摘要
为消除变长长度分组和信元长度对算法性能的影响,提高带宽利用率,以iRGRR算法为例,分析了在分组调度中带宽利用率等性能,并提出了一种简单、硬件易实现的改进策略。仿真结果表明,iRGRR算法使用该策略后,不仅提高了带宽利用率,而且消除了分组变长长度对算法性能的影响。给出了该策略的实现方案。
To achieve simplicity and fine performance,the input-queuing crossbar and its scheduling algorithms always adopt fixed-size switching technology,in which the data unit is called "cell".Incoming IP packets were fragmented into cells and then processed.When the variable size of the packets were not integer times of the cell size,this assembly scheme would lead to the loss of bandwidth and decrease the performance of the scheduling algorithms.Based on iRGRR algorithm,the performance of bandwidth utilization under IP packets traffic was analyzed,and an improved scheme was presented which was simply and easily implemented according to the analysis results.The simulation results show that the negative effect of the variable size packets was removed while the performance of the bandwidth utilization was greatly increased by the improved scheme of iRGRR.Finally,an implementation method in hardware for the advanced scheme was proposed.
出处
《解放军理工大学学报(自然科学版)》
EI
北大核心
2010年第6期617-622,共6页
Journal of PLA University of Science and Technology(Natural Science Edition)
基金
江苏省自然科学基金资助项目(BK200701)
关键词
输入排队crossbar
分组调度
封装
带宽利用率
iRGRR
input-queuing crossbar
packet scheduling
assembly
bandwidth utilization
iterative request-grant-based round-robin