Covering Array Bounds Using Analytical Techniques

Abstract: A $t$-covering array with entries from the alphabet ${\cal Q}={0,1,\ldots,q-1}$ is a $k\times n$ stack, so that for any choice of $t$ (typically non-consecutive) columns, each of the $q^{t}$ possible $t$-letter words over ${\cal Q}$ appear at least once among the rows of the selected columns. We will show how a combination of the Lov\'asz local lemma; combinatorial analysis; Stirling's formula; and Calculus enables one to find better asymptotic bounds for the minimum size of $t$-covering arrays, notably for $t = 3, 4$. Here size is measured in the number of rows, as expressed in terms of the number of columns.