AI-Driven Proactive Access Control for Urban 5G Radio Access Networks

Author (s)

Mabel Ernestine Cole, Saio Alusine Marrah, Karl Mark Arhin, Selorm K. Dzokoto, Gibrilla Deen Kamara, & Mutota Simion Mutota

Abstract

Urban 5G Radio Access Networks (RANs) are increasingly challenged by highly dynamic access demand driven by dense populations, large-scale events, and commuter mobility patterns. Conventional reactive Access Class Barring (ACB) mechanisms respond only after congestion occurs, leading to elevated random-access collisions, excessive access delays, and degraded connection success rates during traffic surges. To address these limitations, this paper proposes AI-PAC, a proactive, AI-driven access control framework that predicts short-term access pressure and dynamically actuates ACB parameters before congestion manifests. AI-PAC integrates multi-source urban mobility and network telemetry with a short-horizon Long Short-Term Memory (LSTM) model to forecast random access load at the cell level with 13-15% lower RMSE compared to statistical baselines. Based on these predictions, the framework proactively adjusts ACB barring factors within predefined safety constraints to balance access efficiency, reliability, and fairness. The proposed system is evaluated using a realistic urban dataset derived from the Milano Grid, under representative high-stress scenarios including stadium events and commuter corridors, using a detailed 5G NR RACH simulation model aligned with 3GPP specifications. Performance is benchmarked against traditional reactive ACB and no-control baselines using key metrics. Experimental results demonstrate that AI-PAC consistently reduces PRACH collisions by 30% compared to reactive schemes, improves 95th-percentile access delay by 34% (from 285ms to 187ms), and enhances RRC setup success rates by 7.3 percentage points under heavy traffic loads, while maintaining high fairness (Jain’s index of 0.94) across cells. An ablation study identifies the 10-minute prediction horizon as optimal, balancing forecast accuracy with actuation latency. The findings confirm that predictive, AI-enabled access control significantly improves network resilience and user experience in dense urban environments. This work highlights the practical feasibility of proactive congestion management in 5G RANs and provides a foundation for future extensions toward autonomous and 6G-ready access control architectures.

Key words: AI-Driven, Proactive Access Control, Urban 5G, Radio Access Networks, Milano Grid.

Download PDF

Cite This Article:

Fathurrohman. Rahayu, E. S., & Marwanti, S. (2026). Blue Ocean Strategy and Smallholder Tea Value Innovation: Empirical Evidence from Batang Regency, Indonesia, Journal of Scientific Reports, 14(1), 274-289. DOI: https://doi.org/10.58970/JSR.1217

About Author (s)

Mabel Ernestine Cole, Information and Communication Engineering at the School of Electronics Information and Communication, Huazhong University of Science and Technology, Wuhan, Hubei China;  Limkokwing University of Creative Technology, Freetown, Sierra Leone. ORCID: https://orcid.org/0009-0002-1956-4474

Saio Alusine Marrah (Corresponding Author), Limkokwing University of Creative Technology, Freetown, Sierra Leone;  School of Software Engineering, University of Electronic Science and Technology of China (UESTC), Chengdu, Sichuan 611731, China. ORCID: https://orcid.org/0009-0008-4727-2480

Karl Mark Arhin, Political Science Department, KAAF University, Ghana. ORCID: https://orcid.org/0009-0007-4362-5800

Selorm K. Dzokoto, School of Software Engineering, University of Electronic Science and Technology of China (UESTC), Chengdu, Sichuan 611731, China. ORCID:  https://orcid.org/0009-0003-1230-0414

Gibrilla Deen Kamara, Limkokwing University of Creative Technology, Freetown, Sierra Leone;  School of Computer Science and Engineering, University of Electronic Science and Technology of China (UESTC), Chengdu, Sichuan 611731, China. ORCID: https://orcid.org/0009-0005-1732-6407

Mutota Simion Mutota, School of Software Engineering, University of Electronic Science and Technology of China (UESTC), Chengdu, Sichuan 611731, China. ORCID: https://orcid.org/0009-0004-3936-0246

Download PDF

DOI: https://doi.org/10.58970/JSR.1217

Summarize this article with: