|Title||Bandwidth Allocation with Fairness in Multipath Networks|
|Authors||Hamzeh, H., Hemmati, M. and Shirmohammadi, S.|
Network resource management and traffic engineering are important subjects in today’s Internet. In terms of traffic engineering, bandwidth allocation and splitting it in a fair manner among different users have become challenging. In addition, optimizing the utilization of network resources, increasing the user utility and throughput are also considerable. So, the user satisfaction with regard to the resource allocation and Quality of Service (QoS) are the most important factors that should be taken into the consideration. At the first step, Network Utility Maximization (NUM) problem has been considered as an initial stage to design any traffic engineering method. In this paper and by considering the mentioned issues, first of all we take into account the NUM problem and optimization decomposition methods by focusing on Traffic Management Using Multipath Protocol (TRUMP), and its weaknesses to tackle the fair resource allocation problem associated with it. We then propose a model to tackle the fair bandwidth allocation issue by implementing an optimized sending rate adaptation model using an intuitive investment method to optimize the link prices (delay and loss) to achieve an efficient fair bandwidth allocation model. The model is evaluated by using different simulations and different topologies under various network conditions. Our results show that the proposed model behaves fairer than TRUMP in certain path selections. As an average from the results and at a minimum point our model achieves 26% improvement in fairness in contrast to TRUMP. In addition, for large networks we can enjoy approximately 90% improvement in fairness measure.
|Journal||International Journal of Computer and Communication Engineering|
|Journal citation||6 (3), pp. 151-160|
|Publisher||International Academy Publishing|
|Digital Object Identifier (DOI)||https://doi.org/10.17706/IJCCE.2017.6.3.151-160|
|Published||30 Jun 2017|