Self-Supervised User Positioning in Cell-Free Massive MIMO Systems
New Note 1
Massive multiple-input multiple-output (MIMO) systems have been extensively investigated in the past decade as one of the most attractive solutions for the fifth generation (5G) of wireless systems. One of the grand challenges in such systems is user positioning, which is critical for emergency services, area-specific advertisements, content caching, and autonomous driving [1,2]. Global navigation satellite systems (GNSS) are being extensively used to provide this kind of information. However, the base station does not have access to GNSS information and satellite-based localization is unreliable indoors and in dense rural scenarios. As a consequence, current research focuses on developing new strategies to use wireless signals to extract positioning data . Emerging approaches that have been proposed for massive MIMO and distributed massive MIMO (DM-MIMO) systems are channel charting , fingerprinting techniques , and triangulation/trilateration . Fingerprinting techniques based on the received signal strength (RSS) and Gaussian process regression (GPR) have been applied to user positioning, but require extensive training with labeled data . Triangulation and trilateration as described in , estimate the user’s position by jointly processing the RSS or time-of-flight information, but require line-of-sight connectivity to at least three base stations. Channel charting, as put forward in , captures the local spatial geometry of an area by extracting relative location information directly from channel state information (CSI). The method collects the data from multiple users over time and generates the channel charts in a self-supervised fashion using machine learning techniques. While all of these methods have been proposed for conventional cellular wireless systems, not much is known about their efficacy for cell-free wireless networks, which consists of a very large number of access points (AP) distributed over a large area, communicating via a centralized processor. Having access to the extreme amount of measurements at the distributed APs has the potential to significantly increase localization accuracy in indoor and rural scenarios, while avoiding the need of labeled training data. This project will investigate self-supervised user positioning in cell-free massive MIMO systems using channel charting. More specifically, we will investigate scalable algorithms that enable accurate user localization from CSI using machine learning techniques that do not require labeled data (i.e., ground truth location information). The goal is to achieve meter-level accuracy in indoor and rural scenarios in a purely data-driven fashion. The techniques to be learned in this project include wireless communication, dimensionality reduction, deep neural networks, and manifold learning.
 V. Savic and E. G. Larsson, "Fingerprinting-Based Positioning in Distributed Massive MIMO Systems," 2015 IEEE 82nd Vehicular Technology Conference (VTC2015-Fall), Boston, MA, 2015, pp. 1-5, doi: 10.1109/VTCFall.2015.7390953.  K. N. R. S. V. Prasad, E. Hossain and V. K. Bhargava, "Machine Learning Methods for RSS-Based User Positioning in Distributed Massive MIMO," in IEEE Transactions on Wireless Communications, vol. 17, no. 12, pp. 8402-8417, Dec. 2018, doi: 10.1109/TWC.2018.2876832.  C. Studer, S. Medjkouh, E. Gonultaş, T. Goldstein and O. Tirkkonen, "Channel Charting: Locating Users Within the Radio Environment Using Channel State Information," in IEEE Access, vol. 6, pp. 47682-47698, 2018, doi: 10.1109/ACCESS.2018.2866979.  N. Garcia, H. Wymeersch, E. G. Larsson, A. M. Haimovich, M. Coulon, “Direct Localization for Massive MIMO,” IEEE Trans. Signal Processing, vol. 65, no. 10, pp. 2475-2487, May 2017.
- Looking for 1-2 Semester/Master students
- Contact: Christoph Studer
- Communication Systems (recommended)
- 20% Literature Research
- 80% System Development
Detailed Task Description
Describe your new note here.