Congestion control algorithms for 4g OFDMA-based systems with mixed traffic scenarios

Abstract

In this master thesis, we discuss an adaptive Congestion Control (CC) framework for networks employing OFDMA in the downlink. The adaptive CC is a generalization of a CC framework proposed before by Rodrigues, E.B and Cavalcanti, F.R.P. This generalization was motivated by the adoption of OFDMA by Fourth Generation (4G) systems, such as WiMAX and LTE-Advanced. Although the performance of the Voice over IP (VoIP) service can be improved with the adoption of adaptive CC, the VoIP Frame Erasure Rate (FER) can be better controlled if we could predict its build-up. One of the factors that can increase the FER of VoIP flows is the packet loss for example, when a packet is not transmitted before the deadline imposed by delay constraints. As a result, besides the framework generalization for OFDMA 4G systems, the main contribution of this master thesis is an additional feature to the adaptive CC framework for overload prediction based on delay. In conclusion, we show that the generalized framework with the additional feature does not only guarantee the Quality of Service (QoS) fulfillment of VoIP, while imposing only a small performance degradation to the World Wide Web (WWW) service, but also reduces FER peaks.