we can directly predict the downlink channel gains from the uplink channel gains, eliminating the downlink training and feedback overhead in co-located/distributed FDD massive MIMO systems. Following [1], we use deep neural networks (DNN) to approximate the mapping between uplink and downlink channel gains, and the freely available Deep MIMO Dataset [2] will be used for training and testing.

Performance Evaluation of AutoEncoder

Following [3], we present a communication system (transmitter, channel and receiver) as an autoencoder, and design an end to end reconstruction task that seeks to jointly optimize transmitter and receiver components in a single process. We compare the performance of end to end autoencoder for five different schemes. The key idea here is to represent transmitter, channel, and receiver as one deep neural network (NN) that can be trained as an autoencoder.

All the details are given in report.pdf.

Code Details

All the scripts are in DeepMIMO folder. testing contains some notebooks where we did some testing of different ideas.

The configuration file is config.yaml. It has configurations for DNN (fnn), autoencoder and varitional autoencoder. The code uses pytorch for implementation.

To run DNN, use

python train.py -c config.yaml

To run autoencoder, use

python trainautoencoder.py

or (V2 is different architecture of AE we tried)

python trainautoencoder_v2.py

To run Variational Autoencoder, use

python train_vae.py -c config.yaml

References

[1] M. Alrabeiah and A. Alkhateeb, “Deep learning for tdd and fdd massive mimo: Map- ping channels in space and frequency,” arXiv preprint arXiv:1905.03761, 2019.

[2] A. Alkhateeb, “Deepmimo: A generic deep learning dataset for millimeter wave and massive mimo applications,” arXiv preprint arXiv:1902.06435, 2019.

[3] J. Xu, W. Chen, B. Ai, R. He, Y. Li, J. Wang, T. Juhana, and A. Kurniawan, “Perfor- mance evaluation of autoencoder for coding and modulation in wireless communica- tions,” in 2019 11th International Conference on Wireless Communications and Signal Processing (WCSP). IEEE, 2019, pp. 1–6.

Thesis Supervisor