This thesis comprehensively examines the rApp lifecycle within the O-RAN Alliance (O- RAN) Non-Real Time RIC (Non-RT RIC) framework, serving as a practical guide for exper- imental research and development. The focus is on the entire lifecycle of rApp development, from designing and onboarding to deployment and execution, using a spectral efficiency op- timization use case to illustrate the process. The study develops and integrates a Spectrum Optimization (SO)rApp employing Reinforcement Learning (RL) techniques, specifically a Deep Q-Network (DQN) model, within the O-RAN architecture. The research highlights how the SOrApp dynamically allocates spectrum resources to enhance network performance under varying demand conditions. Utilizing the Network Simulator (NS)-3 5G-LENA simulator, the thesis replicates diverse service demand scenarios to evaluate the rApp's effectiveness in optimizing spectral efficiency. The findings demonstrate that integrating Artificial Intelligence (AI)-driven rApps within the O-RAN framework significantly improves spectral efficiency and overall network performance, providing valuable insights and methodologies for future research and practical implementations in 5G networking. / Master of Science / As 5G networks grow more advanced, managing them effectively becomes increasingly chal- lenging. This thesis explores a method to improve network performance using specialized software applications, known as rApp, within the O-RAN Non-RT RIC framework. By focusing on the lifecycle of these rApps—how they are created, onboarded, deployed, and executed. To illustrate this process, used a real-world example of optimizing the efficiency of spectrum use, which is crucial for maintaining high-speed data and reliable communication in 5G networks. Our findings show that integrating intelligent rApps driven by Machine Learning (ML) can significantly enhance the performance and efficiency of 5G networks, offering valuable insights for future innovations in this rapidly evolving field.

As 5G and O-RAN become more widely used, the number of user equipment requesting access to the network will increase. This will require operators to expand their 5G solutions by purchasing more hardware to handle the increase in demand. The acquisition of new hardware will have both an economic and an environmental impact. Hardware is costly for operators, both in initial cost and when operating it. There is also a significant energy cost associated, which has a negative environmental impact.     This thesis explores the benefits of more advanced control over the path taken within the Radio Access Network, with the goal of increasing the number of user equipment able to connect to a static set of hardware. The control comes from new algorithms designed with the intuition that providing connections with only the bare essentials and nothing more would, in theory, increase the capacity of the whole network. Three algorithms were tested, with one representing a basic control method of selecting the first valid connection, and the other two were built on the intuition of the worst acceptable connection.     The three algorithms were tested on four different shapes of network configuration at four different sizes. The tests were run on a graph data structure implemented in C++ that represents the logical paths a connection could take. This resulted in a noticeable improvement in networks that exhibited a triangular structure, with more units as one moved toward the edge of the network. The largest improvement observed managed to fit 18.9% more units into the network.

This thesis establishes the need for a standardized, interoperable, front end interface to support the development of open RAN technologies, and establishes the viability and desirability of the VITA 49 interface standard as the alternative to other interface technologies. The purpose of this work is to propose a testbed platform for the further development for VITA 49 as a standard frontend interface as other current testbeds are not designed not as well suited to the VITA 49 standard or open RAN architecture. The VITA 49 interface standard provides a packetized interface between the front-end and the digital back-end of a split architecture system in a way that enables hardware interoperability between and within vendor supplies. The VITA 49 Radio Transport standard is ideally appropriate for integration into SDRs [12] due to its flexibility and metadata support. The REDHAWK platform is an integrated development environment which is used to develop a radio system that utilizes a remote radio unit to send and receive signals which transmits it using the VITA 49 protocol to the base band unit for processing. It was found that REDHAWK is better than GNURadio for this purpose, and that VRT technology is a much better than the current CPRI Standard as it provides an open standard, that enables a flexible, scalable interface that enables long-term growth.

As 5G becomes more established and faces widespread roll-out, energy consumption of radio access networks around the globe will increase. Since the high-frequency radio waves used in 5G communication has a shorter effective range compared to the waves used in previous generations, 5G networks will require a higher number of radio units to compensate for their reduced range. Since the transmission of radio waves is a costly procedure in terms of energy consumption, this further increases the relevancy of radio equipment when considering solutions for increasing radio access networks' energy efficiency. This thesis has therefore provided a possible solution for increasing the energy efficiency of an O-RAN compliant radio access network by decreasing the energy consumption of its radio units. If a network's radio units are capable of entering a low-power sleep mode whenever they are left idle, i.e. not handling any traffic, the energy efficiency of a network can thus be increased by forcing its radio units to enter sleep mode as often as possible. This can be done by offloading traffic from radio units with little traffic onto other nearby radio units. The handovers required to perform such offloading, however, need to be predicted on the fly somewhere within the network. The solution proposed within this thesis therefore utilizes a component indigenous to the O-RAN architecture called RIC and its functionality-defining xApps which are capable of automatically detecting situations where radio units can be put to sleep as well as handling the offloading procedures. Through testing inside a simulated network, the set of xApps designed as a solution resulted in a potential 20-35% decrease in energy consumption among a radio access network's radio units.

The growing adoption of millimeter wave (mmWave) turns efficient beamforming and beam management procedures into key enablers for 5th Generation (5G) and Beyond 5G (B5G) mobile networks. Recent research has sought to optimize beam management in modern Radio Access Network (RAN) architectures, where open, virtualized, disaggregated and multi-vendor environments are considered, and management platforms allow the use of Artificial Intelligence (AI) and Machine Learning (ML)-based solutions. Moreover, beam management represents some fundamental use cases defined by Open RAN Alliance (O-RAN). This work analyses beam management strategies in Open RAN and proposes solutions for codebook-based mmWave systems inspired by two use cases from O-RAN: the Grid of Beams (GoB) Optimization and the AI/ML-assisted Beam Selection. For the GoB Optimization use case, a scenario subject to constraints on the use of the full GoB due to overhead during beam selection is considered. An Advantage Actor Critic (A2C) learning-based framework is proposed to optimize the GoB, as well as the transmission power in a mmWave network. The proposed technique improves Energy Efficiency (EE) and ensures fair coverage is maintained. The simulations show that A2C-based joint optimization of GoB and transmission power is more effective than using Equally Spaced Beams (ESB) and fixed power, or the optimization of GoB and transmission power disjointly. Compared to the ESB and fixed transmission power strategy, the proposed approach achieves more than twice the average EE in the scenarios under test, and it is closer to the maximum theoretical EE. In the case of the AI/ML-assisted Beam Selection use case, the overhead during beam selection is addressed by a multi-modal sensing-aided ML-based method. When using sensing information sources external to the RAN in a multi-vendor disaggregated environment, such methods must account for privacy and data ownership issues. A Distributed Machine Learning (DML) strategy based on Split Learning (SL) is proposed to this end. The solution can cope with deployment challenges in novel RAN architectures and is applied to single and multi-level beam selection decisions, where the latter considers hierarchical codebook structures. With the proposed approach, accuracy levels above 90% can be achieved, while overhead decreases by 85% or more. SL achieves performance comparable to the centralized learning-based strategies, with the added value of accounting for privacy and data ownership issues.

Fifth Generation (5G) introduces technologies that expedite the adoption of mobile networks, such as densely connected devices, ultra-fast data rate, low latency and more. With those visions in 5G and 6G in the next step, the need for a higher transmission rate and lower latency is more demanding, possibly breaking Moore’s law. With Artificial Intelligence (AI) techniques becoming mature in the past decade, optimizing resource allocation in the network has become a highly demanding problem for Mobile Network Operators (MNOs) to provide better Quality of Service (QoS) with less cost. This thesis proposes a Reinforcement Learning (RL) solution on bandwidth allocation for network slicing integration in disaggregated Open Radio Access Network (O-RAN) architecture. O-RAN redefines traditional Radio Access Network (RAN) elements into smaller components with detailed functional specifications. The concept of open modularization leads to greater potential for managing resources of different network slices. In 5G mobile networks, there are three major types of network slices, Enhanced Mobile Broadband (eMBB), Ultra-Reliable Low Latency Communications (URLLC), and Massive Machine Type Communications (mMTC). Each network slice has different features in the 5G network; therefore, the resources can be relocated depending on different needs. The virtualization of O-RAN divides the RAN into smaller function groups. This helps the network slices to divide the shared resources further down. Compared to traditional sequential signal processing, allocating dedicated resources for each network slice can improve the performance individually. In addition, shared resources can be customized statically based on the feature requirement of each slice. To further enhance the bandwidth utilization on the disaggregated O-RAN, a RL algorithm is proposed in this thesis on midhaul bandwidth allocation shared between Centralized Unit (CU) and Distributed Unit (DU). A Python-based simulator has been implemented considering several types of mobile User Equipment (UE)s for this thesis. The simulator is later integrated with the proposed Q-learning model. The RL model finds the optimization on bandwidth allocation in midhaul between Edge Open Cloud (O-Cloud)s (DUs) and Regional O-Cloud (CU). The results show up to 50% improvement in the throughput of the targeted slice, fairness to other slices, and overall bandwidth utilization on the O-Clouds. In addition, the UE QoS has a significant improvement in terms of transmission time.