Adaptive Power and Subchannel Allocation for Dual-Class OFDMA Packet Data Networks

  • Authors:
  • Antonis G. Gotsis;Nektarios Th. Koutsokeras;Philip Constantinou

  • Affiliations:
  • Mobile Radio-Communications Laboratory, ECE School, National Technical University of Athens, Greece, Attica, Greece 15773;Mobile Radio-Communications Laboratory, ECE School, National Technical University of Athens, Greece, Attica, Greece 15773;Mobile Radio-Communications Laboratory, ECE School, National Technical University of Athens, Greece, Attica, Greece 15773

  • Venue:
  • Wireless Personal Communications: An International Journal
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

Adaptive Radio Resource Allocation exploiting the inherent frequency selectivity of the wireless medium as well as the multi-user diversity effect is expected to play a crucial role in providing high QoS on emerging OFDMA-based wireless networks. Although a plethora of studies concerning exclusively constant bit rate (CBR) or variable bit rate/best effort (BE) traffic has been published to date, limited amount of work has been devoted to the more practical mixed CBR-BE data traffic scenario over OFDM radio access networks. In this paper we attempt to deal with the specific heterogeneous allocation problem, namely the maximization of elastic users' sum-throughput while providing minimum data rate service to a subset of non-elastic users. The contribution of this work is twofold. First, due to the high complexity of the resource allocation problem, we propose a relaxation method based on the prioritization of CBR- over BE-class users during the subchannel allocation procedure. We devise a method for obtaining the exact performance penalty induced by the specific hypothesis when compared to the optimal (unprioritized) decision. Secondly, we develop a polynomial complexity approximation algorithm for allocating power and bandwidth, that employs the CBR-prioritization idea. The scheme is shown to experience a relatively low performance penalty compared to its upper bound and to outperform two representative algorithms from the literature.