Open Collections

Abstract

This thesis investigates the implementation of flexible functional splits in Cell-Free Massive MIMO (CF-mMIMO) systems under the Open Radio Access Network (O-RAN) framework. CF-mMIMO is a promising architecture for next-generation wireless networks, offering enhanced spectral efficiency (SE), energy efficiency (EE), and consistent service quality. With rising demands from smart devices, the Internet of Things (IoT), and immersive applications like augmented and virtual reality (AR/VR), scalable and efficient wireless solutions are essential. However, integrating CF-mMIMO into O-RAN introduces challenges, including high fronthaul bandwidth due to centralized coordination, increased computational burden at distributed units, and complexity in selecting functional splits that balance latency, power consumption, and processing load. The O-RAN architecture disaggregates conventional baseband processing into standardized functional splits (Split 1 through Split 8), allowing flexible allocation of tasks between centralized and distributed units. This thesis focuses on Split 7.2, which places partial baseband processing at the Radio Unit (O-RU), and Split 8, which centralizes all processing at the Central Unit (O-DU/O-CU). A simulation framework is developed to assess SE, EE, and Jain’s Fairness Index (JFI) across various access point (AP) and antenna configurations using a static user model and coherence block assumptions. Fronthaul limitations and power models are incorporated, and optimization is performed using the Concave-Convex Procedure (CCP). Simulation results show that Split 8 provides superior SE, EE, and fairness due to centralized coordination but requires higher fronthaul capacity and computing resources. Split 7.2, in contrast, reduces fronthaul load and latency, making it suitable for bandwidth-limited or delay-sensitive deploy ments. A fully decentralized architecture is also evaluated and shows significantly poorer performance across all metrics. These findings offer insight into the trade-offs of functional split selection and emphasize how architectural decisions impact network performance. The results support scalable, energy-efficient, and QoS-aware CF-mMIMO networks aligned with the O-RAN paradigm.