Ensuring data speed and resiliency in a mobile IoT application

Control supplies racing teams with telemetry devices that carry three modems and can switch between mobile networks up to 15 times a second. Powerful — but only if each device picks the right network. Getting that decision right, automatically, at every racetrack, was the problem to solve.

The challenge

When a device arrived at a new track, it connected to the first available network rather than the optimal one — and the resulting connectivity issues could degrade performance by as much as 23%. The fix was manual and resource-intensive: roughly an hour of engineering time per session to configure 30 vehicles by hand. That created bottlenecks, led to missed configurations, and left customers at unsupported test days exposed.

The solution

Control partnered with Quix to build a machine-learning system that detects network performance in real time and optimizes connectivity by reconfiguring devices dynamically. Quix connected to Control's Azure Event Hub infrastructure, transformed raw messages into structured data with standardized business semantics, and contextualized it so each vehicle maintained its own data stream.

Models were trained on high-quality historical vehicle data and deployed inside a stream-processing pipeline that ranks network connectivity and updates each device's configuration table in real time. In total, Control created 82 specialized models — one for each venue and transmission mode — to optimize performance across every environment.

The results

Vehicle connectivity configuration became fully automatic, eliminating the manual setup entirely. Control developed, trained, tested and deployed all 82 ML models in just two weeks — and Quix's resilient architecture, with built-in data replication and sharding, delivered consistent performance without demanding extra monitoring resources.