Partial Go back N Scheme for Occupancy Control of Reordering Buffer in 3GPP ARQ

3GPP ARQ에서 재정렬 버퍼의 점유량 조절을 위한 부분 Go back N 방식

  • Shin, Woo-Cheol (Department of Electronics and Computer Engineering, Graduate School Dankook University) ;
  • Park, Jin-Kyung (Department of Electronics and Computer Engineering, Graduate School Dankook University) ;
  • Ha, Jun (Department of Electronics and Computer Engineering, Graduate School Dankook University) ;
  • Choi, Cheon-Won (Department of Electronics and Computer Engineering, Graduate School Dankook University)
  • 신우철 (단국대학교 대학원 전자컴퓨터공학과) ;
  • 박진경 (단국대학교 대학원 전자컴퓨터공학과) ;
  • 하준 (단국대학교 대학원 전자컴퓨터공학과) ;
  • 최천원 (단국대학교 대학원 전자컴퓨터공학과)
  • Published : 2003.11.01

Abstract

3GPP RLC protocol specification adopted an error control scheme based on selective repeat ARQ. In the 3GPP ARQ, distinctive windows are provided at transmitting and receiving stations so that those stations are prohibited to send or receive data PDU's out of window. An increase in window size enhances delay performance. Such an increase, however, raises the occupancy at re-ordering buffer, which results in a long re-ordering time. Aiming at suppressing the occupancy at re-ordering buffer, we propose partial go back N scheme in this paper In the partial go back N scheme, the receiving station regards all data PDU's between the first (lowest sequence numbered) error-detected PDU and last (highest sequence numbered) error-detected PDU. By the employment of the partial go back N scheme, the occupancy at the re-ordering buffer is apparently reduced, while the delay and throughput performance may be degraded due to the remaining properties of go back N. We thus consider peak occupancy of re-ordering buffer, mean sojourn time at re-ordering buffer, mean delay time, and maximum throughput as measures to evaluate tile proposed scheme and investigate such performance by using a simulation method. From numerical examples, we observe a trade-off among performance measures and conclude that the partial go back N scheme is able to effectively reduce the occupancy of re-ordering buffer.

Keywords