Adaptive Scheduling over Multiple Heterogeneous Mobile Access Technologies Using Quality Estimation

Swen Kühnlein

Abstract

In many regions, multiple technologies or network operators are available to provide mobile data access. Bonding multiple of these transmission paths to a single, virtual data link, oers the potential to increase achievable data rates or reliability. Due to the prevalent usage of TCP for data transmission, multi-path scheduling algorithms need to be carefully designed in order to not interfere with its congestion avoidance. In this thesis, a scheduling algorithm based on packet-train throughput estimations is presented and evaluated in a simulation scenario with dynamically changing channel characteristics. It has shown to be capable of aggregating data rates of multiple interfaces while guaranteeing in-order delivery of packets. Further research is, however, necessary in order to minimize drop rates resulting from this guarantee.

Keywords

Heterogeneous wireless network scheduling, bandwidth aggregation, packet striping, connection striping

x