北京航空航天大学学报 ›› 2016, Vol. 42 ›› Issue (7): 1335-1343.doi: 10.13700/j.bh.1001-5965.2015.0482

• 论文 • 上一篇    下一篇

基于网络编码的无线NoC多播映射

陈亦欧, 谭芳, 凌翔   

  1. 电子科技大学 通信抗干扰技术国家级重点实验室, 成都 611731
  • 收稿日期:2015-07-17 出版日期:2016-07-20 发布日期:2016-07-29
  • 通讯作者: 陈亦欧,Tel.: 028-61830326 E-mail: chenyiou@uestc.edu.cn E-mail:chenyiou@uestc.edu.cn
  • 作者简介:陈亦欧 女,博士,副教授。主要研究方向:通信集成电路、信号处理与多核并行技术。Tel.: 028-61830326 E-mail: chenyiou@uestc.edu.cn;凌翔 男,博士,教授,博士生导师。主要研究方向:无线通信与VLSI设计。Tel.: 028-61831270 E-mail: xiangling@uestc.edu.cn
  • 基金资助:
    中央高校基本科研业务费专项资金(ZYGX2014J011);国家重大科技专项(2013ZX03003015);国家“863”计划(2015AA01A701)

Multicast wireless NoC mapping based on network coding

CHEN Yiou, TAN Fang, LING Xiang   

  1. National Key Laboratory of Science and Technology on Communications, University of Electronic Science and Technology of China, Chengdu 611731, China
  • Received:2015-07-17 Online:2016-07-20 Published:2016-07-29

摘要: 具有高带宽、QoS保障和广播等优点的无线片上网络(NoC)能为通信信号系统的多核并行处理提供平台,而网络编码能够以较小功耗开销为代价显著提高NoC的传输效率。本文从通信信号系统的多播特性与需求出发,为无线NoC建立合适的架构与映射模型,在多约束条件下设计满足网络吞吐率性能最大化及核间通信延时、数据传输和编码功耗等开销最小化的任务映射与网络编码技术联合优化方法,实现了任务的高效并行与多处理核的协同工作。实验表明,本文提出的网络架构及映射算法与传统方式相比能节约6%的功耗,得到16%的吞吐率增益,并且路由节点硬件实现复杂度低。

关键词: 无线片上网络, 网络编码, 映射, 多播, 低功耗

Abstract: Featured by high bandwidth, QoS guarantee, and broadcasting, wireless network on chip(NoC) can provide an efficient platform for multicore and parallel processing for communication and signal processing systems. In addition, network coding can greatly increase the transmission efficiency at the cost of small power consumption in wireless NoC. On this basis, we proceeded from the features and requirements of communication and signal processing systems, and built proper architectures and mapping models for wireless NoCs. We also proposed a joint optimization method including task mapping and network coding mapping, to meet the specialized constraints and requirements of wireless NoC multicast services, such as maximizing network throughput, data transmission, and minimizing communication latency and power consumption, so as to achieve efficient task paralleling and multi-processor cooperation. Experimental results prove that the proposed architecture and mapping algorithm can achieve at least 6% power consumption saving and 16% throughput gain, with acceptable physical complexity increasement than the traditional methods.

Key words: wireless network on chip, network coding, mapping, multicast, low power

中图分类号: 


版权所有 © 《北京航空航天大学学报》编辑部
通讯地址:北京市海淀区学院路37号 北京航空航天大学学报编辑部 邮编:100191 E-mail:jbuaa@buaa.edu.cn
本系统由北京玛格泰克科技发展有限公司设计开发