Realization of an Artificial Neural Network Demodulator PoC

Celestia Antwerp Celestia Antwerp is a leading company in Belgium, active in the field of satellite ground communication technology. It is part of the Celestia Technologies Group.
Celestia Antwerp’s core business is in the delivery of satellite communication solutions for the ground segment extension with products and systems. Celestia Antwerp operates both on the commercial and institutional markets, supporting space agency programs, as well as industrial company projects worldwide. The company activities have been located in Antwerp since 1962, when it was founded as part of Bell Telephone.

Type and Description

Type: Internship

Description: The candidate will join the department developing satellite communication ground products such as the Satellite High-Speed Data Receiver (OMNISAT) for earth observation, the Integrated BaseBand system (IBB) for tracking and command (TT&C). He/she will be in charge of introducing artificial intelligence algorithms in the IBB product to increase the data speed. About the candidate: The candidate shall have strong DSP knowledge applied to wireless communications. Knowledge in artificial intelligence techniques is considered as an asset.

Objective and Deliverables

The objective of this internship is to design & develop (in C/C++) an artificial neural networks (ANNs) (O)QPSK demodulator. ANNs are known to be a good approximation of nonlinear functions, and can therefore be used for LLR calculation, non-linear equalization etc. It is expected that ANNs will reduce the complexity compared to traditional algorithms, so that higher data rates can be targeted. A dedicated chipset or GPU could be used to run the ANNs Tasks to perform are: - ANN state of the art analysis for DSP, in particular high-speed demodulator
- Scenario definition and consolidation
- Design and implementation of the ANN learning methods, including definition of the learning method
- Performance comparison between ANNs and traditional algorithms
- Suggestions for performance optimization
- Testing on a real-time testbed
More information about the IBB can be found at

Expected deliverables

The following deliverables are expected:
• Detailed Design Description documentation
• Scripts (source code and user manual)
• C++ library implementing the ANNs
• Final Report / Presentation


3-6 months.

Evaluation Criteria

The following will be taken into account for the final evaluation:
• Demonstrated technical skills
• Timeline of deliverables vs agreed plan
• Quality of deliverable
• Level of autonomy shown by the candidate
• Motivation and initiatives taken


Berchem (Antwerp), Belgium

More Information about Celestia Antwerp and other potential internship or open vacancies:

Interested in this internship?

Would you be interested in this internship or would you like to get more information about this internship, please take contact with