Migrating State Information in Mobile Environments

Several studies point out the deficiencies of the transport protocol TCP in wireless scenarios. Falsely, TCP’s congestion avoidance mechanisms are triggered by packets that are destroyed due to the high bit error probability of wireless channels or loss due to disruptions caused by handoffs. This can be avoided, if the indirect transport protocol model approach is applied. It subdivides the end-to-end transport connection into two connections, one operating over the wired subpath, the other over the wireless subpath. A so called transport gateway is used to interconnect both transport connections. If the geographical distance between the transport gateway and the mobile station increases, the active transport layer entities should be migrated to another transport gateway closer to the mobile host. Depending on the amount of transferred state information the approach presented in [2] interrupts transport layer communication up to 1.4 seconds. Our approach cuts down interruption time to a constant amount of time. Special requirements on a mobility supporting network layer, e.g. Mobile IP, are needed [6]. Within this paper we focus on the migration of buffer contents to the new transport gateway. Strategies to minimize the communication overhead and migration time are discussed in detail.