세계의 AI 및 RAN 트래픽 최적화 - 기술 및 시장

AI has made a decisive entry into the RAN, transforming how traffic is managed and optimized. After years of anticipation, AI is now actively present in RAN, reshaping its structure and capabilities. Our comprehensive new report, "AI and RAN Traffic Optimization- Technologies and Markets," delves into this transformative journey, offering key insights and market forecasts focused on traffic optimization.

AI is enhancing RAN traffic management by improving efficiency, reducing latency, and optimizing network resources. This transformation is facilitated by the transition of RAN from a rigid, monolithic structure to a more disaggregated, agile, and open system. The roles of Software-Defined Networking (SDN), Network Functions Virtualization (NFV), Cloud-Native Functions (CNF), and Open RAN (O-RAN) are crucial in enabling AI's impact on RAN traffic optimization. These technological advancements provide the foundation for AI to optimize traffic within RAN, leading to significant improvements in network performance.

Highlights:

• Insight Research breaks down the market for AI in RAN traffic optimization two criteria- mobility generation and geographical regions.
• Insight Research considers two mobility generations- 5G and others; and four geographical regions- NA, EMEA, APAC and CALA.

Table of Contents

1. Executive Summary

• 1.1. Key observations
• 1.2. Quantitative Forecast Taxonomy
• 1.3. Report Organization

2. AI/ML/DL Key Concepts Explainer

• 2.1. Artificial Intelligence
• 2.2. Machine Learning (ML)
  • 2.2.1. Supervised Machine Learning
  • 2.2.2. Unsupervised Machine Learning
  • 2.2.3. Reinforced Machine Learning
  • 2.2.4. K-Nearest Neighbor
• 2.3. Deep Learning Neural Network (DLNN)
• 2.4. Noteworthy ML and DL Algorithms
  • 2.4.1. Anomaly Detection
  • 2.4.2. Artificial Neural Networks (ANN)
  • 2.4.3. Bagged Trees
  • 2.4.4. CART and SVM Algorithms
  • 2.4.5. Clustering
  • 2.4.6. Conditional Variational Autoencoder
  • 2.4.7. Convolutional Neural Network
  • 2.4.8. Correlation and Clustering
  • 2.4.9. Evolutionary Algorithms and Distributed Learning
  • 2.4.10. Feed Forward Neural Network
  • 2.4.11. Graph Neural Networks
  • 2.4.12. Hybrid Cognitive Engine (HCE)
  • 2.4.13. Kalman Filter
  • 2.4.14. Markov Decision Processes
  • 2.4.15. Multilayer Perceptron
  • 2.4.16. Naive Bayes
  • 2.4.17. Radial Basis Function
  • 2.4.18. Random Forest
  • 2.4.19. Recurrent Neural Network
  • 2.4.20. Reinforced Neural Network
  • 2.4.21. SOM Algorithm
  • 2.4.22. Sparse Bayesian Learning

3. Virtualization of the RAN

• 3.1. The RAN and its Evolution
  • 3.1.1. Closer Look at E-UTRAN
  • 3.1.2. 5G- NR, NSA and SA
  • 3.1.3. MEC
  • 3.1.4. The Rigid CPRI
• 3.2. The Progression of the RAN to the vRAN
• 3.3. How VM-based and Container-based vRANs Compare?
  • 3.3.1. NFV architecture
  • 3.3.2. The Need for Containers
  • 3.3.3. Microservices
  • 3.3.4. Container Morphology
  • 3.3.5. Container Deployment Methodologies
  • 3.3.6. Stateful and Stateless Containers
  • 3.3.7. Advantage Containers
  • 3.3.8. Challenges Confronting Containers
• 3.4. RAN Virtualization A Story of Alliances
  • 3.4.1. O-RAN Architecture Overview
  • 3.4.2. History of O-RAN
  • 3.4.3. Workgroups of O-RAN
  • 3.4.4. Open vRAN (O-vRAN)
  • 3.4.5. Telecom Infra Project (TIP) OpenRAN

4. AI and RAN Traffic Optimization

• 4.1. O-RAN and AI
  • 4.1.1. Introduction
  • 4.1.2. RIC, xApps and rApps
  • 4.1.3. WG2 and ML
• 4.2. AI Use-Case - Traffic Optimization
  • 4.2.1. Background
  • 4.2.2. Methodologies and Challenges
  • 4.2.3. AI-based Approaches

5. Vendor Initiatives for AI in the RAN

• 5.1. Introduction
• 5.2. Salient Observations
• 5.3. Company and Organization Summary
• 5.4. Aira Channel Prediction xApp
• 5.5. Aira Dynamic Radio Network Management rApp
• 5.6. AirHop Auptim
• 5.7. Aspire Anomaly Detection rApp
• 5.8. Cisco Ultra Traffic Optimization
• 5.9. Capgemini RIC
• 5.10. Cohere MU-MIMO Scheduler
• 5.11. DeepSig OmniSig
• 5.12. Deepsig OmniPHY
• 5.13. Ericsson Radio System
• 5.14. Ericsson RIC
• 5.15. Fujitsu Open RAN Compliant RUs
• 5.16. HCL iDES rApp
• 5.17. Huawei PowerStar
• 5.18. Juniper RIC/Rakuten Symphony Symworld
• 5.19. Mavenir mMIMO 64TRX
• 5.20. Mavenir RIC
• 5.21. Net AI xUPscaler Traffic Predictor xApp
• 5.22. Nokia RAN Intelligent Controller
• 5.23. Nokia AVA
• 5.24. Nokia ReefShark Soc
• 5.25. Nvidia AI-on-5G platform
• 5.26. Opanga Networks
• 5.27. P.I. Works Intelligent PCI Collision and Confusion Detection rApp
• 5.28. Qualcomm RIC
• 5.29. Qualcomm Cellwize CHIME
• 5.30. Qualcomm Traffic Management Solutions
• 5.31. Rimedo Policy-controlled Traffic Steering xApp
• 5.32. Samsung Network Slice Manager
• 5.33. ZTE PowerPilot
• 5.34. VMware RIC

6. Telco Initiatives for AI in the RAN

• 6.1. Introduction
• 6.2. Salient Observations
• 6.3. Company and Organization Summary
• 6.4. AT&T Inc
• 6.5. Axiata Group Berhad
• 6.6. Bharti Airtel
• 6.7. China Mobile
• 6.8. China Telecom
• 6.9. China Unicom
• 6.10. CK Hutchison Holdings
• 6.11. Deutsche Telekom
• 6.12. Etisalat
• 6.13. Globe Telecom Inc
• 6.14. NTT DoCoMo
• 6.15. MTN Group
• 6.16. Ooredoo
• 6.17. Orange
• 6.18. PLDT Inc
• 6.19. Rakuten Mobile
• 6.20. Reliance Jio
• 6.21. Saudi Telecom Company
• 6.22. Singtel
• 6.23. SK Telecom
• 6.24. Softbank
• 6.25. Telefonica
• 6.26. Telenor
• 6.27. Telkomsel
• 6.28. T-Mobile US
• 6.29. Verizon
• 6.30. Viettel Group
• 6.31. Vodafone

7. Quantitative Analysis and Forecasts

• 7.1. Research Methodology
• 7.2. Quantitative Forecasts
  • 7.2.1. Overall Market
  • 7.2.2. Mobile Telephony Generations
  • 7.2.3. Geographical Regions

Tables and Figures

• Figure 3-1: VNF versus CNF Stacks
• Figure 3-2: O-RAN High-Level Architecture
• Figure 3-3: O-RAN High-Level Architecture
• Figure 3-4: Architecture of vRAN Base Station as Visualized by TIP
• Figure 4-1: Reinforcement learning model training and actor locations per O-RAN WG2
• Figure 4-2: AI/ML Workflow in the O-RAN RIC as proposed O-RAN WG2
• Figure 4-3: AI/ML deployment scenarios
• Table 5-1: AI in RAN Product and Solution Vendor Summary
• Figure 5-1: The Aira channel detection xApp functional blocks
• Figure 5-2: Modules of the Aspire Anomaly Detection rApp
• Figure 5-3: OmniPHY Module Drop in Typical vRAN Stack Overview
• Figure 5-4: Ericsson IAP
• Figure 5-5: HCL iDES rApp Architecture
• Figure 5-6: Working of the Net Ai xUPscaler
• Figure 5-7: Nokia RIC programmability via AI/ML and Customized Applications
• Figure 5-8: Timesharing the GPU in Nvidia Aerial A100
• Figure 5-8: Rimedo TS xApp in the O-RAN architecture
• Figure 5-9: Rimedo TS xApp in the VMware RIC
• Figure 5-10: PowerPilot Solution Evolution
• Table 6-1: AI in RAN Telco Profile Snapshot
• Table 7-1: Addressable Market in Traffic Optimization End-Application in Mobile RAN for AI and Related Technologies 2023-2028 ($ million)
• Table 7-2: Addressable Market in Traffic Optimization Application in Mobile RAN for AI and Related Technologies; by Mobile Telephony Generation 2023-2028 ($ million)
• Figure 7-1: Share of Addressable Market in Traffic Optimization End-Application in Mobile RAN for AI and Related Technologies; by Mobile Telephony Generation 2023-2028
• Table 7-3: Addressable Market in Traffic Optimization End-Application Mobile RAN for AI and Related Technologies; by Geographical Region 2023-2028 ($ million)
• Figure 7-2: Share of Addressable Market in Traffic Optimization End-Application Mobile RAN for AI and Related Technologies; by Geographical Region 2023-2028