본문 바로가기
  • Home

Optimizing MIMO Link Scheduling

  • Journal of Knowledge Information Technology and Systems
  • Abbr : JKITS
  • 2020, 15(5), pp.649-658
  • DOI : 10.34163/jkits.2020.15.5.008
  • Publisher : Korea Knowledge Information Technology Society
  • Research Area : Interdisciplinary Studies > Interdisciplinary Research
  • Received : September 29, 2020
  • Accepted : October 13, 2020
  • Published : October 31, 2020

Young-myoung Kang 1

1삼성전자

Accredited

ABSTRACT

In conventional communication systems, throughput optimization problems have been mainly dealt with in terms of protocol designs such as routing, scheduling, and encoding/decoding efficiency. However a new optimization solution has been required and widely studied for complex multivariate systems such as multi-hop MIMO networks. Cross-layer optimization has been introduced as the basic framework for the theoretical solutions to improve the network performance, and there has been a plethora of approaches to extend the cross-layer optimization processes to the multi-hop MIMO networks. Contrary to the legacy SISO, if the cross-layer optimization is applied to MIMO networks without appropriate variables reduction, the computation complexity increases exponentially. Linking the physical layer operations of MU-MIMO to other layer features such as MAC scheduling may cause a tremendous computational overhead. The problem itself has characteristics that cannot be easily expressed in LP, and even if expressed in LP, numerous variables presented in the system prevent a practical optimization. The purpose of this paper is to solve the problem of minimum length scheduling that satisfies a given traffic demand in a multi-hop MIMO network with a cross-layer optimization scheme. The solution we proposed is to devise an optimized scheduling algorithm based on LP decomposition through column generation. Through various mathematical analysis and test results, we have confirmed that the proposed method greatly improves the system performance significantly.

Citation status

* References for papers published after 2023 are currently being built.