Even if they are similar in size and equally probable, they will follow a
Bernoulli distribution, which does not guarantee equal numbers and a uniform distribution.
Let [pi] be the
Bernoulli distribution on [S.sup.Z] defined by [pi]([s.sub.i]) [??] [p.sub.i] for all i, and let [mu] be the additive quantity defined by [mu]([s.sub.i]) [??] - log [p.sub.i].
Raghavan and Upfal [1999] considered the model in which n users generate messages according to a
Bernoulli distribution with total generation rate up to about 1/10.
This distribution looks very different from the
Bernoulli distribution shown in Figure 1a, which has the same mean.
The packet dropouts would be described as a binary sequence which is subject to a
Bernoulli distribution taking the value of one or zero with certain probability.
Bernoulli distribution with parameter p [member of] (0, 1), Bern(p).