Join us!

Location: Lund, Sweden

Preferred starting date: Jan. 2025

Extent: 1-2 students, 30hp.

Thesis description

Multiple input multiple output (MIMO) systems play a very important role in modern wireless communications. Versatile approaches of MIMO detection algorithms have been developed for MIMO systems such as in the fifth generation new-radio (5G-NR) to achieve a good trade-off between complexity and performance. However, the performance of MIMO detector highly depends on the detection algorithms, and the accuracies of estimated channel, interference and noise characteristics. On the other hand, AI-based methods are particularly well suited for situations characterized by algorithmic deficits, i.e., for scenarios in which optimal algorithms are unknown or with prohibitive. It is known that Graph neural network (GNN) and Transformer architectures may be good candidates for MIMO detection, in combination with super-resolution and convolutional neural network (SR-CNN) based channel estimate (CE).

Project Scope

In this master thesis project, we will investigate the state of art of deep learning methods by applying GNN and/or Transformer to MIMO detection, and evaluate the performance and complexity compared to conventional MIMO detection. The student(s) will conduct research under supervisions, including the following main tasks:

• Implement basic GNN/Transformer based MIMO detection in OFDM system.
• Generate train labels, train networks, and perform tests under different conditions.
• Analyze the accuracy of learning and sensitivity of model to mismatches.
• Optimize NN architecture to achieve the best performance, and provide complexity analysis in terms of the number of trainable parameters and FLOPs.

In this project, the candidates will gain a comprehensive understanding of system-level wireless communications and hands-on AI’s applications to signal processing. Both ability and skills of coding and debugging will be improved.

Qualifications

• Master student in Mathematics, Computer Science, Electrical Engineering or equivalent.
• Theoretical background in areas such as wireless communications, AI, and optimization.
• Hands-on experience in Python, MATLAB, and libraries such as Tensorflow/Pytorch.

Contact person

Please send your detailed Resume/CV and transcript of records in English via e-mail to Chris Qin (chrisqin@huawei.com）

The e-mail should be titled “AI Receiver for Next Generation Wireless Communication

+ your name”.