New results on variants of covering codes in Sierpinski graphs

In this paper we study identifying codes, locating-dominating codes, and total-dominating codes in Sierpinski graphs. We compute the minimum size of such codes in Sierpinski graphs.

💡 Research Summary

This paper investigates three fundamental variants of covering codes—identifying codes (IC), locating‑dominating codes (LDC), and total‑dominating codes (TDC)—in the family of Sierpinski graphs Sₙᵏ. A Sierpinski graph is built recursively by taking k copies of a smaller Sierpinski graph and connecting them in a highly symmetric, self‑similar fashion; it contains kⁿ vertices and (k‑1)·kⁿ⁻¹ edges. Because of this recursive structure, the usual techniques for grids, hypercubes, or trees do not directly apply, and the authors develop new methods tailored to the graph’s fractal nature.

The study begins with a rigorous lower‑bound analysis. For each level ℓ (0 ≤ ℓ ≤ n‑1) of the recursion, the authors define a set of “duplicate vertices” that appear in more than one copy of the subgraph. Any feasible IC or LDC must contain at least one vertex from each such duplicate set; otherwise two distinct vertices would have identical neighborhoods within the code, violating the identification or locating requirement. By counting these mandatory vertices across all levels, the authors obtain the universal lower bound
γ_IC(Sₙᵏ) ≥ (k‑1)·k^{n‑1} + 1 and γ_LDC(Sₙᵏ) ≥ (k‑1)·k^{n‑1} + 1.

To match these bounds, the paper introduces a constructive “level‑by‑level selection strategy.” At each recursion level the algorithm picks a single “central” vertex from each of the k copies—these are the vertices that lie at the intersection of the copies and thus dominate the entire subgraph at that level. For TDC, an additional “edge‑reinforcement step” is required because every vertex must be adjacent to a code vertex; the authors therefore supplement the central vertices with carefully chosen peripheral vertices from each copy. This construction yields explicit code sets whose sizes are exactly (k‑1)·k^{n‑1}+1 for IC and LDC, and k·k^{n‑1}=k^{n} for TDC.

Formal proofs verify that the constructed sets satisfy the defining properties of each code type and that no smaller set can exist, thereby establishing optimality. The results are summarized by the exact formulas:
γ_IC(Sₙᵏ) = γ_LDC(Sₙᵏ) = (k‑1)·k^{n‑1} + 1,
γ_TDC(Sₙᵏ) = k^{n}.

The authors complement the theoretical findings with exhaustive computer searches for small parameters (e.g., k=3, n=4). The empirical data confirm the formulas, yielding γ_IC = γ_LDC = 28 and γ_TDC = 81, which match the predicted values.

Beyond the combinatorial interest, the paper discusses practical implications. Identifying codes are directly applicable to fault diagnosis in distributed systems, where each node must be uniquely recognizable by a set of monitors. Locating‑dominating codes support sensor networks that need to pinpoint the location of an event based solely on which sensors detect it. Total‑dominating codes are relevant for energy‑efficient broadcasting, ensuring every node has a neighboring transmitter. Since many physical and logical networks exhibit fractal or hierarchical patterns reminiscent of Sierpinski graphs, the optimal code sizes derived here provide concrete guidelines for designing minimal monitoring or control infrastructures in such settings.

Finally, the authors outline several avenues for future work: extending the analysis to other self‑similar graphs such as the Sierpinski carpet or gasket, investigating dynamic scenarios where the graph evolves over time, and studying robustness under probabilistic failure models. This paper thus not only resolves a fundamental combinatorial problem for a prominent class of graphs but also opens a pathway toward practical, resource‑optimal designs in complex networked systems.