The transition of fifth generation (5G) cellular systems to softwarized, programmable, and intelligent networks depends on successfully enabling public and private 5G deployments that are (i) fully software-driven and (ii) with a performance at par with that of traditional monolithic systems. This requires hardware acceleration to scale the Physical (PHY) layer performance, end-to-end integration and testing, and careful planning of the Radio Frequency (RF) environment. In this paper, we describe how the X5G testbed at Northeastern University has addressed these challenges through the first 8-node network deployment of the NVIDIA Aerial RAN CoLab (ARC), with the Aerial Software Development Kit (SDK) for the PHY layer, accelerated on Graphics Processing Unit (GPU), and through its integration with higher layers from the OpenAirInterface (OAI) open-source project through the Small Cell Forum (SCF) Functional Application Platform Interface (FAPI). We discuss software integration, the network infrastructure, and a digital twin framework for RF planning. We then profile the performance with up to 4 Commercial Off-the-Shelf (COTS) smartphones for each base station with iPerf and video streaming applications, measuring a cell rate higher than 500 Mbps in downlink and 45 Mbps in uplink.

As wireless networks evolve towards open architectures like O-RAN, testing, and integration platforms are crucial to address challenges like interoperability. This paper describes ARA-O-RAN, a novel O-RAN testbed established through the NSF Platforms for Advanced Wireless Research (PAWR) ARA platform. ARA provides an at-scale rural wireless living lab focused on technologies for digital agriculture and rural communities. As an O-RAN Alliance certified Open Testing and Integration Centre (OTIC), ARA launched ARA-O-RAN - the first public O-RAN testbed tailored to rural and agriculture use cases, together with the end-to-end, whole-stack programmability. ARA-O-RAN uniquely combines support for outdoor testing across a university campus, surrounding farmlands, and rural communities with a 50-node indoor sandbox. The testbed facilitates vital R&D to implement open architectures that can meet rural connectivity needs. The paper outlines ARA-O-RAN's hardware system design, software architecture, and enabled research experiments. It also discusses plans aligned with national spectrum policy and rural spectrum innovation. ARA-O-RAN exemplifies the value of purpose-built wireless testbeds in accelerating impactful wireless research.

Connected autonomous vehicles (CAVs) are crucial to a future society that embraces advanced technologies. For these vehicles to effectively share information with nearby vehicles or infrastructures through vehicle-to-everything (V2X) interfaces, stable mmWave communication is essential yet challenging. This paper presents an innovative approach to millimeter-wave beam management for CAVs, utilizing open radio access network (O-RAN) architecture to improve beam and handover efficiency in diverse road scenarios. Our approach uniquely combines CAV application data with mobility management to predict vehicles' location. Our findings show that this method substantially surpasses the traditional beam sweeping method by consistently maintaining a higher signal-to-noise ratio (SNR), even in challenging scenarios such as vehicle obstruction at intersections. This research underscores the potential of integrating O-RAN with vehicle-to-infrastructure (V2I) communication in CAVs, paving the way for future advancements in autonomous transportation technology.

In modern Open RAN architectures, the classic gNB radio protocol stack is disaggregated and implemented in different virtualized components, the Centralized Unit (CU), the Distributed Unit (DU), and the Radio Unit (RU). Each of these units is deployed throughout the cloud-enabled RAN infrastructure in order to achieve users' required Quality of Service (QoS). Within this framework, our study is dedicated to maximizing the admission of User Equipments (UEs) into the system while ensuring their specific QoS needs. We focus on two primary tasks: (i) establishing an association between UEs and RUs and (ii) placing CUs and DUs across the network's cloud hosts. We initially address these tasks by formulating the joint association-placement optimization problem, subject to the system's available resources and QoS-related constraints. Although it is an NP-Hard problem, we discuss how we can frame it into an Integer Linear Programming (ILP) model. Then, we propose an approximation algorithm based on the decomposition of the original ILP model. We show through exhaustive simulations that our proposed ILP model provides higher admissibility levels than other baseline models. Moreover, it significantly minimizes the deployment costs and increases the overall fairness. Finally, we remark that our decomposition algorithm presents a short optimality gap in practice, with up to 6% less admissions, while reducing the solution time by up to 98%.

5G has revolutionised mobile communication networks; however, it poses significant challenges due to the increased number of connected devices and the escalating data demands from applications. The Open Radio Access Network (O-RAN) architecture has emerged as a solution, characterised by open and standardised interfaces that foster interoperability among diverse vendors and enable the implementation of innovative solutions. In this context, the RAN Intelligent Controller (RIC) emerges as an intelligent control entity that empowers the efficient management and optimisation of the 5G RAN. Central to this orchestration are xApps, which can be developed and executed within the RIC. These applications possess the potential to drive innovation and substantially enhance the operation of 5G networks. As primary objective, this paper demonstrates the feasibility of employing a monitoring xApp within the Near-RT RIC to support the 5G core. This contributes to a better selection of user profiles, resulting in a better management of allocated resources to each user, and improved Quality of Service (QoS). By collecting and analysing real-time data, an Orchestrator enables proactive management and informed decision-making to optimise the Core performance, QoS, and resource utilisation. Specifically, Machine Learning (ML)-processed data is leveraged to select QoS profiles and assign them to individual users with the assistance of the Core Network (CN) agent. The results demonstrate the system's capability to efficiently collect and process real-time RAN data, to make user profile category predictions, and to allocate resources accordingly.

As radio access networks (RANs) evolve to incorporate new paradigms of programmability and control, the Open RAN (or O-RAN) architecture is rapidly becoming the de facto blueprint for designing next-generation cellular networks. While O-RAN has brought forth unprecedented flexibility to the RAN, it has also led to an increase in the management complexity due to the introduction of several new network functions and interfaces, ultimately leading to a disconnected RAN performance optimization (RPO) landscape. To that end, with a view to overcoming the limitations of O-RAN-driven RPO, this paper introduces IntegRAN, an enabling study that makes the case for holistic programmable RPO. Key highlights include a detailed component-level description of the IntegRAN architecture including a set of novel rApps and xApps, along with the implementation of IntegRAN in support of RAN slice lifecycle management, addressing key themes relating to end-to-end slice orchestration and application-specific RAN customization. Furthermore, the paper also includes a comprehensive experimental evaluation on an over-the-air testbed that showcases the operational efficacy of IntegRAN in improving RAN performance while seamlessly leveraging all possible O-RAN functions and interfaces in an integrated manner.

Cloud Radio Access Network (CRAN) is an architecture for wireless communication networks, particularly in the context of WiFi, cellular networks, and beyond. The main objective of CRAN is to centralize and virtualize the baseband processing functions of the network to provide several benefits in terms of scalability, efficiency, and flexibility. At first, this paper suggests a design to enhance the core capabilities of CRAN. As a result, two Radio Frequency Systems on Chip (RFSoCs) devices are used to communicate between the Radio Equipment Controller (REC) and the Radio Equipment (RE) through the Common Public Radio Interface (CPRI). The designed flow and the required FPGA resources are discussed. Secondly, the main purpose of this paper is to perform the Fast Fourier Transform (FFT) on the Orthogonal Frequency Division Multiplexing (OFDM) data when the slave sends back OFDM data to the master.

Forecasting is instrumental to efficiently manage network resources. In this workshop paper, we make the following contributions. First, we carry out the first assessment of recently proposed advanced forecasting techniques by the AI community, namely DLinear and PatchTST, when applied to the prediction of mobile traffic load and number of users connected to a single Base Station (BS). We compare these techniques with the well-known Long-Short Term Memory (LSTM) models that are widely adopted in mobile network tasks. Second, we analyze the accuracy tradeoff of these Artificial Intelligence (AI) techniques for single-and multi-step prediction horizons. Third, we profile the operation of all these black-box predictors with an EXplainable Artificial Intelligence (XAI) lens by using AIChronoLens, a new tool that links legacy XAI explanations with the temporal properties of the input sequences. We find that DLinear excels in single-step horizon predictions while PatchTST and LSTM are more accurate in multi-step horizon predictions. Our XAI study reveals that, unlike PatchTST and LSTM, DLinear focuses its prediction decisions on a few key samples of the input sequences, which ultimately lets it match the ground truth closely.

Towards Beyond 5G and 6G, Open Radio Access Network (Open-RAN) is a recent RAN architecture that promotes the decoupling of RAN components, virtualization, open interfaces, and the use of machine learning-based intelligent models. Operators can benefit from this architecture to optimize the network performance and reduce deployment and operation costs. Open-RAN paves the way for RAN-sharing, where multiple operators can share the same infrastructure instead of deploying their own. In this paper, we model the problem of allocating radio and computing resources to multiple operators with different services as an Integer Linear Programming (ILP) problem aiming to satisfy users' demands. Due to the high complexity of solving an ILP problem, we develop a policy-gradient-based Reinforcement Learning (RL) model that aims to dynamically allocate resources to operators. The simulation results demonstrate the higher efficiency of our RAN-sharing RL model as it improves the radio and CPU resource utilization compared to No-Sharing models that deploy double the amount of provisioned resources, as each operator has its own infrastructure, with its own base stations and computing resources. In the considered scenario, RL demonstrates up to 19.5% more RBs utilization and 27.4% more CPU utilization. This highlights the ability of the RL model to reduce operational and deployment costs. Additionally, the RL model outperforms static RAN-sharing algorithms thanks to its dynamic adaptation to operators' varying traffic. Precisely, it scores up to 12.3% and 17.6% more RBs and CPU utilization, respectively.

Digital network twin (DNT) is a promising paradigm to replicate real-world cellular networks toward continual assessment, proactive management, and what-if analysis. Existing discussions have been focusing on using only deep learning techniques to build DNTs, which raises widespread concerns regarding their generalization, explainability, and transparency. In this paper, we explore an alternative approach to augment network simulators with context-aware neural agents. The main challenge lies in the non-trivial simulation-to-reality (sim-to-real) discrepancy between offline simulators and real-world networks. To solve the challenge, we propose a new learn-to-bridge algorithm to cost-efficiently bridge the sim-to-real discrepancy in two alternative stages. In the first stage, we select states to query performances in real-world networks by using newly-designed cost-aware Bayesian optimization. In the second stage, we train the neural agent to learn the state context and bridge the probabilistic discrepancy based on Bayesian neural networks (BNN). In addition, we build a small-scale end-to-end network testbed based on OpenAirInterface RAN and Core with USRP B210 and a smartphone, and replicate the network in NS-3. The evaluation results show that, our proposed solution substantially outperforms existing methods, with more than 92% reduction in the sim-to-real discrepancy.

To address the evolving and diverse Quality of Service (QoS) demands in modern cellular networks, an imperative for a Machine Learning (ML) optimized programmable Radio Access Network (RAN) has become evident. This study focuses on the Radio Resource Management (RRM) aspect of this paradigm by introducing a configurable QoS-aware scheduling heuristic optimized through Deep Reinforcement Learning (DRL). The proposed framework dynamically optimizes its policies by weighing and combining multiple scheduling metrics to adapt to changing RAN demands. It exhibits tremendous promise by outperforming existing heuristic benchmarks while retaining increased flexibility compared to other low level scheduler approaches that utilize DRL. The findings underscore the potential of our approach as a mid-level DRL-scheduling technique, well-positioned to meet the evolving QoS demands towards 6G cellular networks.

The open fronthaul interface defined by O-RAN ALLIANCE aims to support the interoperability between multi-vendor open radio access network (O-RAN) radio units (O-RU) and O-RAN distributed units (O-DU). This paper introduces a new tool that could be used to evaluate Denial-of-Service (DoS) attacks against the open fronthaul interface. We launched an array of control/user planes (C/U-Planes) attacks with the tool under different traffic types and data rates, and we evaluated their impacts on the throughput and block error rate (BLER) of real-world O-RAN systems with commercial hardware.

The 6G Massive Multiple-Input Multiple-Output (MMIMO) networks can follow the so-called User-Centric Cell-Free (UCCF) architecture, where a single user is served by multiple Access Points (APs) coordinated by the Central Processing Unit (CPU). In this paper, we propose how O-RAN functionalities, i.e., rApp-xApp pair, can be used for energy-efficient Serving Cluster Formulation (SCF). Simulation studies show up to 37% gain in Energy Efficiency (EE) of the proposed solution over the state-of-the-art Network-Centric (NC) designs.

The evolution of 5G and beyond wireless communications and upcoming 6G communications have enabled many critical applications. Telerobotics over tactile Internet is one such application that involves ultra-low latency for its control and also needs bandwidth-intensive visual or multi-sensory feedback to the remote control center for critical operations involving human-in-the-loop. In this study, we have conducted an experimental evaluation of closed-loop telerobotic operations over a software-defined radio (SDR) based 5G wireless testbed and demonstrate the advantages and limitations of the current SDR technology in context of these applications.

To efficiently operate 5G radio access networks (RAN) under a variety of environments and use cases that change over time, it is important to intelligently manage the RAN and reprogram its configuration dynamically as needed. The Non-Real-Time RAN Intelligent Controllers (Non-RT RIC) as defined by O-RAN ALLIANCE can serve a key role towards such programmable RAN, by supporting AI-based rApps that can infer the best RAN operating configurations based on gathered information from RAN. In this demo, we present a series of rApps that we have developed based on O-RAN Software Community (O-RAN SC)'s Non-RT RIC. These rApps together provide localization, UE throughput prediction, and cross-cell interference management functionalities. We have successfully integrated our rApps with a commercial-grade 5G system that is deployed in our campus. Our demo shows significant end-to-end performance gain that can be obtained by using AI-driven rApps in a real-world dynamic environment.