An Adaptive CQI-Based Algorithm for HSDPA Flow Control

One of the key elements of the High Speed Downlink Packet Access (HSDPA) is the fast scheduler located at the base station. For the scheduler to operate efficiently, sufficient amount of traffic needs to be buffered at the base station. However, the users' buffers at the base station are constrained by a flow control protocol to minimize data loss during handover. We evaluate the impact of this flow control on the Proportional Fair scheduler by comparing two flow control algorithms to the hypothetical case when flow control is eliminated. The results indicate that eliminating the flow control greatly improves the scheduler performance and increases the overall system throughput at low traffic load. However, at high traffic load, data loss during handover increases to a point that negates the positive improvement of the scheduler performance and reduces the effective system throughput. In light of these findings, we propose a new proactive algorithm that utilizes information about the channel quality to predict potential handover conditions and tune buffers' levels accordingly. In our simulation, the algorithm improved the scheduler performance and reduced the handover data loss by at least 44% (in our reference scenario) compared to the no-flow-control case, and hence improved the effective TCP throughput.