简体繁体 English

在python中模拟UDP中的数据包丢失

[英]Simulate packet loss in UDP in python

原文 2017-03-25 13:41:39 4 1 python/ random/ packet-loss

I am supposed to simulate a packet loss rate of 10^-2 in a Stop-and-wait protocol, that is 0.01, which means that 1 out of 100 packets transmitted get lost. 我应该假设在Stop-and-wait协议中将数据包丢失率模拟为10 ^ -2，即0.01，这意味着传输的100个数据包中有1个丢失了。 Suppose I'm sending 1000 packets, how to drop exactly 1 random packet out of the 100 packets sent throughout the transmission ? 假设我要发送1000个数据包，如何在整个传输过程中发送的100个数据包中准确地丢弃1个随机数据包？

1 个解决方案

Having a rate of 0.01 doesn't mean that exactly 1 out of 100 packets gets dropped. 速率为0.01并不意味着完全丢弃了100个数据包中的1个。 It means that each packet has a 1% chance of getting lost. 这意味着每个数据包都有1％的丢失机会。 Under the assumption that losses are independent of each other, the actual number of lost packets will follow a binomial distribution . 在损失彼此独立的假设下，实际丢失的数据包数量将遵循二项式分布。

For each packet you generate, check if a random Uniform(0,1) is less than or equal to the proportion of losses p , in your case 0.01. 对于生成的每个数据包，检查随机Uniform（0,1）是否小于或等于损耗p的比例，在您的情况下为0.01。 If it is, that packet is lost, otherwise it goes through. 如果是，则该数据包丢失，否则它将通过。 This approach scales if you increase or decrease the N , the total number of packets. 如果增加或减少数据包总数N ，则此方法可扩展。 The expected number of losses will be N * p , but if you repeat the experiment multiple times there will be variability. 预期的损失数将为N * p ，但是，如果您重复多次实验，将存在可变性。