NetSFC

NetSFC is a campus map system for Keio University Shonan Fujisawa Campus (SFC). It combines facility information and real-time Wi-Fi measurements in one map.

Server Status

Server	Status
Keio SFC Server
Backend

Current Status

The project is already in a usable state:

• Frontend and backend are separated.
• Map view, POI panels, and category filters are implemented.
• Wi-Fi heatmap snapshot, timeline replay, and live WebSocket updates are implemented.
• Measurement page (location + network test) is implemented.
• AI campus advisor with local tool-calling is implemented.
• Production deployment is running.

Main Features

• Interactive campus map with buildings and facility markers
• Classroom/facility detail panels with images and metadata
• Real-time Wi-Fi heatmap
• Heatmap timeline playback
• Measurement ingestion endpoint: POST /api/measurements
• Live update endpoint: WS /ws/heatmap
• AI assistant endpoint: POST /api/assistant/chat

Architecture

• Backend: FastAPI (main.py)
• Database: SQLite
• Frontend: Vanilla HTML/CSS/JavaScript (frontend/src)
• Static assets: served under /data/images

Frontend uses window.ENV.API_HOST in frontend/src/config.js to connect to the backend.

Detailed design can be found here: https://notes.tianyibrad.com/s/nMXo2ggUt

Quick Start

1. Install dependencies

python3 -m venv .netsfc_pyvenv
source .netsfc_pyvenv/bin/activate
pip install -r requirements.txt

or simply run run.sh as it would automatically configure for you.

./run.sh

2. Create .env

Create a .env file in the project root: An example would be:

APITITLE=NetSFC API
VERSION=1.0.0
DB_NAME=netsfc.db
HOST=0.0.0.0
PORT=8080
RUNTIME=PRODUCTION
CORS_ORIGINS=http://localhost:5500,http://127.0.0.1:5500

# AI Advisor - pick one provider and set its key. See .env.example for all options.
AI_PROVIDER=openai
OPENAI_API_KEY=...
# OPENAI_MODEL=gpt-4o-mini

Change RUNTIME from PRODUCTION to DEBUG for development.

AI Advisor provider

The AI Advisor (/api/assistant/chat) supports multiple LLM providers - pick one with AI_PROVIDER and set the matching API key. See .env.example for the full list of env vars and model overrides for each provider:

AI_PROVIDER	API key env var	Get a key
openai (default)	OPENAI_API_KEY	https://platform.openai.com/api-keys
grok	GROK_API_KEY	https://console.x.ai
gemini	GEMINI_API_KEY	https://aistudio.google.com/apikey
claude	ANTHROPIC_API_KEY	https://console.anthropic.com

If AI_PROVIDER is unset it defaults to openai. The legacy CHATGPT_API_KEY / CHATGPT_MODEL vars still work as aliases for OPENAI_API_KEY / OPENAI_MODEL.

3. Run backend

./run.sh

Default backend URL: http://localhost:8080

4. Open frontend pages

• frontend/src/homepage.html for the measurement page
• frontend/src/index.html for the map page

frontend/src can be hosted by any static file server.

Documentation

Communication format standard can be found in the following file:

• API: docs/API.md
• Database schema: docs/DATABASE.md

Production URLs

• Map: https://ais-official.sfc.keio.ac.jp/map/
• API: https://netsfc-api.tianyibrad.com