Turning the Map Stack Upside Down
Turning the Map Upside Down
How NexusLRS Blends Mobile & Web to Put Mapping Logic on the Client
1 | Why flip the architecture at all?
Most GIS tools still push every click or GPS ping back to a server and wait for geometry to come down the wire. NexusLRS reverses that flow. We push the brain of the LRS—routes, measures, and even dynamic-segmentation math—directly to the device so the map can answer questions in milliseconds without phoning home. The payoff is huge: practically no bandwidth, no dead-zone failures, and infrastructure you can run on a shoestring.
2 | A single data foundation for both screens
| Piece | What it is | Why it matters |
|---|---|---|
| Measure-aware vector tiles | We extend the Mapbox spec so each tile already knows its route IDs, calibration points, and M values. | The browser or phone can calculate a milepoint locally—no geometry in the API payload. |
| Tile-local indexes | On the web we build a WASM hash-map at runtime; on mobile we pre-index each tile into 15-ft geohash-9 cells. | Both deliver O(1) look-ups with <10 data-preserve-html-node="true" MB RAM per state. |
| Edge-side dynamic segmentation | A WebAssembly module in the browser and a Go micro-server on the phone dynamically segment events live. | Users drag a CSV on screen or stream GPS and see results in ~4 ms. |
3 | Mobile: MBTiles + Go = 4 ms response
- Statewide network fits in a 60 MB zoom-14 MBTiles bundle.
- A 3 MB Go server starts on app launch, loads the current tile, and builds a geohash-9 index (~0.8 ms).
- Every GPS tick (1 Hz) or tap is answered in ~4 ms—fast enough to update the UI before the next frame is drawn.
- Works 100 RPS per device, fully offline.
4 | Web: WASM tiles that think for themselves
- The page streams only the tiles in view; each tile’s payload already contains measures and calibration.
- A lightweight WASM module resolves hover milepoints, Smart-Road-Search, and straight-line-diagram syncing in single-digit milliseconds.
- Dynamic segmentation runs in a web-worker, so 100 k-row event tables draw without freezing the UI.
- Because geometry stays on the client, normal map interactions need zero round-trips.
5 | Shared wins for agencies
- Real-time milepoints everywhere — phone, tablet, laptop.
- Offline resiliency — tunnels, mountains, or post-storm blackouts don’t interrupt workflows.
- Minimal cloud footprint — WVDOT runs ~2 k monthly users on infra that costs ≈ $50/month.
- Security by default — data and calculations stay on the endpoint; only static tiles traverse the network.
6 | The bigger picture
By putting measure-aware tiles and spatial math on the edge, NexusLRS collapses wait-time, slashes bandwidth, and scales simply by adding more users—not more servers. Whether you open the app in a browser or on a rugged field tablet, the experience is the same: instant, real-time LRS at your fingertips—even when the map is upside down.
