Beacon | Disaster Intelligence Platform

Beacon is a next-generation disaster intelligence system designed to reduce response times from hours to seconds. By fusing real-time data from global sensor arrays, satellite imagery, and social signal processing, Beacon provides verified, location-aware situational awareness for emergency response teams.

---

🧠 proprietary AI Risk Scoring Engine

Beacon's core differentiator is its Dynamic Risk Coefficient Algorithm, which processes thousands of data points to assign a unified Threat Score (0-100) to any geolocation.

How it Works

The scoring engine operates on a weighted multi-modal fusion model:

1. Signal Ingestion:

   • Seismic Sensors (35% Weight): Real-time USGS and EMSC feeds.
   • Atmospheric Data (25% Weight): API-linked weather stations detecting pressure drops and wind anomalies.
   • Satellite Telemetry (20% Weight): Thermal imaging for wildfire perimeter tracking.
   • Social Verification (20% Weight): Natural Language Processing (NLP) on crowdsourced reports for impact validation.

2. Confidence Staging:

   • Level 1 (Info): Raw data detection (unverified).
   • Level 2 (Watch): Multi-source correlation (e.g., Seismic Alert + Social Spikes).
   • Level 3 (Warning): Confirmed threat with high confidence intervals.

🧪 The Math Behind the Magic: Adaptive Risk Quantification

The Risk Coefficient ($R_c$) isn't just a static average. It's a living probability engine that adapts to data velocity and veracity using a custom hyperbolic tangent weighting function.

Core Equation

$$R_c = \tanh\left( \sum_{i=1}^{n} (w_i \cdot S_i) \cdot D(t) \cdot P(d) \right) \times 100$$

Where:

• $S_i$ (Severity Signal): Normalized intensity from source $i$ (e.g., Richter scale 7.0 $\rightarrow$ 0.9).
• $w_i$ (Reliability Weight): Trust score of the source (e.g., USGS seismometer = 0.95, Twitter/X report = 0.35).
• $D(t)$ (Temporal Decay): $$e^{-\lambda t}$$ — Information creates entropy over time; data older than 15 minutes decays in influence exponentially.
• $P(d)$ (Proximity Factor): Inverse-square law applied to the user's geolocation relative to the disaster epicenter.

Data Vector Weights

Impact Vector	Weight ($w$)	Data Source
Geophysical	0.95	USGS / EMSC Seismometers
Atmospheric	0.85	NOAA / OpenWeather Satellites
Infrastructure	0.70	Traffic Cams / Power Grid Net
Social Sentiment	0.40	NLP on Geo-tagged Social Posts

🧠 NLP Sentiment Velocity

Beacon doesn't just count keywords; it measures the velocity of panic.

• Zero-Shot Classification: Uses a distilled BERT model to categorize social posts into Panic, Observation, or Irrelevant.
• Burst Detection: Triggers an alert only if the rate of Panic signals exceeds $\Delta s / \Delta t > \sigma$ (standard deviation threshold).

🗺️ Geospatial Clustering (DBSCAN)

To prevent false positives from scattered reports, we utilize Density-Based Spatial Clustering of Applications with Noise (DBSCAN):

• Epsilon ($\epsilon$): 5km radius.
• MinPoints: Minimum 3 verified sensors/reports required to form a cluster core.
• This ensures that a single faulty sensor or a localized social media rumor is treated as "Noise" and does not trigger a global Warning state.

📊 Risk Score Tiers & Operational Protocols

The final $R_c$ value determines the system's autonomous response state:

Score Range	Risk Level	Status Indicator	Operational Protocol
0 - 35	Low	🟢 Nominal	Monitor Only: Routine data logging. No user alerts sent. Background anomaly detection active.
36 - 70	Medium	🟡 Elevated	Human-in-the-Loop: Anomaly detected. Admin dashboard flagged. "Watch" notification sent to local emergency ops.
71 - 89	High	🟠 Danger	Active Warning: Confirmed threat. Instant push notifications to civilians in the 50km radius. Evacuation routes pre-calculated.
90 - 100	Critical	🔴 Extreme	Automated Defense: Multi-vector containment protocol. Public broadcast API triggered (CAP/IPAWS). All-hands emergency mobilization.

---

🚀 Key Features

• 📍 Hyper-Local Intelligence: Automatically detects user location (City/Town level) and filters global noise to show only relevant threats within a 100km radius.
• ⚡ Real-Time Dashboard: "Glass cock-pit" interface showing live updates with <1s latency.
• 🛡️ Multi-Source Fusion: Aggregates disparate data streams into a single, cohesive operational picture.
• 📱 Progressive Web App: Offline-capable architecture with Service Worker supported push notifications.
• 👮 Admin Command Center: Secure interface for manual alert validation and public broadcast authorization.

---

🛠️ Technology Stack

• Framework: Next.js 15 (App Router)
• Styling: Tailwind CSS with Custom "Maggie" Design System (Teal/Orange/Mint)
• Animation: GSAP (GreenSock) for high-performance interactions
• Database: SQLite (better-sqlite3) with Serverless Fallback logic (Vercel-ready)
• Intelligence: Apify Actors & OpenAI Integration

---

📦 Getting Started

1. Clone the Repository

   git clone https://github.com/jrdevadattan/beacon.git
   cd beacon

2. Install Dependencies

   npm install

3. Configure Environment Create a .env.local file:

   APIFY_API_TOKEN=your_token_here
   NEXT_PUBLIC_VAPID_PUBLIC_KEY=your_key

4. Run Development Server

   npm run dev

---