Master Thesis - Information Theory Enhanced Channel Charting

The Fraunhofer-Gesellschaft (www.fraunhofer.com) currently operates 76 institutes and research institutions throughout Germany and is the world’s leading applied research organization. Around 30 000 employees work with an annual research budget of 2.9 billion euros.

You are interested in bridging information theory and machine learning and you think localization is exciting?

Motivation. The demand for increased radio communication and number of terminals thereof, in future wireless communication systems, requires the use of new technologies, e.g., massive multiple-input / multiple-output, millimeter-wave bands that rely on knowledge of transmitter locations to solve the challenges encountered in their practical application [1]. In contrast to supervised wireless localization frameworks such as fingerprinting, Channel Charting (CC) is system-level unsupervised and operates directly on Channel State Information (CSI) that are passively collected at multi-antenna base stations (BS). As such, it avoids the expensive data collection phase of their supervised counterparts. While promising results in simulated, static environments have demonstrated the potential feasibility of the method [1, 2, 4], generalization and robustness in complex, dynamic, and realistic environments with potentially sparse UE coverage have not yet been addressed in public literature. And beyond that, the existing evaluation criteria of potential CC techniques essentially rely on qualitative inspection and thus, suffer from a lack of well-calibrated and understood performance metrics. For example, trustworthiness (TW) and continuity (CT) impede the automatic comparison required for hyperparameter optimization.

Two core ideas to address these challenges. To improve the robustness and generalizability of CC, we should examine Siamese or Triplet Neural Networks (SNN) in detail. SNNs allow fusion of absolute position information with locally accurate CSI similarities to locate UEs for which no position is known. Further improvements of the method may be achievable by integrating knowledge of the environment using location density preferences. We may use prior knowledge of the environment that is implicitly derivable from available maps around a BS and impose preferences on the distribution of UEs in the CC. For instance, busy streets have higher UE density than a remote industrial area. Finding appropriate loss functions that reflect such preferences in a differentiable, numerically stable way represents a new, promising research direction for channel charting. Ultimately, complex location density preferences should be expressible, e.g., via 2D-heatmaps.

What you will do

You define, implement and evaluate loss functions that express location density preferences for channel charting
You study and propose more efficient evaluation metrics (TW, CT, alternatives)
You will write an academic paper

What you bring to the table

You study Information Technology, Data Science or equivalent in a master program
You have programming experience with Machine Learning in Python
You hold preferably experience in working with common libraries such as numpy, pandas, matplotlib
You are ideally experienced with PyTorch or Tensorflow
You have preferably knowledge on radio communication

What you can expect

Open and friendly team work
Networking with scientists
Active contribution in applied research
Interesting an innovative projects
Mentoring program »josephine®« for talented female students

If you have any questions about this opening, please contact: Christian Nickel (christian.nickel@iis.fraunhofer.de).

The thesis will be assigned and carried out in accordance with the rules of your university. For this reason, please discuss the thesis with a professor who can advise you over the course of the project. The duration for the thesis should be 6 months.

We value and promote the diversity of our employees' skills and therefore welcome all applications - regardless of age, gender, nationality, ethnic and social origin, religion, ideology, disability, sexual orientation and identity. Severely disabled persons are given preference in the event of equal suitability.

With its focus on developing key technologies that are vital for the future and enabling the commercial utilization of this work by business and industry, Fraunhofer plays a central role in the innovation process. As a pioneer and catalyst for groundbreaking developments and scientific excellence, Fraunhofer helps shape society now and in the future.

Interested? Apply online now (PDF: Cover Letter, CV, transcripts). We look forward to getting to know you!

Maximilian Müller

Fraunhofer Institute for Integrated Circuits IIS

www.iis.fraunhofer.de

Requisition Number: 29004 Application Deadline: None

References
[1] Studer, C., Medjkouh, S., Gonulta¸s, E., Goldstein, T., & Tirkkonen, O. (2018). Channel charting: Locating users within the radio environment using channel state information. IEEE Access, 6(?), pp. 47682-47698.
[2] Deng, J., Medjkouh, S., Malm, N., Tirkkonen, O., & Studer, C. (2018, October). Multipoint channel charting for wireless networks. In: Asilomar Conf. on Signals, Systems, and Computers (Asilomar), pp. 286-290.
[3] Lei, E., Casta˜neda, O., Tirkkonen, O., Goldstein, T., & Studer, C. (2019, September). Siamese neural networks for wireless positioning and channel charting. In: Annual Allerton Conf. on Communication, Control, and Computing (Allerton), pp. 200-207.
[4] Agostini, P., Utkovski, Z., & Sta´nczak, S. (2020, May). Channel charting: an Euclidean distance matrix completion perspective. In: Intl. Conf. on Acoustics, Speech and Signal Processing (ICASSP), pp. 5010-5014.
[5] Pihlajasalo, J., Koivisto, M., Talvitie, J., Ali-Loeytty, S., & Valkama, M. (2020). Absolute Positioning with Unsupervised Multipoint Channel Charting for 5G Networks. In IEEE Vehicular Technology Conference (VTC2020-Fall) (pp. 1-5). IEEE.

Job Segment: Network, Telecom, Telecommunications, Inspector, Technology, Research, Quality

Apply now »