Applying Neural Networks in Optical Communication Systems: Possible Pitfalls

By Tobias A. Eriksson, Henning Bülow, Andreas Leven. 2017

143 citations recorded by the source corpus.

We investigate the risk of overestimating the performance gain when applying neural network-based receivers in systems with pseudorandom bit sequences or with limited memory depths, resulting in repeated short patterns. We show that with such sequences, a large artificial gain can be obtained, which comes from pattern prediction rather than predicting or compensating the studied channel/phenomena.

Subfields: Artificial Intelligence, Electrical and Electronic Engineering

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Ala Al-Fuqaha
2. Mohsen Guizani
3. Sameh Sorour
4. Tobias A. Eriksson
5. Henning Bülow
6. M. Mohammadi
7. Ekram Hossain
8. Sanjay Chawla

In Electrical and Electronic Engineering

1. Ala Al-Fuqaha
2. M. Mohammadi
3. Mohsen Guizani
4. Sameh Sorour
5. Laurent Schmalen
6. Polina Bayvel

Citations in the corpus

• Deep Learning for IoT Big Data and Streaming Analytics: A Survey (2017, background): “[181] showed that using pseudo random bit sequences or short repeated sequences can lead to overestimating the signal-to-noise ratio.”
• End-to-End Deep Learning of Optical Fiber Communications (2018, methodology): “In the experimental validation, we generated a long random sequence (not a PRBS, as suggest in [12]) which is processed by the transmitter ANN to generate a waveform, loaded (after filtering and resampling) into the DAC, and transmitted multiple times, to capture differ-”