Edge Agent

Overview

The edge-agent is a high-performance, Rust-based edge node designed to capture audio data, evaluate local anomaly thresholds (in dB), and securely push anomaly payloads to RabbitMQ over AMQP. It acts as the distributed sensory layer of the acoustic monitoring infrastructure.

Audio Modes

The agent supports two distinct modes of operation, configurable via the AUDIO_SOURCE environment variable:

• FILE: Reads audio from pre-recorded .wav files (specified by AUDIO_FILE_PATH). This mode is primarily used for automated Docker Compose simulations and testing.
• MICROPHONE (or MIC): Captures live audio streams directly from the device's hardware.

Warning

Crucial Note for Windows Users (Microphone Mode): For accurate ML inference and anomaly detection, you MUST disable all "Audio Enhancements" and AI noise suppression features. Navigate to the Windows Sound Control Panel -> Recording -> Properties (of your microphone) -> Enhancements, and check "Disable all enhancements". Failing to do so will distort the raw acoustic data and degrade classification accuracy.

Environment Variables

Configure the edge agent using an .env file or direct environment variables. The following are available:

Variable	Description
AMQP_URI	The RabbitMQ connection URI (e.g., amqp://guest:guest@localhost:5672/%2f).
DEVICE_ID	A unique identifier for this specific edge node (e.g., sensor-01).
LATITUDE	The geographical latitude of the sensor deployment.
LONGITUDE	The geographical longitude of the sensor deployment.
AUDIO_SOURCE	Sets the audio capture mode. Use FILE or MIC.
AUDIO_FILE_PATH	The path to the .wav file (used when AUDIO_SOURCE=FILE).
ANOMALY_THRESHOLD_DB	The local decibel threshold to trigger an anomaly event.
TELEMETRY_INTERVAL_SECS	The interval, in seconds, at which telemetry data is dispatched.
DEFAULT_SAMPLE_RATE	The sample rate for audio capture (default: 44100).
COOLDOWN_DURATION_SECS	Seconds to wait between consecutive anomaly triggers.

Setup & Running

Local Development

To run the agent locally during development, ensure you have Rust and Cargo installed, then execute:

cargo run

Release Build

For production deployment, compile a highly optimized release binary:

cargo build --release

The resulting executable will be located in target/release/edge-agent.