Low-Latency Sensor Integration for Critical Security Infrastructure

The Problem

A Reading-based security company providing automated protection for high-sensitivity sites needed to integrate a new sensor vendor's hardware into their existing ecosystem. The sites included decommissioned Ministry of Defence facilities, former nuclear power plants, and commercial locations including Heathrow Airport.

The Technical Challenge

• New sensor hubs: Devices consolidating motion detectors, light sensors, and infrared detection
• MQTT integration: Sensors publishing data via secured WiFi that needed processing
• Low-latency requirement: Security responses must be near-instantaneous
• Downstream automation: Events trigger drone deployment and ground vehicle response

The Documentation Problem

The sensor vendor provided engineering specifications written for electrical engineers, not software developers. Translation from hardware specs to software requirements added significant complexity.

Our Approach

We built a C# console application for MQTT event processing with a focus on latency-critical design.

Event Processing Pipeline

• MQTT subscription: Secure connection to sensor hub broadcasts
• Threshold configuration: Customisable triggers for motion, light, and infrared readings
• Event standardisation: Raw sensor data transformed into consistent event formats
• Downstream republishing: Standardised events triggering automated response systems

AI-Augmented Discovery

This was the first production engagement using AI-assisted discovery. ChatGPT was used to translate electrical engineering documentation into development requirements - accelerating understanding of hardware interfaces significantly.

The actual integration code was developed without AI assistance due to restricted work environment requirements, but the discovery phase demonstrated how LLMs can bridge domain expertise gaps.

Physical Testing

The sensor equipment was tested with physical hardware, watching real devices respond to integration work in real-time. This hardware-in-the-loop approach ensured the integration worked with actual equipment, not just simulated inputs.