自相似业务下光缓存设计

提出了在数据包长可变的自相似业务下研究光纤延时线所组成的光缓存问题的解析模型.利用Pareto到达过程来模拟自相似业务.研究表明,缓存深度越大,包丢失率越低,而且存在与业务强度对应的最佳光纤延时线单位,使得在给定的缓存深度下,数据包丢失率最小,在业务强度和缓存深度相同时,Pareto到达比Poisson到达有更大的包丢失率.

OPTICAL BUFFER DESIGN FOR SELF-SIMILAR TRAFFIC

The approximation model is proposed to address the problem of dimensioning fiber delay lines for optical buffers for self-similar traffic with variable packet length.Pareto arrivals are used to model the self-similar traffic.As expected,packet loss probability decreases with buffer depth increasing.Optimal granularity of fiber delay lines exists for given traffic load.Under the optimal granularity,the lowest packet loss probability is achieved for given buffer depth.Compared with Poisson arrivals,Pareto arrivals correspond to larger packet loss probability under the same traffic load and buffer depth.