The evolution toward 5G networks and beyond has brought unprecedented complexity in performance management. In multi-operator environments, where different providers, mobility modes, and traffic profiles converge, accurately predicting throughput becomes a critical challenge for resource orchestration. Traditional monolithic machine learning models fail to generalize across operators, mobility modes (pedestrians, buses, rapid transit), and traffic types (persistent download, upload, adaptive streaming). This stochastic gap between signal conditions and actual throughput limits the network’s ability to adapt dynamically. To overcome these limitations in heterogeneous urban environments, we propose a Tiered Multi-Agent System (TMAS) that dynamically routes edge telemetry to specialized domain micro-agents. Validated with almost 50,000 samples from a real city, the system achieves a coefficient of determination of up to 0.931 and a mean absolute error of 0.53 Mbps, with routing latencies below 0.13 ms. These figures make it a promising candidate for the response times required by next-generation networks.
Behind this innovation lies a design philosophy that combines distributed artificial intelligence with modular software architectures. Instead of a single model trying to cover everything, TMAS deploys specialized agents per operator, mobility mode, and traffic type. Each micro-agent trains quickly on specific data, reducing bias and improving accuracy. From a business perspective, this means telecom companies can integrate this technology into their network management platforms without large hardware investments. This is where custom software developed by Q2BSTUDIO allows tailoring the system to each operator’s specific needs, whether for real-time monitoring, historical analysis, or decision automation.
The key to TMAS’s success lies in its ability to manage heterogeneity. In a city like Sunway City, Malaysia, data was collected from three Tier-1 operators, three mobility modes (elevated pedestrian walkway, ground-level shuttle bus, and elevated bus rapid transit), and three traffic profiles. Conventional models could not predict reliably because radio conditions varied drastically between scenarios. However, by decomposing the problem into manageable subdomains, the micro-agents capture local patterns that elude a global approach. Artificial intelligence here is not just about prediction algorithms but also about dynamically routing inference requests to the most suitable agent, a concept aligned with the AI agents that Q2BSTUDIO implements in its enterprise solutions. These agents not only predict but can also recommend network reconfiguration actions in milliseconds.
From an infrastructure standpoint, deploying a multi-agent system at the edge requires a robust and secure cloud platform. Throughput predictions must be processed close to the access network to minimize latency, demanding cloud AWS/Azure services with edge computing capabilities. Q2BSTUDIO offers native integration with these providers, allowing the system to scale from a few base stations to nationwide coverage. Furthermore, cybersecurity is non-negotiable: telemetry data is sensitive and its transmission must be protected. Q2BSTUDIO’s cybersecurity solutions ensure that both data traffic and AI models are shielded from attacks, complying with regulations such as GDPR.
Another fundamental pillar is business analytics. Operators need to visualize performance predictions and correlate them with business indicators. This is where BI / Power BI comes in: Q2BSTUDIO develops customized dashboards that allow network and business teams to make data-driven decisions. For example, an operator can identify areas where throughput predictions are low and launch marketing campaigns to compensate, or dynamically adjust network resources. Integrating TMAS with Power BI turns complex data into actionable insights, closing the loop between artificial intelligence and business strategy.
Finally, TMAS is not just a technical advancement; it represents a paradigm shift in how prediction is approached in heterogeneous networks. The combination of autonomous micro-agents, dynamic routing, and edge computing capabilities paves the way for self-organizing networks (SON) of the sixth generation. Companies like Q2BSTUDIO, with its expertise in custom software development, artificial intelligence, cloud, cybersecurity, and business intelligence, are capable of bringing this technology from research to production. If your organization seeks to optimize 5G network performance or prepare for 6G, having a technology partner that understands both the algorithmic and operational sides is essential. TMAS demonstrates that, with the right approach, it is possible to close the stochastic gap and deliver a consistent user experience regardless of operator or environment.
