Fig 2 - uploaded by Myeonggi Jeong
Content may be subject to copyright.
Source publication
Contexts in source publication
Context 1
... trained a sample ML model with Acumos On-board client library to get a model dump and metadata. ( in Figure 2) Then, we uploaded them to the on-board system to get an Acumos ML container. ...
Context 2
... we modified Acumos Model Runner to handle those libraries. The proposed builder also creates xApp Description [16] which is required for xApp On-boarder to deploy xApp on RIC. ( in Figure 2) Additionally, the latest Acumos release (Cilo) packages multiple modules that are not needed for xApps such as gimp toolkit (for GUI support that are not needed for xApps) and Gunicorn (for distributed computing that RIC platform should provide). We excluded these modules and reduced the size of ML xApp package from 3GB to 1GB. ...
Context 3
... TP ML Inference xApp sends current traffic and forecasts as VES and hence they are collected by Prometheus first and then transmitted by VESPA to VES Collector in ONAP. ( in Figure 2) The collected VES is published to Data Movement as a Platform (DMaaP) [9], which is the message router of ONAP. As there is no reference implementation of rApp yet, we modified Message Router Subscription (MR Sub) module to use it as an rApp that can subscribe to VES events published. ...
Context 4
... also added additional capabilities to this rApp (modification of MR Sub) to judge the quality of the forecasts and create new policies if necessary. The created policy is pushed to Policy Agent ( in Figure 2) to be sent to through Policy Agent, A1 Controller of Non-RT RIC ( in Figure 2), and A1 Mediator of Near-RT RIC to TP Control xApp eventually. This completes the control path from Non-RT RIC and Near-RT RIC; we could confirm that the routed new policy was received by TP Control. ...
Context 5
... also added additional capabilities to this rApp (modification of MR Sub) to judge the quality of the forecasts and create new policies if necessary. The created policy is pushed to Policy Agent ( in Figure 2) to be sent to through Policy Agent, A1 Controller of Non-RT RIC ( in Figure 2), and A1 Mediator of Near-RT RIC to TP Control xApp eventually. This completes the control path from Non-RT RIC and Near-RT RIC; we could confirm that the routed new policy was received by TP Control. ...
Similar publications
Citations
... Principal initiators for this vendor-independent and AI-enabled network movement are the Telecom Infra Project (TIP) and the O-RAN Alliance. Other key communities are the O-RAN Software Community for the Development of Open RAN Solutions (O-RAN-SC), Open Network Automation Platform (ONAP), Acumos for the manufacture and deployment of AI applications, and so on [2]. However, the O-RAN specifications do not explicitly exhibit a detailed fault management architecture for open RAN specific environments. ...
In this article, we propose a multi-stage AI-based framework called “open Fault Management” or “openFM,” for end-to-end autonomous fault management of the open radio access network (open RAN) that is also applicable to 5G and beyond 5G next-generation networks. A unified vendor-independent fault management system architecture that can be seamlessly integrated with open RAN and other parallel evolving networks is outlined in this article. An optimized combination of classical machine learning and artificial neural network (ANN)-based deep learning is implemented and analyzed in different stages of the autonomous fault management process, and their accuracy and various other metrics of performance are presented.
... However, a challenge for the Open RAN system is orchestrating AI/ML models [2]. There are still concerns about determining the right AI/ML model, choosing an appropriate deployment location and resource requirement, and considering the time scale to make input available [3], [13], [17], [20]. Another major challenge in intelligence management is handling fault prediction (i.e., drifting) of AI/ML results. ...
... In [2], the authors propose OrchestRAN, an intelligent orchestration framework, to select the optimal location for the RAN function deployment. To address the memory optimization challenges when deploying AI/ML-based network solutions, the authors in [20] propose removing unused program modules and frameworks during the programming phase. The authors in [21] suggest selecting appropriate data for training rather than using all the data for faster convergence of AI/ML model and accurate results. ...
Due to the tightly-coupled hardware and software architecture of existing RAN systems and their non-flexibility, disaggregation of software and hardware can bring many unprecedented opportunities regarding enabling the entrance of multiple small-scale infrastructure providers to enter the RAN market, which creates more competitive and innovative RAN ecosystem. Moreover, mobile network operators (MNOs) will also have the advantage of selecting the services according to their network requirements. Open Radio Access Network (O-RAN) builds a multi-vendor RAN ecosystem and utilizes openness and intelligence to address the complexity of network functionalities, increase development agility, and provide more cost-effective platforms due to softwarization and the avoidance of dedicated hardware. However, O-RAN faces many challenges, such as interoperability, convergence, and AI/ML management which still need to be addressed before its wide deployment. This paper surveys the existing issues in the Open RAN ecosystem and explores existing solutions.
... Closed-loop controlling enables xAPPs to track the measurement information in real time and dynamically configure the gNB. Some key issues remain to be addressed when incorporating artificial intelligence in O-RAN, such as what kind of model should be deployed in a specific scenario, which dataset is suitable for learning, and how to design a solution with the lowest complexity for the real world systems, etc. [27], [28]. ML-based automation of O-RAN modules has enabled complex controlling through internal processing and interactions in a black-box model [29]. ...
Massive random access of devices in the emerging Open Radio Access Network (O-RAN) brings great challenge to the access control and management. Exploiting the bursting nature of the access requests, sparse active user detection (SAUD) is an efficient enabler towards efficient access management, but the sparsity might be deteriorated in case of uncoordinated massive access requests. To dynamically preserve the sparsity of access requests, a reinforcement-learning (RL)-assisted scheme of closed-loop access control utilizing the access class barring technique is proposed, where the RL policy is determined through continuous interaction between the RL agent, i.e., a next generation node base (gNB), and the environment. The proposed scheme can be implemented by the near-real-time RAN intelligent controller (near-RT RIC) in O-RAN, supporting rapid switching between heterogeneous vertical applications, such as mMTC and uRLLC services. Moreover, a data-driven scheme of deep-RL-assisted SAUD is proposed to resolve highly complex environments with continuous and high-dimensional state and action spaces, where a replay buffer is applied for automatic large-scale data collection. An actor-critic framework is formulated to incorporate the strategy-learning modules into the near-RT RIC. Simulation results show that the proposed schemes can achieve superior performance in both access efficiency and user detection accuracy over the benchmark scheme for different heterogeneous services with massive access requests.
... The key objective of O-RAN is to support 5G systems and improve traditional RAN performance by leveraging its elements' virtualization and adopting Machine Learning (ML)/Artificial Intelligence (AI) techniques into its core structure [2]. Two novel modules, the near-Real-Time RAN Intelligent Controller (near-RT RIC), and non-RT RIC were developed and tasked with centralized network abstraction to reduce further cost, network complexity, and human interaction [3]. In addition, these modules enable ML-based algorithms to be introduced to any layer of the RAN. ...
... In Eq. 1, the algorithm is provided with K=2 input variables x and corresponding target variable t. µ is the class-specific mean vector, and σ is the class-specific covariance matrix. With Equation 1 and using Bayes' theorem, we calculate the class posterior in Equation 2, then x is classified into a class in Equation 3. The two classes are congested and not congested. ...
The goal of Next-Generation Networks is to improve upon the current networking paradigm, especially in providing higher data rates, near-real-time latencies, and near-perfect quality of service. However, existing radio access network (RAN) architectures lack sufficient flexibility and intelligence to meet those demands. Open RAN (O-RAN) is a promising paradigm for building a virtualized and intelligent RAN architecture. This paper presents a Machine Learning (ML)-based Traffic Steering (TS) scheme to predict network congestion and then proactively steer O-RAN traffic to avoid it and reduce the expected queuing delay. To achieve this, we propose an optimized setup focusing on safeguarding both latency and reliability to serve URLLC applications. The proposed solution consists of a two-tiered ML strategy based on Naive Bayes Classifier and deep Q-learning. Our solution is evaluated against traditional reactive TS approaches that are offered as xApps in O-RAN and shows an average of 15.81 percent decrease in queuing delay across all deployed SFCs.
... Principal initiators for this vendor-independent and AI-enabled network movement are the Telecom Infra Project (TIP) and the O-RAN Alliance. Other key communities are the O-RAN Software Community for the Development of Open RAN Solutions (O-RAN-SC), Open Network Automation Platform (ONAP), Acumos for the manufacture and deployment of AI applications, etc. [2]. However, the O-RAN specifications do not explicitly exhibit a detailed fault management architecture for open RAN specific environments. ...
In this paper, we propose a multi-stage AI-based framework called 'open Fault Management' or 'openFM', for end-to-end autonomous fault management of the open Radio Access Network (open RAN) which is also applicable to 5G and beyond 5G next-generation networks. Open RAN is the subsequent stage of evolution for the next-generation radio access network. A unified vendor-independent fault management system architecture that can be seamlessly integrated with open RAN and other parallel evolving networks has been outlined in this paper. An optimized combination of classical machine learning and artificial neural network (ANN) based deep learning technique has been implemented and analyzed in different stages of the autonomous fault management process, and their accuracy and various other metrics of performance have been presented. We were able to achieve a classification accuracy of 98% for the prediction of false alarms and 96% for the prediction of alarm-specific suggestive actions. The Random Forest based classifier outperformed five other classifiers by being consistently accurate with a precision rate of more than 97% and a recall score of more than 99%, also with a smaller training time. Index Terms-AI-based network maintenance, autonomous fault management, AI for open RAN, RAN Intelligent Controller (RIC), 6G
... A. Reinforcement Learning Algorithm for Solving L-SP (11) The flow-split decision β k [t] in problem (11) can be estimated separably by minimizing L k [t]. ...
To enable an intelligent, programmable and multi-vendor radio access network (RAN) for 6G networks, considerable efforts have been made in standardization and development of open RAN (ORAN). So far, however, the applicability of ORAN in controlling and optimizing RAN functions has not been widely investigated. In this paper, we jointly optimize the flow-split distribution, congestion control and scheduling (JFCS) to enable an intelligent traffic steering application in ORAN. Combining tools from network utility maximization and stochastic optimization, we introduce a multi-layer optimization framework that provides fast convergence, long-term utility-optimality and significant delay reduction compared to the state-of-the-art and baseline RAN approaches. Our main contributions are three-fold: i) we propose the novel JFCS framework to efficiently and adaptively direct traffic to appropriate radio units; ii) we develop low-complexity algorithms based on the reinforcement learning, inner approximation and bisection search methods to effectively solve the JFCS problem in different time scales; and iii) the rigorous theoretical performance results are analyzed to show that there exists a scaling factor to improve the tradeoff between delay and utility-optimization. Collectively, the insights in this work will open the door towards fully automated networks with enhanced control and flexibility. Numerical results are provided to demonstrate the effectiveness of the proposed algorithms in terms of the convergence rate, long-term utility-optimality and delay reduction.
... Other papers [12,[14][15][16][17][18][19][20] introduce the O-RAN building blocks and architecture, with use cases mostly related to the application of machine learning to the RAN. The literature on Open RAN also includes several high-level white papers that summarize different elements of the O-RAN architecture [10,[21][22][23][24][25][26][27][28][29][30][31][32]. ...
... The O-RAN Alliance has collected an extensive list of 19 exemplary use cases for Open RAN deployments in [141,142], and the literature further discusses at a high level some of these in [10,[15][16][17][18][21][22][23][24][25][26][27][28][29][30][31][32]. At a high level, the scenarios and use cases can be classified in different ways, e.g., by considering the control knob or inference target, or the domain that is being controlled or optimized (e.g., a UE, a slice, a RAN node, or the whole network). ...
... O-RAN white papers and surveys. Finally, overviews of O-RAN and of its components are given by Lee et al. in [15], which implements AI/ML workflows through open-source software frameworks; by Abdalla et al. in [16], which reviews O-RAN capabilities and shortcomings; by Garcia-Saavedra and Costa-Perez in [17], which gives a succinct overview of O-RAN building blocks, interfaces and services; and by Brik et al. in [18] and Arnaz et al. in [19], which discuss deep learning and artificial intelligence applications for the Open RAN. [14] discusses open source software that can be used to deploy 5G and Open RAN networks. ...
The Open Radio Access Network (RAN) and its embodiment through the O-RAN Alliance specifications are poised to revolutionize the telecom ecosystem. O-RAN promotes virtualized RANs where disaggregated components are connected via open interfaces and optimized by intelligent controllers. The result is a new paradigm for the RAN design, deployment, and operations: O-RAN networks can be built with multivendor, interoperable components, and can be programmatically optimized through a centralized abstraction layer and datadriven closed-loop control. Therefore, understanding O-RAN, its architecture, its interfaces, and workflows is key for researchers and practitioners in the wireless community. In this article, we present the first detailed tutorial on O-RAN. We also discuss the main research challenges and review early research results. We provide a deep dive of the O-RAN specifications, describing its architecture, design principles, and the O-RAN interfaces. We then describe how the O-RAN RAN Intelligent Controllers (RICs) can be used to effectively control and manage 3GPP-defined RANs. Based on this, we discuss innovations and challenges of O-RAN networks, including the Artificial Intelligence (AI) and Machine Learning (ML) workflows that the architecture and interfaces enable, security, and standardization issues. Finally, we review experimental research platforms that can be used to design and test O-RAN networks, along with recent research results, and we outline future directions for O-RAN development.
... The importance of private 5G networks in terms of security and growth potential has been recognized by many equipment vendors, mobile operators, and organizations. To this end, various authorities and stakeholders have a will both to open up competition through open platforms (e.g., OpenCore and OpenRAN [3]- [5] and to secure networks by using local talents and expertise [6]. ...
To enable emerging applications of fifth generation (5G) systems for vertical scenarios like mobile broadband and Internet of things (IoT) for vehicles and maritime sectors, a new trend towards softwarization and establishment of private networks is gaining momentum. Different from earlier generations of mobile systems where network functions are traditionally operated based on dedicated hardware, 5G infrastructures can be virtualized and services can be provided based on software, common hardware and servers. In this paper, we present an open-source based prototype private network which has been implemented in the framework of an EEA research project SOLID-B5G. The prototype network is a 5G non-standalone (NSA) network with a fourth generation (4G) evolved packet core (EPC) connecting both evolved NodeB (eNB) and next generation NodeB (gNB) to commercial off-the-shelf (COTS) user equipment (UE). The prototype is implemented based on two popular open-source software platforms, namely OAI RAN and srsRAN. Based on our implementations, extensive experiments have been conducted to assess the performance of the prototype network in terms of downlink/uplink throughput and latency between different parts of the system.
... APIs exist for the integration of ML approaches that are able to infer in near-real-time manner, and decide on parameters such as slice allocation, for the connected network users. Leveraging on these contributions, authors in [15] develop their ML approach that enable near-real-time allocation of the networking resources. ...
5G Core Network (CN) has been designed for operating in a cloud-native manner, dynamically managed and micro-services oriented. This has been widely accepted in industry as the essential solution for realizing 5G networks, due to its potential to deliver scalability and dynamic configurability. In the case of 5G networks, orchestrators need to handle Virtual Network Functions which have different requirements when compared to cloud-based services and need to dynamically adapt/reconfigure based on the demand. In this work, we detail the potential challenges when deploying a cloud-native 5G CN, and develop schemes for dynamic scaling of network functions using Artificial Intelligence (AI). The need for a customized orchestrator deployment is demonstrated in a real test-bed setup, with the case of the Access and Mobility Function. Moreover, a Deep Learning AI approach is applied for proactively scaling network functions, providing two major benefits: A significantly higher number of users can be admitted to the 5G Network and traffic is balanced evenly among the different AMF replicas.
... [31] describes design 5G-LENA, an ns-3 module that supports end-to-end NR system-level simulations. [32] describes FlexRIC, a platform for real-time RAN control applications, [33] describes SD-RAN, which provides an O-RAN compliant cloud-native platform that can host RAN control applications, and [34] describes how an AI/ML model workflow can be deployed in O-RAN SC Near-RT RIC platform. [36]. ...
The telecommunication system being a critical pillar of emergency management, intelligent deployment and management of slices in an affected area will help emergency responders. Techniques such as automated management of Machine Learning (ML) pipelines across the edge and emergency responder devices, usage of hierarchical closed-loops, and offloading inference tasks closer to the edge can minimize latencies for first responders in case of emergencies. This study describes the major results from building a Proof of Concept (PoC) for network resource allocation for emergency management using a hierarchical autonomous Artificial Intelligence (AI)/ML-based closed-loops in the mobile network, organized by the Internal Telecommunication Union Focus Group on Autonomous Networks (ITU FG-AN). The background scenario for this PoC included the interaction between a higher closed-loop in the Operations Support System (OSS) and a lower closed-loop in Radio Access Network (RAN) to intelligently share RAN resources between the public and the emergency responder slice. Representation of closed-loop "controllers" in a declarative fashion (intent), triggering "imperative actions" in the "underlay" based on the intent, setup of a data pipeline between various components, and methods of "influencing" lower layer loops using specific logic/models, were some of the essential aspects investigated by various teams. The main conclusions are summarised in this paper, including the significant observations and limitations from the PoC as well as future directions.
