1. On the performance of TCP Vegas over UMTS/WCDMA channels with large round-trip time variations
- Author
-
Lo, Anthony, Heijenk, Geert, Niemegeers, I.G.M.M., prof.dr.ir. Niemegeers, I.G.M.M., Heemstra de Groot, S.M., and Dr.ir. Heemstra de Groot, S.M.
- Subjects
TCP Vegas ,EWI-2850 ,Transmission Control Protocol ,business.industry ,Code division multiple access ,Computer science ,ComputerSystemsOrganization_COMPUTER-COMMUNICATIONNETWORKS ,Congestion window ,Throughput ,Round-trip delay time ,Radio Link Control ,Universal Mobile Telecommunications System ,METIS-219900 ,W-CDMA ,Cellular network ,Zeta-TCP ,IR-62891 ,business ,UMTS frequency bands ,Computer network - Abstract
Universal Mobile Telecommunications System (UMTS) is a third-generation cellular network that enables high-speed mobile Internet access. This paper evaluates and compares the performance of two well-known versions of Transmission Control Protocol (TCP), namely, Vegas and Reno, in a UMTS environment. Bulk data transfer was considered in the simulation with varying radio channel conditions. We assume that data losses are only due to the radio channel. Simulation results show that the performance of Vegas is worse than Reno even though data losses incurred by the radio channel are completely recovered by the UMTS radio link control layer. This has led us to conduct a thorough investigation on the behavior of Vegas in order to identify the cause of performance degradation in Vegas. The poor performance of Vegas is attributed to the UMTS radio interface characteristics which resulted in large and highly variable TCP round-trip times. Vegas would interpret the round-trip time variation as a sign of congestion, and consequently, shrink its window size which reduces the transmission rate. Furthermore, a sudden increase in the instantaneous round-trip time can trigger spurious timeouts at the TCP sender using Vegas which performs unnecessary retransmissions. Spurious timeouts can lead to significant throughput reduction. Reno, on the other hand, does not show any abnormality and delivers the expected performance.
- Published
- 2004
- Full Text
- View/download PDF