Interference management via rate splitting and HARQ over time-varying fading channels

  • Authors:
  • Marco Levorato;Osvaldo Simeone;Urbashi Mitra

  • Affiliations:
  • University of Padova, Padova, Italy;New Jersey Institute of Technology, Newark, NJ, USA;University of Southern California, Los Angeles, CA, USA

  • Venue:
  • Proceedings of the 2009 ACM workshop on Cognitive radio networks
  • Year:
  • 2009

Quantified Score

Hi-index 0.00

Visualization

Abstract

The coexistence of two unlicensed links is considered, where one link interferes with the transmission of the other, over a time-varying, block-fading channel. In the absence of fading, standard information-theoretic results for the scenario at hand prove the sum-rate optimality of single-user detection or joint decoding at the interfered receiver, depending on the instantaneous value of the interference power. These results hinge critically on the presence of full channel state information at some of the nodes. In this paper, the problem is revisited with quasi-static fading and in the absence of channel state information at the nodes by assuming that: (i) HARQ type-I is used at both links; (ii) the channels exhibit Markovian memory over the time-slots; (iii) ACK/ NACK messages of the two HARQ processes are received at the interfering transmitter and used to adapt the current access strategy and transmission parameters. The problem is formulated as a Partially Observable Markov Decision Process (POMDP) and a greedy solution is proposed. Numerical results highlight, on the one hand, the differences between the optimal design with or without channel state information and, on the other, the advantages of exploiting channel memory for interference management.