AI could flag network trouble through message and call fallbacks
A telecom systems engineer is exploring how AI models can use overlooked network signals, such as RCS messages falling back to SMS or VoLTE calls dropping to older systems, to detect instability. These fallback events can reveal congestion or emerging problems even when customers do not notice a service disruption. The approach depends on integrating data across network silos and analyzing granular signaling and transport data.
Telecom systems engineer Srinivasa Maringanti, who has spent over ten years working on radio access, core and O-RAN networks, is developing AI models as an independent project. His aim is linking network performance measurements with fine-grained geographic data.
He notes that RCS and VoLTE fallbacks generate signals about underlying network conditions, yet these events are invisible to subscribers because messages and calls still complete. Because such indicators exist at individual message or call level, manual monitoring is impractical, making automated analysis a possible route to spotting congestion before it becomes disruptive.
If such techniques mature, they could let operators address congestion before customers experience failures, potentially improving reliability for everyone relying on mobile messaging and voice. Network engineers and operations teams may gain earlier, more localized warnings, while subscribers might see fewer silent degradations. The benefits would likely depend on whether carriers can combine data now held in separate systems, and on how such monitoring is governed, since more granular network and location analysis could raise privacy considerations.