How to Minimize Network Costs in Cloud Video Live Streaming via Smart Bandwidth Scheduling

Problem Background

In video live‑streaming scenarios, network cost is a key factor affecting service operating expenses. Optimizing cloud resource planning and traffic scheduling can save billions while delivering stable, smooth user experiences.

Problem Description

There are M client nodes and N edge nodes. At each time slot, the algorithm must decide how to allocate each client’s bandwidth demand to edge nodes. Allocation is allowed only when the QoS between a client and an edge node is below a given threshold, and the total demand assigned to an edge node cannot exceed its bandwidth limit. The objective is to allocate demands across all time slots so that the total bandwidth cost is minimized.

Scoring Rules

The judging program reads the contestant’s solution.txt, computes the total bandwidth cost and records the program’s runtime in milliseconds. A lower total cost wins; if costs tie, the shorter runtime wins; if still tied, the earlier submission wins. The total runtime for all computation steps on each data set must not exceed 300 seconds. Any timeout, error, or invalid solution results in no score.

Dataset Information

Practice Stage : Two data sets are provided – a non‑downloadable practice set for online verification and a downloadable test set for offline debugging.

Formal Competition Stage : One or more official data sets are used; the sum of scores across all sets determines the final ranking.

Timeline

Mar 14 09:00 – Contest problem released and data sets made available.

Mar 14 09:00 – Mar 24 18:00 – Practice phase.

Mar 24 18:00 – Mar 25 09:00 – Online training submission window closed.

Mar 25 09:00 – Mar 27 18:00 – Initial competition submission phase.

Mar 27 18:00 – System locks submissions; best result during the formal phase becomes the final score.

Mar 29 15:00 – Initial competition results announced.

Mar 30 15:00 – Final competition results announced.

Notes

Supported programming languages are C/C++, Java, and Python (CPython, PyPy). Submission limits are 20 per day during both practice and formal phases. Teams are reminded to manage version control carefully.