Meridian 🌍📍

Meridian is a specialized ingestion engine for moving assets. It solves the "Range Query" problem at scale: checking if thousands of moving trucks have entered arbitrary polygons (Geofences) in real-time.

Standard databases (PostGIS) are excellent for storage but struggle to trigger events for millions of updates per second. Meridian moves the spatial indexing logic into the application memory using Rust.

🏗️ Architecture: The Spatial Index

1. In-Memory Quadtree

Meridian maintains a lock-free, concurrent Quadtree structure in RAM.

• Ingestion: When a GPS coordinate arrives, it is mapped to a leaf node in $O(\log N)$.
• Evaluation: Point-in-Polygon checks are performed using SIMD-optimized ray casting algorithms against active geofences.

2. Write-Heavy Optimization

To handle the massive write load of fleet telemetry:

• Journaling: Updates are first appended to an immutable commit log (Kafka).
• Batching: A Rust consumer batches updates and flushes them to Cassandra (for location history) and Redis (for last-known location) to minimize IOPS.

🛠️ Tech Stack

• Language: Rust
• Spatial Index: Custom Quadtree / Geohash
• Event Log: Apache Kafka
• History Store: Apache Cassandra (ScyllaDB compatible)
• API: gRPC

🚀 Usage

// Defining a Geofence via the Rust SDK
let warehouse_zone = Polygon::new(vec![
    Point::new(44.42, 26.10),
    Point::new(44.43, 26.11),
    // ...
]);

engine.register_fence("zone_101", warehouse_zone, |event| {
    println!("Truck {} entered Warehouse!", event.asset_id);
});