Sat, 04/28/2007 - 01:16 by Damien Leroy • Categories:

Currently, broadband wireless access is gaining a great deal of interest from the networking research community. Particularly, the recently standardized WiMAX is going to serve as a wireless extension or alternative to cable and DSL for broadband access. Particularly for end users in rural, sparsely populated areas or in areas where laying cable is difficult or expensive. WiMAX will provide a new broadband access path to Internet. But companies and communities along will benefit from WiMAX as well, if they require mobile networks that cover a wider area than Wi-Fi.

Most of the wireless technologies have incorporated TCP/IP into their protocol stacks. However, TCP was designed for wired networks. Its sliding window and congestion avoidance mechanisms were designed to avoid routers congestion more than wireless connectivity and signal strength problems. TCP assumes that packet losses are caused by network congestion. While this is an acceptable assumption for many networks, segments may be lost for other reasons, such as poor data link layer transmission quality. Thus, TCP can perform poorly in situations with poor reception, such as wireless networks where the weather conditions and some other factors could affect the signal link quality. In general, the use of TCP in wireless environment has poorer and worse performance than in wired environments, since even moderate packet loss due to wireless errors has severe effects on Internet protocols.

There are several current implementations of TCP like, BIC TCP, Fast TCP and TCP Westwood that improve the congestion control mechanisms of the standard TCP implementations. However, these versions of TCP mainly focus on the improvement of data transfer in high-speed long distance networks than in wireless networks. So the behavior of TCP in a last mile broadband wireless connection is still unpredictable even with the use of a newer TCP version.

For all of these reasons and considering that nowadays WiMAX is the most important and promising technology for broadband wireless access, it is very important to study TCP real behavior in one of the pre-WiMAX implementations.

We have deployed, in Université Catholique de Louvain, a pre-WiMAX testbed that allows us to perform active measurements in order to understand TCP behavior in such an environment. Our golden dream is to reproduce TCP behavior in our lab with a model implemented in an emulator, such as dummynet.

Short Bibliography

  • Wimax forum.
  • WiMax 802.16 - site d'information et √©change sur le WiMax.
  • IEEE 802.16 Specifications.
  • G. Xylomenos and G. Polyzos. TCP and UDP Performance over a Wireless LAN. in Proc. IEEE INFOCOM. Mar. 1999.
  • J. Cheng, W. David and L. Steven. Fast TCP: Motivation, Architecture, Algorithm and Performance. in Proc. IEEE INFOCOM. Mar. 2004.
  • X. Lisong, H. Khaled and R. Injong. Binary Increase Congestion Control for Fast Long-Distance Networks. in Proc. IEEE INFOCOM. Mar. 2004.
  • H. Balakrishan, S. Padmanabhan and R. Katz. A Comparison of Mechanisms for Improving TCP Performance over Wireless. in Proc. ACM SIGCOMM. Aug. 1996.
  • S. Ramachandran, C. W. Bostian and S. F. Midkiff. Performance Evaluation of IEEE 802.16 for Broadband Wireless Access. in Proc. OPNETWORK. Aug. 2002.