Burst Aware Forecasting of User Traffic Demand in LEO Satellite Networks

May 14, 2026 ยท View on GitHub

This is the repo for the ``Burst Aware Forecasting of User Traffic Demand in LEO Satellite Networks" paper.

Required packages

Python 3.7.17 is used alongside the libraries provided in requirements.txt

Changes over standard Informer architecture

Three main changes can be seen under

  1. models/embed.py and data/data_loader.py for Peak embedding
  2. models/model.py for the additional Fully Connected Layers
  3. exp/exp_informer.py for the asymmetric training loss

The rest of the codebase is in parallel with standard informer architecture

Dataset

Used dataset is provided under data/ETT as the 1.04_750_10ms.csv and bursts177.csv get_bursts.py is used to find the burst locations

Results

Forecasting accuracies Burst forecasts within a prediction horizon Ablation results

Acknowledgement

Most of the codebase is derived from the following main repo:

https://github.com/zhouhaoyi/Informer2020