Title:Tight Probability Bounds with Pairwise Independence

Abstract:Probability bounds on the sum of $n$ pairwise independent Bernoulli random variables exceeding an integer $k$ have been extensively studied in the literature. However, these bounds are not tight in general. In this paper, we provide three results towards finding tight probability bounds for the sum of Bernoulli random variables with pairwise independence. First, for the sum of identically distributed random variables, we provide the tightest upper bound on the probability that the sum exceeds $k$, for any $k$. Second, when the distributions are non-identical, we provide the tightest upper bound on the probability that the sum exceeds $k = 1$. Lastly, we provide new upper bounds on the probability that the sum exceeds $k$, for any $k \geq 2$, by exploiting ordering of the probabilities. These bounds improve on existing bounds and under certain conditions are shown to be tight. In each case, we prove tightness by identifying a distribution that attains the bound. Numerical examples are provided to illustrate when the bounds provide significant improvements.