The Construction of Near-optimal Universal Coding of Integers

Abstract: Universal Coding of Integers (UCI) is suitable for discrete memoryless sources with unknown probability distributions and infinitely countable alphabet sizes. The UCI is a class of prefix codes, such that the ratio of the average codeword length to $\max{1, H(P)}$ is within a constant expansion factor $K_{\mathcal{C}}$ for any decreasing probability distribution $P$, where $H(P)$ is the entropy of $P$. For any UCI code $\mathcal{C}$, define \emph{the minimum expansion factor} $K_{\mathcal{C}}^{*}$ to represent the infimum of the set of extension factors of $\mathcal{C}$. Each $\mathcal{C}$ has a unique corresponding $K_{\mathcal{C}}^{*}$, and the smaller $K_{\mathcal{C}}^{*}$ is, the better the compression performance of $\mathcal{C}$ is. A class of UCI $\mathcal{C}$ (or family ${\mathcal{C}i}{i=1}^{\infty}$) achieving the smallest $K_{\mathcal{C}}^{*}$ is defined as the \emph{optimal UCI}. The best result currently is that the range of $C_{\mathcal{C}}^{*}$ for the optimal UCI is $2\leq C_{\mathcal{C}}^{*}\leq 2.5$. In this paper, we prove that there exists a class of near-optimal UCIs, called $\nu$ code, to achieve $K_\nu=2.0386$. This narrows the range of the minimum expansion factor for optimal UCI to $2\leq C_{\mathcal{C}}^{*}\leq 2.0386$. Another new class of UCI, called $\Delta\delta$ code, is specifically constructed. We show that the $\Delta\delta$ code and $\nu$ code are currently optimal in terms of minimum expansion factor. In addition, we propose a new proof that shows the minimum expansion factor of the optimal UCI is lower bounded by $2$.