On the structure of data losses induced by an overflowed buffer
作者：
Highlights：
• The autocorrelation of the interarrival times of a stream of packets has a deep impact on the structure of losses induced by an overflowed buffer.
• The burst ratio parameter can be effectively computed for an autocorrelated stream using the Markov-modulated Poisson process (MMPP) model.
• The autocorrelation makes the dependence of the burst ratio on the system load to be complicated, with muliple local extrema.
• For small or moderate buffers, the burst ratio in an uncorrelated model can be higher than in its autocorrelated counterpart.
• The burst ratio depends severely on the service time distribution, no matter if arrivals are positively autocorrelated, or not.
摘要
•The autocorrelation of the interarrival times of a stream of packets has a deep impact on the structure of losses induced by an overflowed buffer.•The burst ratio parameter can be effectively computed for an autocorrelated stream using the Markov-modulated Poisson process (MMPP) model.•The autocorrelation makes the dependence of the burst ratio on the system load to be complicated, with muliple local extrema.•For small or moderate buffers, the burst ratio in an uncorrelated model can be higher than in its autocorrelated counterpart.•The burst ratio depends severely on the service time distribution, no matter if arrivals are positively autocorrelated, or not.
论文关键词：Packetized data loss,Buffer overflow,Loss ratio,Burst ratio,Autocorrelated arrival stream
论文评审过程：Received 5 July 2021, Revised 31 August 2021, Accepted 24 September 2021, Available online 18 October 2021, Version of Record 18 October 2021.
论文官网地址：https://doi.org/10.1016/j.amc.2021.126724
