Joint Dynamic Resource Allocation for QoS Provisioning in Multi-Access and Multi-Service Wireless Systems

  • Authors:
  • Daniel Calabuig;José F. Monserrat;David Martín-Sacristán;Narcís Cardona

  • Affiliations:
  • iTEAM Research Institute, Polytechnic University of Valencia (UPV), Valencia, Spain;iTEAM Research Institute, Polytechnic University of Valencia (UPV), Valencia, Spain;iTEAM Research Institute, Polytechnic University of Valencia (UPV), Valencia, Spain;iTEAM Research Institute, Polytechnic University of Valencia (UPV), Valencia, Spain

  • Venue:
  • Mobile Networks and Applications
  • Year:
  • 2010

Quantified Score

Hi-index 0.00

Visualization

Abstract

This paper proposes a Joint Dynamic Resource Allocation (JDRA) algorithm that allocates simultaneously the best-suited Radio Access Technologies (RATs) and amount of resources to all the users active in a multi-access wireless system. Both distributions are performed at the same time so as to make the most of the heterogeneous network. In this scenario users can connect to several RATs but not simultaneously and, therefore, the JDRA algorithm is able to consider the required handover time in the decision making. Moreover, the algorithm guarantees the Quality of Service (QoS) provision in terms of delay and bit rate in a multi-service scenario where different users may have different QoS requirements. Such a complex optimization problem has been tackled using a Hopfield Neural Network (HNN) formulation. These neural networks have fast response times once hardware implemented, which is very significant since current and future wireless networks must rapidly adapt to changing circumstances in wireless environment and traffic. Results prove the benefits achieved by the usage of the HNN-based JDRA algorithm. Firstly, the joint decision outperforms a two-steps procedure in which, after the RAT selection, the same uni-RAT DRA algorithm is applied. Secondly, the proposed algorithm can deal with different levels of congestion and load distribution among RATs in a much better way that other reference algorithms specifically designed for multi-service scenarios.