Cognitive RIC-SON: A Reinforcement-Learning xApp/rApp Framework for Multi-Vendor 5G Standalone and Open RAN Networks in Saudi Arabia

Authors

  • Mohammad Firoz Independent Researcher – Advanced Technologies Network Performance Specialist, Riyadh (KSA)

DOI:

https://doi.org/10.47577/ijitss.v5i.173

Keywords:

Self-Optimizing Networks, Cognitive SON, Open RAN, RAN Intelligent Controller, xApps, 5G Standalone, Multi-Vendor RAN, Reinforcement Learning, Conflict Management, Saudi Vision 2030, stc, Mobily; Zain, Nokia MantaRay, Hajj

Abstract

Self-Optimizing Networks (SON) have travelled a long road since 3GPP first standardized them for LTE in Release 8. The destination is now in sight: closed-loop, AI-driven, multi-vendor radio access networks that adjust themselves in real time to load, mobility, and energy conditions. Saudi Arabia has been one of the most active proving grounds for that destination. In July 2024, stc Group and Nokia announced the world's first live deployment of a cognitive, AI-powered SON module — Nokia's MantaRay Cognitive SON — on stc's commercial network during Hajj 2024 (Nokia, 2024; Telecom Review Middle East, 2025). The deployment processed more than 10,000 autonomous actions in 15-minute optimization intervals, lifted loaded-cell utilization by approximately 30 percent, improved user throughput by about 10 percent on average, and held service quality steady through a 40 percent traffic surge driven by over a million pilgrims.

That deployment is a strong validation that cognitive SON works at scale. It also exposes the next problem clearly. MantaRay is a vendor-specific platform, and stc, like any Tier-1 operator in the Kingdom, runs a multi-vendor network across Huawei, Nokia, and Ericsson. The same multi-vendor reality holds at Mobily, which signed an O-RAN MoU with Ericsson in March 2024, and at Zain KSA, which signed a Cloud RAN and AI-RAN MoU with Nokia at LEAP 2025 and launched commercial 5G SA with Huawei on the 600 MHz band in October 2025. Solutions by stc became Saudi Arabia's first commercial Open RAN operator with Mavenir in February 2024. ACES, with Radisys, completed an NTDP-funded ORAN small-cell partnership in November 2024. The Kingdom is becoming an Open RAN multi-vendor environment far more quickly than most regions of the world.

This paper proposes RIC-SON-KSA, a multi-vendor Cognitive Self-Optimizing Network framework that re-expresses the classical SON functions defined in 3GPP TS 28.313 as O-RAN xApps and rApps running over the Near-RT and Non-RT RIC respectively. The framework introduces a Conflict-Aware Coordinator (CAC) that detects and mitigates the three principal conflict classes identified in the recent O-RAN literature — direct, indirect, and implicit — across xApps and rApps from different vendors. The framework is anchored in 3GPP TS 28.313 SON specifications, 3GPP TS 28.552 PM specifications, the O-RAN Alliance WG1, WG2, and WG3 specifications including E2SM-KPM and E2SM-RC service models, and the policy frameworks issued by the Communications, Space and Technology Commission (CST), the Saudi Data and Artificial Intelligence Authority (SDAIA), and the TM Forum Autonomous Network Levels.

The methodology integrates literature review (2020 to 2026), public industry milestones from stc, Mobily, Zain, ACES, Solutions by stc, Mavenir, Nokia, Ericsson, and Huawei, and analytical projection of RIC-SON-KSA performance using real KPI outcomes from the Hajj 2024 MantaRay deployment as the empirical anchor. Where multi-vendor performance data is unavailable from public sources, illustrative projections are constructed and explicitly identified.

Findings indicate three things. First, classical SON functions map cleanly onto the xApp/rApp construct, with most mobility functions (MRO, MLB, RACH) suited to xApps and most coverage and energy functions (ANR, CCO, ES, PCI) suited to rApps. Second, the multi-vendor conflict surface is the single biggest unsolved problem in production RIC deployments, and the CAC's three-class taxonomy is sufficient to cover the conflict modes documented in Wadud et al. (2025) and the O-RAN nGRG conflict-management report. Third, projected RIC-SON-KSA performance gains over single-vendor cognitive SON are most pronounced in capacity utilization and multi-vendor integration cost, with smaller gains in handover success rate where the MantaRay baseline is already strong.

The paper concludes that the next stage of Saudi 5G excellence does not come from buying more capacity. It comes from operating the capacity already in the ground better. RIC-SON-KSA gives the Kingdom a structured, vendor-neutral, AI-native blueprint for that operating excellence, aligned with Vision 2030 and ready to scale across Hajj 2026, Riyadh Season, Expo 2030, and FIFA World Cup 2034.

Downloads

Download data is not yet available.

Downloads

Published

2026-05-08

How to Cite

Firoz, M. (2026). Cognitive RIC-SON: A Reinforcement-Learning xApp/rApp Framework for Multi-Vendor 5G Standalone and Open RAN Networks in Saudi Arabia. International Journal of Instruction, Technology and Social Sciences, 5, 56–78. https://doi.org/10.47577/ijitss.v5i.173