next up previous contents
Next: TCP Congestion Control Up: Extending TCP for Wireless Previous: Contents

Introduction

  Recent activity in mobile computing and wireless networks give a strong indication that mobile computers with wireless communication links will be an integral part of future internetworks. Communication over wireless links is characterized by limited capacity, sporadic high bit-error rates and temporary disconnections. The problem of mobile hosts with unreliable channels connecting them to their base stations has been attacked by various researchers in different levels.

Several Mobile-IP proposals [4,5] have addressed the problem of delivering IP packets to mobile hosts regardless of their location. In theory, one can use existing fixed network transport protocols such as UDP and TCP on the mobile hosts to communicate with the fixed network. Such proposals attempt to keep the mobility, disconnection and other features of wireless and mobile hosts transparent to the layers above the network layer, which might lead to degraded performance at these layers. On the other hand, use of a new protocol stack leads to interoperability problems.

One other approach has been to been to hide any non-congestion related losses from the TCP sender by making the low quality wireless link appear as a high quality one. The intuition behind this approach is that since the problem is local, it should be handled locally. Example of this approach include wireless links with reliable link-layer protocols such as AIRMAIL [1]. In this mechanism, the transport layer need not be aware of the characteristics of the individual links. The link quality is improved by repeated re-transmissions in the link layer level of corrupted or lost frames, or using more bits for the frame error checksum. However, since TCP uses its own end-to-end retransmission protocol, such independent re-transmission protocols might lead to degraded performance, specially as error rates become significant [3].

Similar work has been reported in the implementation of I-TCP [2], where a TCP connection is split into two separate interactions - on between the mobile host and its support router (base station) over the wireless medium, and the other between the base station and the stationary host across a wire-lined network. This allows the separation of flow control and congestion control on the wireless link from that on the fixed network. This separation is desirable because of the vastly different characteristics of the two kinds of links - the fixed links are becoming faster and more reliable every day, where as the mobile, wireless link is very slow and is extremely vulnerable to noise and loss of signal due to fading which result in high error rates. The problems associated with this approach include loss of semantics - where packets may be acknowledged before they are received by the mobile host, and software overhead - a packet needs to go through the TCP protocol stack and the associated overhead four times, once at the sender, twice at the base station and once at the mobile host.


next up previous contents
Next: TCP Congestion Control Up: Extending TCP for Wireless Previous: Contents

Suman Banerjee
Wed May 21 00:53:15 EDT 1997