On the Fault Tolerance and Hamiltonicity of the Optical Transpose Interconnection System of Non-Hamiltonian Base Graphs

Hamiltonicity is an important property in parallel and distributed computation. Existence of Hamiltonian cycle allows efficient emulation of distributed algorithms on a network wherever such algorithm exists for linear-array and ring, and can ensure deadlock freedom in some routing algorithms in hierarchical interconnection networks. Hamiltonicity can also be used for construction of independent spanning tree and leads to designing fault tolerant protocols. Optical Transpose Interconnection Systems or OTIS (also referred to as two-level swapped network) is a widely studied interconnection network topology which is popular due to high degree of scalability, regularity, modularity and package ability. Surprisingly, to our knowledge, only one strong result is known regarding Hamiltonicity of OTIS - showing that OTIS graph built of Hamiltonian base graphs are Hamiltonian. In this work we consider Hamiltonicity of OTIS networks, built on Non-Hamiltonian base and answer some important questions. First, we prove that Hamiltonicity of base graph is not a necessary condition for the OTIS to be Hamiltonian. We present an infinite family of Hamiltonian OTIS graphs composed on Non-Hamiltonian base graphs. We further show that, it is not sufficient for the base graph to have Hamiltonian path for the OTIS constructed on it to be Hamiltonian. We give constructive proof of Hamiltonicity for a large family of Butterfly-OTIS. This proof leads to an alternate efficient algorithm for independent spanning trees construction on this class of OTIS graphs. Our algorithm is linear in the number of vertices as opposed to the generalized algorithm, which is linear in the number of edges of the graph.

💡 Research Summary

The paper investigates Hamiltonicity and fault‑tolerance of Optical Transpose Interconnection Systems (OTIS), a two‑level swapped network widely used in parallel and distributed computing. An OTIS consists of n clusters, each isomorphic to a base graph G. While it is known that a Hamiltonian base graph guarantees a Hamiltonian OTIS, the converse has not been studied. The authors address three fundamental questions: (i) Is Hamiltonicity of the base graph necessary for the OTIS to be Hamiltonian? (ii) Is the existence of a Hamiltonian path in the base graph sufficient? (iii) Can we construct independent spanning trees (ISTs) efficiently on Hamiltonian OTIS instances?

Main contributions

1. Hamiltonicity without a Hamiltonian base – The authors define a generalized “bow‑tie” graph BF(m,n) consisting of two cycles C_m and C_n sharing a single cut‑vertex. When m and n are odd (i.e., BF(2m+1,2n+1) or BF(2m+1,2k)), the base graph is non‑Hamiltonian. Nevertheless, they prove that OTIS built on such bases is Hamiltonian. The proof proceeds by explicitly identifying a set of “key non‑Hamiltonian edges” inside each cluster, deleting them, and then applying two inference rules: (IR‑1) a vertex of degree ≥3 that becomes saturated forces the remaining incident edges to be non‑Hamiltonian; (IR‑2) if a non‑Hamiltonian edge connects two degree‑3 vertices, all other incident edges of those vertices must be Hamiltonian. By systematic application of these rules, a Hamiltonian cycle spanning all n² vertices is obtained. The authors also show that, because the minimum degree δ of the resulting OTIS is 2, at most ⌊δ/2⌋ = 1 edge‑disjoint Hamiltonian cycle can exist, confirming that the construction yields a unique Hamiltonian cycle.

2. Hamiltonian path is not sufficient – Counter‑examples are provided using BF(4,4) and BF(4,6). Both base graphs possess Hamiltonian paths, yet the corresponding OTIS graphs are proven to be non‑Hamiltonian. This demonstrates that a Hamiltonian path in the base does not guarantee Hamiltonicity of the OTIS, highlighting the intricate interplay between intra‑cluster and inter‑cluster connections.

3. Hamiltonicity of OTIS on the 4‑regular butterfly graph – The paper also treats the butterfly family BF(n) introduced in prior work, where vertices are labeled as (α; x_{n‑1},…,x_0) with α∈{0,…,n‑1} and each x_i∈{0,1}. This graph is 4‑regular and highly symmetric. The authors construct a Hamiltonian cycle for OTIS built on BF(n), extending the applicability of their methods beyond the bow‑tie family.

4. Linear‑time construction of two independent spanning trees – Leveraging the Hamiltonian cycle, the authors design an algorithm that produces two edge‑disjoint ISTs in O(|V|) time, where |V| is the number of vertices of the OTIS (i.e., n²). Traditional algorithms (e.g., Itai‑Rodeh) run in O(|E|) time, which can be substantially larger for dense OTIS instances. The new algorithm works by first deleting the identified non‑Hamiltonian intra‑cluster edges, then traversing the remaining structure directly. When at least one vertex of degree 2 remains in each cluster, the algorithm’s performance improves further. This result is significant for fault‑tolerant routing, as each IST provides a set of node‑disjoint paths guaranteeing resilience against processor or link failures.

5. Structural properties and fault‑tolerance – The paper restates known OTIS properties (degree increase by one, diameter doubling plus one, preservation of connectivity) and combines them with the new Hamiltonicity results. It confirms that OTIS can achieve maximal vertex‑disjoint path connectivity (k‑disjoint paths for a k‑connected base) while also supporting Hamiltonian cycles, thereby offering both high bandwidth and strong fault tolerance.

Implications and future work
The findings broaden the design space for OTIS networks: system architects are no longer constrained to Hamiltonian base graphs. The constructive proofs give explicit wiring patterns that can be implemented in hardware or simulated in software. Future research directions suggested include extending the Hamiltonicity analysis to broader classes of non‑Hamiltonian bases, exploring the existence of multiple edge‑disjoint Hamiltonian cycles (and consequently more than two ISTs), and experimentally validating the linear‑time IST algorithm on realistic optical‑electronic interconnect platforms with asymmetric fault models.