TCP congestion control over a live UMTS network [1][edit]

We have performed an extensive campaign of measurements that involved around 3000 flows and more than 40 hours of measurements using three different TCP stacks:

• TCP NewReno: the congestion control algorithm standardized by IETF
• TCP BIC: the default congestion control algorithm adopted by the Linux o.s.
• TCP Westwood+: available in the Linux kernel (both 2.4 and 2.6 kernels).

The experimental evaluation has been carried out by accessing the public Internet using an UMTS card (see figure below).

The following instantaneous variable measurements have been collected by using the libnetmeas library:

• Goodput
• Round Trip Time
• packet loss ratios
• number of timeouts

The main findings of the investigation can be summarized as follows:

1. the considered TCP congestion control algorithms performed similarly both in downlink and uplink scenarios;
2. the UMTS uplink channel did not exhibit any remarkable issues, providing good channel utilization and very low number of timeouts and packet retransmissions;
3. a very high number of timeouts has been observed in our measurements in the case of downlink channel that does not seem to be caused by congestion;
4. the UMTS downlink channel utilization is poor in the single flow case because of the joint effect of the very high number of retransmissions and the inflated RTT due to queuing.

We report cumulative distribution function plots:

• downlink and uplink channel goodput CDFs
• downlink and uplink RTT and Queuing time CDFs

More can be found in the following paper:

• L. De Cicco and S. Mascolo
  TCP Congestion Control over 3G Communication Systems: an Experimental Evaluation of New Reno, BIC and Westwood+
  in Proceedings of NEW2AN 2007, St. Petersburg, Russia, 2007 (PDF)