Network Traffic and Congestion Control

When more data flows through a network than its links can carry, packets are dropped or delayed — a phenomenon called congestion — and the algorithms that prevent or recover from this state sit at the heart of how the internet functions at scale. Researchers study how transport protocols like TCP detect and respond to congestion, how routers can actively manage their queues to signal overload before buffers overflow, and how bandwidth and delay can be accurately estimated across paths that span heterogeneous infrastructure. Understanding the underlying topology of the internet matters here too, since the shape of the network influences where bottlenecks form and how traffic from multicast or quality-of-service applications should be routed. Open questions include how to design congestion control that remains fair and stable as wireless links, satellite connections, and vastly different round-trip times become increasingly common on the same shared infrastructure.

Works

75,969

Total citations

868,582

Keywords

Congestion ControlTCPActive Queue ManagementNetwork PerformanceBandwidth EstimationInternet Topology