Public defence in Signal Processing Technology, M.Sc. Kameswara Atchutaram Kocharlakota

The title of the thesis: Resource optimization for massive MIMO systems

Thesis defender: Kameswara Atchutaram Kocharlakota
Opponents:
Prof. Gayan Baduge, Southern Illinois University, US
Prof. Angelo Coluccia, University of Salento, Lecce, Italy
Custos: Prof. Sergiy Vorobyov, Aalto University School of Electrical Engineering, Department of Information and Communications Engineering 

The doctoral research explores critical challenges in the implementation of advanced wireless communication systems, specifically massive multiple-inout multiple output (MIMO) and Cell-Free Massive MIMO (CFmMIMO), technologies pivotal to 5G and beyond wireless networks. The study investigates how pilot contamination impacts Spectral Efficiency (SE) in Massive MIMO systems and develops innovative analytical tools and machine learning-based solutions to enhance system performance. 

Massive MIMO systems leverage large antenna arrays to provide unparalleled SE, but their performance is hindered by pilot contamination, particularly when pilot sequences are reused across cells. The research derives closed-form SE expressions that account for imperfect covariance matrix estimation, shedding light on the trade-offs between pilot overhead and estimation accuracy. This analysis enables more informed strategies for balancing pilot resources with overall system efficiency. 

In CFmMIMO systems, where a network of distributed antennas collaboratively serves users, achieving fairness in downlink power control is computationally challenging due to the non-convex nature of the problem. The thesis introduces a cutting-edge solution utilizing transformer neural networks, inspired by GPT, BERT, and Generative AI (Gen AI). These attention-based architectures effectively address the power control problem, offering scalability and robustness while significantly reducing computational costs compared to traditional optimization methods. 

By addressing these challenges, the research advances the theoretical understanding and practical application of Massive MIMO and CFmMIMO systems. The findings are crucial for optimizing SE, improving resource allocation, and enabling reliable and efficient next-generation wireless networks. The methodologies and insights presented provide a roadmap for overcoming critical barriers in deploying 5G and beyond technologies, positioning the research at the forefront of wireless innovation.

Keywords: Massive Multiple-Input Multiple-Output, Cell-Free Massive Multiple-Input Multiple-Output (CFmMIMO), Spectral Efficiency (SE), Pilot Contamination, Covariance Matrix Estimation, Downlink Power Control, Transformer Neural Networks, Generative Pre-trained Transformer (GPT), Bidirectional Encoder Representations from Transformers (BERT), Generative Artificial Intelligence (Gen AI), Resource Allocation, 5G and Beyond Wireless Networks

Thesis available for public display 10 days prior to the defence at: https://aaltodoc.aalto.fi/doc_public/eonly/riiputus/

Contact:

Doctoral theses in the School of Electrical Engineering: https://aaltodoc.aalto.fi/handle/123456789/53