A Network monitor, That monitors your network speed, Ping, And other stats and Much more. Made in Go-Lang.
Go Network Mon, Use Ping, It`s Super effective.
Note
The name is inspired by Pokémon.
Important
This project is primary build to be hosted by docker and used in linux. I do not know how to support windows.
- Telegram notification: Notify you using Telegram if a bad internet speed is found.
- Dashboard: A dashboard for the stats and visualize the result.
- Docker: Easily depolyable by Docker.
- Re-connect Network: Some router has a timeout in which we have to Re-Authenticate to use network. Automate it.
- Machine Learning: A ML model that will be trainned on the data to predict the correct network time frame for the next day or today.
This project is built in two distinct phases: separating the data collection engine from the predictive analytics pipeline.
A Docker/Background worker that re-connectes, re-loggin if the router needs auth. And logs the data for ML model
- Automated Re-Authentication: Detects captive portal timeouts and automatically injects login payloads to keep the internet connection alive.
- Loggin: Logs ping, packet loss, bandwidth drops, and network state to a Database at a X-minute intervals.
- Telegram Notifications: Pushes instant alerts when a bad internet speed is detected or the connection drops.
- Dashboard: A visual interface to track current network stats and uptime.
- Dockerized: Easily deployable in a lightweight, self-contained environment.
A forecasting model to forecast the network status of the next day, it takes, Day, Holiday, Time, and other stuff in considration to predict.
- Progressive Modeling: Starts with a lightweight baseline model (e.g., Prophet or Random Forest) for short-term predictions, scaling to advanced models (XGBoost) as the dataset grows across months.
- Holiday & Calendar Integration: Injects calendar features to accurately forecast bandwidth shifts during weekends, exam weeks, and national holidays.
- Garbage data Allows users to manually flag and skip "garbage data" (e.g., campus-wide power outages or router hardware failures) so the model does not train on false drops.