
Among the most technical projects I've worked on are SNMP sub-agents for public-address systems used on Italian railway networks: embedded devices that must be monitorable and diagnosable remotely around the clock, in a context where reliability is non-negotiable.
SNMP remains the de-facto standard for network infrastructure monitoring, and in the railway sector it allows diagnostic devices to integrate with the supervision systems already in place in control rooms. Developing a custom sub-agent means exposing the real hardware state through proprietary MIBs: audio levels, amplifier status, temperature, connectivity.
Besides the SNMP sub-agent for diagnostics, the same devices integrated a SIP client for the actual sound broadcasting: automated announcements and communications from the control room to the platforms. Running both protocols on the same embedded system required particular attention to resource management and to the robustness of both stacks in case of connectivity loss.
In a system that must operate in stations and tunnels, often with unstable connectivity, every sub-agent must degrade predictably: keep responding with the last known state instead of hanging, and resume synchronization as soon as the network is available again. Testing on real hardware, under degraded network conditions, was just as essential as development itself.
Projects like this show how embedded development for critical infrastructure requires a different approach compared to application software: fewer features, more reliability, extensive testing under real conditions. It's an area I keep following with interest for anyone with similar needs in the transportation sector.
Tell me about your idea: let's find the best solution together.
Contact me