A novel bandwidth estimation algorithm for IEEE 802.11 TCP data transmissions

Abstract

Efficient bandwidth estimation is significant for Quality of Service (QoS) of multimedia services in IEEE 802.11 WLANs. Many bandwidth estimation solutions have been developed such as probing-based technique and cross-layer scheme. However, these solutions either introduce additional traffic or require modification to the standard protocols. This paper develops a model based bandwidth estimation algorithm at application layer when TCP traffic is delivered in the IEEE 802.11 networks. The proposed model firstly considers both TCP congestion control mechanism and IEEE 802.11 contention-based channel access mechanism. The bandwidth estimation process at the server estimates the achievable bandwidth using two types of parameters: data size to be sent and the feedback information from receivers, i.e. packet loss rate and the number of clients. The two tailed T-test analysis demonstrates that there is a 95% confidence level that there is no statistical difference between the results from the proposed bandwidth estimation algorithm and the results from the real test. Additionally, the proposed algorithm achieves higher accuracy and lower standard deviation of bandwidth, in comparison with other state-of-the art bandwidth estimation schemes.