论文部分内容阅读
针对当前分布式抽样方法不能较好地解决抽取样本的随机性和估计无偏性的问题,采取对被抽样的数据包头中没有实际意义的Flag预留位做标记的方法,以保证各测量点抽取样本的一致性;在入口点处采取泊松抽样保证样本的随机性和估计无偏性.理论推导和实验验证均表明:所提方法实现简单、准确性高,抽取样本能较为准确地估计出流量总体特征,并且部署灵活,适合于工程应用.
Aiming at the problem that the current distributed sampling method can not solve the randomness and unbiased estimation of sampled samples well, we take the method of marking the non-meaningful Flag reserved bits in the sampled data packet header to ensure that each measuring point The randomness of sampling and Poisson sampling are guaranteed at the entry point.The theoretical derivation and experimental verification show that the proposed method is simple, accurate and the extracted samples can be more accurately estimated The overall characteristics of the outflow, and flexible deployment, suitable for engineering applications.