Addressing Legacy Challenge of RTT Unfairness using Sender-Side Taxation (2024)

Undergraduate: Kimberly Brown

Faculty Advisor: Jasleen Kaur
Department: Computer Science

---

As data is transmitted, networks run into the issue of how to fairly allocate resources. Congestion control mechanisms are required to manage the allocation of resources and ensure data can move through the network by responding to potential overload. One well-established problem with current TCP protocols is that they are not RTT (round-trip time) fair. Flows of packets with smaller RTTs have an unfair advantage when sending packets to the network because they are able to receive ACKs quicker than other flows and are therefore able to increase their sending rate earlier.
We explored how to implement sender-side taxation to address this challenge of RTT unfairness and compared the fairness results achieved by Cebinae, a proposal to augment existing networks with flows by taxing switches. We found that sender-side taxation results in flows that reach greater fairness while maintaining high throughput than the default TCP congestion control algorithms or with Cebinae.