User-centric Data Traffic Engineering with Karma

Abstract

Imagine this: you are hosting a movie night with friends, but the video keeps buffering due to limited bandwidth. You may consider purchasing a high-speed bundle, but prioritizing Internet access based on monetary payment can undermine Net Neutrality. To date, there are no known fair means to elicit truthful demand information, leading to a polarization of existing bandwidth allocation schemes that either assume truthfulness or are fully agnostic to demand. In this project, we design a non-monetary karma economy for bandwidth allocation to fill this gap. A significant challenge is to choose a social welfare function that is suitable for the infinitely divisible nature of bandwidth, for which we propose and axiomatically justify weighted proportional fairness. With this function, we prove the well-posedness of our karma-based model and demonstrate our allocation system's efficiency and fairness in various numerical experiments.