RelayZero: Offline Edge-AI Semantic Compression for Disaster Mesh Networks
The Problem: When the Grid Goes Dark When catastrophic natural disasters strike, the first thing to collapse is centralized telecommunications. Cell towers lose power, fiber lines are severed, and dialing emergency numbers fails because there is no base station to route the call. In these blackouts, search-and-rescue teams rely on decentralized, sub-GHz mesh radios (like LoRa). These radios are…
When natural disasters cause widespread power outages, traditional telecommunications networks collapse, leaving search and rescue teams without a way to communicate. Sub-GHz mesh radios, like LoRa, offer a solution as they require no central towers and can work through rubble. However, these radios have a fundamental limitation - they only transmit at bytes per second, which is far too low to support high-bandwidth applications such as voice calls or lengthy text messages.
To address this issue, the RelayZero project developed an offline, edge-AI semantic compression protocol. This protocol intercepts emergency messages at the edge and utilizes artificial intelligence to compress the information into a dense 25-byte telemetry packet. The key components of RelayZero include:
1. Edge Transceiver (Laptop A): This device runs a Python backend and displays a tactical UI. It uses the Gemma 4 : E2B AI model to process incoming emergency messages, extracting critical information such as priority, location, hazard type, casualty count, and required resources. The processed data is then serialized into the compact 25-byte format.
2. Tactical Command Dashboard (Laptop B): Built with Python and Streamlit, this dashboard features a daemon Flask background listener on port 5000 to receive incoming compressed packets. It also includes geospatial triage capabilities, where incoming coordinates are mapped onto an offline OpenStreetMap using Folium to create a dynamic 150-meter danger perimeter around the incident.
3. Zero-Network Alerts: When a packet is flagged as CRITICAL, the dashboard automatically generates a two-second police siren sound using the browser's native Web Audio API, eliminating the need for external audio files.
4. Post-Disaster Logging: In the event of a disaster, triage officers can export the real-time packet queue to a structured CSV file for post-operation analysis and forensic purposes.
The project simulated a physical mesh layer using a private, zero-internet local TCP socket network, although the actual data payload remains byte-for-byte identical to what would be transmitted over a LoRa transceiver. The full implementation details and source code are available on the GitHub repository linked in the presentation.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
