Efficient Management of Flexible Functional Splits in 5G Second Phase Networks

• Main Author: Pires, Ivan Micael Vicente
• Publication Date: 2022
• Format: Master thesis
• Language: eng
• Source: Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)
• Download full: http://hdl.handle.net/10400.6/13088
• topic: 5g; 5g-Air-Simulator; Cost/Revenue Trade-Off; Delay; Fairness; Functional Splits; Goodput; Open Ran; Packet Loss Ratio; Performance Evaluation; Small Cells

The fifth mobile network generation (5G), which offers better data speeds, reduced latency, and a huge number of network connections, promises to improve the performance of the cellular network in practically every way available. A portion of the network operations are deployed in a centralized unit in the 5G radio access network (RAN) partially centralized design. By centralizing these functions, operational expenses are decreased and coordinating strategies are made possible. To link centralized units (CU) and distributed units (DU), and the DU to remote radio units (RRU), both the midhaul and fronthaul networks must have higher capacity. The necessary fronthaul capacity is also influenced by the fluctuating instantaneous user traffic. Consequently, the 5G RAN must be able to dynamically change its centralization level to the user traffic to maximize its performance. To try to relieve this fronthaul capacity it has been considered a more flexible distribution between the base band unit (BBU) (or CU and DU if enhanced common public radio interface (eCPRI) is considered) and the RRU. It may be challenging to provide high-speed data services in crowded areas, particularly when there is imperfect coverage or significant interference. Because of this, the macrocell deployment is insufficient. This problem for outdoor users could be resolved by the introduction of low-power nodes with a limited coverage area. In this context, this MSc dissertation explores, in an urban micro cell scenario model A (UMi_A) for three frequency bands (2.6 GHz, 3.5 GHz, and 5.62 GHz), the highest data rate achievable when a numerology zero is used. For this, it was necessary the implementation of the UMi_A in the 5G-air-simulator. Allowing the determination of the saturation level using the results for the packet loss ratio (PLR=2%). By assuming Open RAN (O-RAN) and functional splitting, the performance of two schedulers in terms of quality-of-service (QoS) were also studied. The QoS-aware modified largest weighted delay first (M-LWDF) scheduler and the QoS-unaware proportional fair (PF) scheduler. PLR was evaluated for both schedulers, whilst analyzing the impact of break point distance while changing the frequency band. The costs, revenues, profit in percentage terms, and other metrics were also estimated for the PF and M-LWDF schedulers when used video (VID) and video plus best effort (VID+BE), with or without consideration of the functional splits 7.2 and 6, for the three frequency bands. One concluded that the profit in percentage terms with functional split option 7.2 applied is always slightly higher than with functional split option 6. It reaches a maximum profit of 366.92% in the case of the M-LWDF scheduler, and 305.51% in the case of the PF scheduler, at a cell radius of 0.4 km for the 2.6 GHz frequency band, considering a price of the traffic of 0.0002 €/min.