AI-Driven 5G Advanced Network Optimization: Smartphone AI Agents | Crowdsourced QoE Analytics | Network Digital Twin

Introduction

As 5G Advanced networks emerge, ensuring Quality of Experience (QoE) for users demands intelligent, adaptive monitoring and optimization. Along with price sensitivity, poor QoE has been identified as a leading reason for customer churn in the telecom market, currently running at over 20% per annum in competitive markets for some operators (E Amiot et al, 2023 and G Miltos, 2025).

Traditional network assessments rely on centralized infrastructure, but with AI inferencing on smartphones (Lee J. etal, 2019), a new paradigm is possible—leveraging crowdsourced AI-driven QoE analytics (Hoßfeld, T. and Wunderer, S. (eds.), 2020) to enhance connectivity dynamically through autonomous networks (K McDonnald, Z Guangying, 2022). 

This is aligned and builds on much of the work being performed within 5G Advanced and future 6G standardization (Xingqin Lin., 2023).

This article presents an AI-powered framework combining:

1. Edge AI Agents on smartphones that assess QoE in real time through structured data collection and inferencing.

2. Centralized AI Agents that orchestrate optimization using multi-source analytics and generative AI for human-driven queries and operational insights.

AI Agents at the Edge: Smartphone-Based QoE Analytics

On-Device AI Inferencing

The AI Agent embedded in smartphones continuously evaluates network conditions, including signal strength, latency, bandwidth, and user interactions (e.g., video buffering, call quality). Using on-device inferencing (Lee J. etal, 2019), it processes this data locally, minimizing network overhead.

Core AI Techniques Used on Smartphones

📊 Time-Series Analysis – Detects evolving patterns in network latency and stability. 

📈 Anomaly Detection – Flags disruptions using trained machine learning models (Zamanzadeh Darban, Z., et al, 2024). 

⚡ Signal Processing Algorithms – Enhances understanding of network congestion, packet loss, and jitter. 

🔒 Privacy-Preserving Analytics – Ensures QoE data is encrypted and anonymized before transmission.

Data Crowdsourcing & Secure Transmission

Smartphone AI Agents selectively upload anonymized QoE metrics to a central AI repository, creating a real-time, crowdsourced network health map (Hoßfeld, T. and Wunderer, S. (eds.), 2020) far surpassing traditional drive tests.

✅ Real-time network visibility through large-scale data crowdsourcing. 

✅ Automated identification of weak coverage zones. 

✅ Privacy-first: Secure data aggregation prevents individual tracking.

Centralized AI Agents: Multi-Source Analysis 

AI-Powered QoE Analytics

Once smartphone AI Agents transmit structured data, centralized AI systems perform multi-source analysis, integrating: 

📡 Network metrics – Network-wide service performance data. 

📶 Network alarms and logs – External infrastructure monitoring. 

👥 User feedback & trends – Sentiment-based QoE metrics.

Advanced AI Techniques in Centralized Analysis

🔹 Predictive Modelling – Forecasts congestion hotspots based on historical trends. 

🔹 Graph-Based Network Analysis – Maps QoE data across regions to pinpoint performance bottlenecks. 

🔹 Reinforcement Learning – AI self-optimizes network parameters by continuously learning from new conditions. 

🔹 Federated Learning (Ligeng Z. etal, 2023) – Processes QoE insights across distributed devices without collecting raw user data, improving models whilst assuring privacy compliance.

🔹 Network Digital Twin (Nguyen H. etal, 2021 and Sanz Rodrigo M. etal, 2023) - With Real-time data collected from the Smartphone AI Agents, supporting simulation and analysis to highlight issues without impact to the real network.

Automated Optimization & Adaptive Engagement

💡 AI-driven analytics empower network operators to: 

✅ Detect & mitigate network degradation proactively. 

✅ Automate real-time network adjustments using predictive models. 

✅ Provide user-facing advisories, ensuring service transparency and trust.

The Future: Self-Optimizing 5G Advanced Networks

By integrating crowdsourced smartphone QoE analytics with centralized AI-driven network orchestration, telecom providers can transition toward autonomous, self-healing networks (K McDonnald, Z Guangying, 2022). AI will continually optimize connectivity, shaping the future of user-driven, AI-enhanced network evolution.

Author : Gareth Price-Jones

References

Lee, J., Chirkov, N., Ignasheva, E., Pisarchyk, Y., Shieh, M., Riccardi, F., Sarokin, R., Kulik, A., and Grundmann, M. (2019) On-Device Neural Net Inference with Mobile GPUs. Available at: Google Research (Accessed: 4 June 2025). https://arxiv.org/pdf/1907.01989 

Hoßfeld, T. and Wunderer, S. (eds.) (2020) Crowdsourced Network and QoE Measurements – Definitions, Use Cases and Challenges. Würzburg, Germany. doi: 10.25972/OPUS-20232. (Accessed: 4 June 2025). https://arxiv.org/pdf/2006.16896 

Zamanzadeh Darban, Z., Webb, G. I., Pan, S., Aggarwal, C., and Salehi, M. (2024) Deep Learning for Time Series Anomaly Detection: A Survey. ACM Computing Surveys, 57(1), Article 15. Available at: ACM Digital Library (Accessed: 4 June 2025). https://dl.acm.org/doi/pdf/10.1145/3691338 

Xingqin Lin. (2023) Artificial Intelligence in 3GPP 5G-Advanced: A Survey, Available at: IEEE ComSoc (Accessed: 6 June 2025) https://www.comsoc.org/publications/ctn/artificial-intelligence-3gpp-5g-advanced-survey 

Ligeng Zhu, Lanxiang Hu, Ji Lin, Wei-Chen Wang, Wei-Ming Chen, Chuang Gan, Song Han. PockEngine: Sparse and Efficient Fine-tuning in a Pocket (2023), Available at: Cornell University (Accessed: 6 June 2025) https://arxiv.org/abs/2310.1775

K McDonnald, Z Guangying, IG1251 Autonomous Networks Reference Architecture v1.0.1 (2022) Available at TMForum, (Accessed 10 June 2025) https://www.tmforum.org/resources/introductory-guide/ig1251-autonomous-networks-reference-architecture-v1-0-1/

Emmanuel Amiot, Pilar de Arriba, Jeff Youssef, Hendrik Willenbruch, Emanuele Raffaele, Beatriz Lacave, and Lorenzo Miláns del Bosch. Navigating Churn In A European Telecom Operators Landscape (2023), Available at: Oliver Wyman  (Accessed 8 June 2025) https://www.oliverwyman.com/our-expertise/insights/2023/oct/european-telecoms-churn-trends.html

Miltos George. Churn Rate Benchmarks by Industry 2025 (2025), Available at growth-onomics (Accessed 8 June 2025) https://growth-onomics.com/churn-rate-benchmarks-by-industry-2025/

Huan X. Nguyen, Ramona Trestian, Duc To, and Mallik Tatipamula Digital Twin for 5G and Beyond (2021), Available at Research Gate (Accessed 24 June 2025) (PDF) Digital Twin for 5G and Beyond 

M. Sanz Rodrigo, D. Rivera, J. I. Moreno, M. Àlvarez-Campana and D. R. López, "Digital Twins for 5G Networks: A Modeling and Deployment Methodology," in IEEE Access, vol. 11, pp. 38112-38126, 2023, Available at IEEE (Accessed 25 June 2025) Digital Twins for 5G Networks: A Modeling and Deployment Methodology | IEEE Journals & Magazine | IEEE Xplore