Public defence in Communications Engineering, M.Sc. Parham Kazemi

The title of the thesis: Channel charting-based radio resource management

Thesis defender: Parham Kazemi
Opponent: Dr. Zoran Utkovski, Frauhofer Heinrich-Hertz Institute, Germany
Custos: Prof. Olav Tirkkonen, Aalto University School of Electrical Engineering

5th Generation (5G) cellular networks are designed to deliver unparalleled performance in mobile environments with three promises: i) increased capacity, ii) ultra-reliable and low-latency connections, and iii) a massive number of connected devices. Achieving these goals necessitates the integration of novel technologies. Millimeter-wave (mmWave) communications enable ultrahigh data rates and low latency, leveraging substantial bandwidth at high frequencies. Beamforming techniques are used in mmWave bands to alleviate path loss. However, challenges exist, primarily the high overhead of finding suitable beams. This thesis addresses key challenges in beam management for 5G and mmWave systems using Channel Charting (CC) and Machine Learning (ML). CC is a self-supervised method that maps high-dimensional Channel State Information (CSI) at a Base Station (BS) into a low-dimensional space representing pseudo positions of User Equipment (UEs). The low-dimensional space preserves the local geometry of UEs, meaning nearby UEs in real space are close on the CC. An offline training phase constructs and annotates CCs with Signal-to-Noise Ratios (SNRs) of neighboring cells/beams. ML algorithms predict the SNR of a user at neighboring cells/beams from its transmission in a massive Multiple Input Multiple Output (mMIMO) system. Predicting signal quality without UE assistance reduces protocol overhead for handover decisions. Both standalone and non-standalone 5G deployments are considered, and the best beam prediction is investigated. Beam tracking based on CC shows that with low beam-search overhead, one can leverage a CC-to-SNR mapping to track strong beams between UEs and the BS. Enhancements in CC construction enable versatile applications across different scenarios. A CSI feature mitigates small-scale fading, yielding robust predictions even with low spatial sampling density. A low complexity Out-of-Sample (OOS) algorithm reduces computational requirements compared to conventional OOS algorithms, making it efficient for practical implementations.

Keywords: Channel charting, beam management, SNR prediction

Thesis available for public display 10 days prior to the defence at Aaltodoc. 

Contact:
email: parham.kazemi@aalto.fi 
phone number: +358504760455 
linkedin: https://www.linkedin.com/in/parham-kazemi-780742121/