1. Abstract & Problem Formulation
Abstract: Natural disasters such as flash floods, landslides, and severe storms present extreme hazards to civilian populations. Conventional navigation platforms direct evacuees using standard shortest-path algorithms (such as classical Dijkstra or A*), which frequently navigate users through high-risk disaster centroids. This paper introduces SafeZone AI, a distributed disaster management framework that dynamically fuses real-time IoT sensor telemetry, weighted environmental risk evaluation, sub-second WebSocket emergency dispatch, and a novel hazard-penalized graph routing algorithm.
Problem Statement: Existing disaster response systems suffer from three primary vulnerabilities:
- Blind Shortest-Path Evacuation: Traditional routing algorithms optimize for Euclidean/geodesic distance rather than safety cost, leading evacuees directly into active hazard zones.
- Telemetry Latency: Absence of automated continuous IoT sensor polling delays critical disaster warnings.
- Uncoordinated Emergency Dispatch: Lack of real-time bidirectional synchronization between distressed citizens and municipal command centers.
2. Mathematical Formulations & Algorithms
A. Multi-Factor Composite Risk Model
Environmental risk score \(R \in [0, 1]\) is computed using a weighted linear combination of sensor telemetry:
Where \(w_1 = 0.40\) (Rainfall Intensity), \(w_2 = 0.45\) (Water Level Threshold), and \(w_3 = 0.15\) (Wind Velocity). Risk tiers are categorized as: LOW (\(R < 0.25\)), MODERATE (\(0.25 \le R < 0.50\)), HIGH (\(0.50 \le R < 0.75\)), and CRITICAL (\(R \ge 0.75\)).
B. Hazard Avoidance Routing Penalty Function
To bypass active disaster polygons, edge costs \(W(u, v)\) in the road graph are dynamically scaled by proximity to hazard centroids:
Where \(M_{uv}\) is the midpoint of edge \((u, v)\), \(H_h\) is the centroid of hazard \(h\), \(D\) is the Haversine distance, and \(\alpha\) is the hazard penalty multiplier. If \(D(M_{uv}, H_h) \le r_h\) for a CRITICAL hazard zone, edge weight \(W'(u, v) = \infty\) (hard block).
3. Distributed System Microservices Architecture
Backend Microservices
- FastAPI / PHP Engine: Asynchronous REST API serving shelter lookups, hazard query, and SOS updates.
- WebSocket Dispatcher: Low-latency bidirectional channel pushing instant SOS events to responders.
- Sensor Telemetry Worker: Simulates & polls IoT water-level gauges and rainfall meters.
Frontend Applications
- Citizen Evacuation Client: Leaflet map UI with 1-click SOS button, risk gauge, and detour path visualization.
- Authority Command Room: Operations console with live emergency SOS table, status toggles, and capacity monitors.
4. Live Interactive Prototype Sandbox
Demonstrate live system behavior during your presentation below.
5. Experimental Evaluation & Empirical Results
Average response time under load
Zero routes through active flood zones
WebSocket broadcast latency