Time-Varying Graphs and Dynamic Networks

The past few years have seen intensive research efforts carried out in some apparently unrelated areas of dynamic systems – delay-tolerant networks, opportunistic-mobility networks, social networks – obtaining closely related insights. Indeed, the concepts discovered in these investigations can be viewed as parts of the same conceptual universe; and the formal models proposed so far to express some specific concepts are components of a larger formal description of this universe. The main contribution of this paper is to integrate the vast collection of concepts, formalisms, and results found in the literature into a unified framework, which we call TVG (for time-varying graphs). Using this framework, it is possible to express directly in the same formalism not only the concepts common to all those different areas, but also those specific to each. Based on this definitional work, employing both existing results and original observations, we present a hierarchical classification of TVGs; each class corresponds to a significant property examined in the distributed computing literature. We then examine how TVGs can be used to study the evolution of network properties, and propose different techniques, depending on whether the indicators for these properties are a-temporal (as in the majority of existing studies) or temporal. Finally, we briefly discuss the introduction of randomness in TVGs.

💡 Research Summary

The paper addresses the fragmented landscape of research on networks whose topology changes over time, encompassing delay‑tolerant networks, opportunistic mobility networks, and social contact networks. Although each community has developed its own models and terminology, the underlying concepts are closely related. To unify these efforts, the authors introduce the Time‑Varying Graph (TVG) formalism. A TVG consists of a static vertex set V and edge set E together with two time‑dependent functions: a presence function ρ(e, t) indicating whether edge e exists at time t, and a labeling function ζ(e, t) that can encode latency, capacity, or any other edge attribute. This compact definition captures both the structural and temporal dimensions of a dynamic network and can represent all previously proposed models as special cases.

Building on this foundation, the paper proposes a hierarchical classification of TVGs based on four fundamental properties that have been central in distributed computing literature. The first class, temporal connectivity, concerns whether a time‑respecting path exists between any pair of nodes over the observation interval. The second class, minimum‑delay, focuses on the earliest arrival time between node pairs, enabling the formulation of latency‑optimal routing problems. The third class, dynamic isomorphism, formalizes how the network’s topology evolves, allowing the identification of recurring structural patterns. The fourth class introduces randomness by treating ρ and ζ as stochastic processes, thereby modeling uncertainty in edge appearance and attributes. Each class subsumes existing results, showing that many known theorems are special instances of more general statements within the TVG framework.

The authors then turn to the analysis of network property evolution. They distinguish between a‑temporal indicators—traditional static metrics such as degree, clustering coefficient, or global efficiency computed on successive time slices—and truly temporal indicators that incorporate the ordering of events. Temporal centrality, temporal path length, and temporal community structure are defined directly from the presence and labeling functions, providing a finer‑grained view of information flow and structural change. The paper outlines algorithmic techniques for computing both families of metrics, emphasizing the trade‑off between computational simplicity and expressive power.

Finally, the paper sketches how randomness can be integrated into TVGs. By defining probabilistic presence and labeling functions, one can derive expected values for temporal metrics, establish concentration bounds, and study phase transitions in connectivity or epidemic spread. This stochastic extension opens the door to rigorous analysis of networks where mobility, link failures, or user behavior are inherently uncertain.

In summary, the contribution of the work is threefold: (1) a unifying mathematical model that captures the essence of a wide range of dynamic networks; (2) a systematic classification that links TVG properties to well‑studied concepts in distributed computing; and (3) a set of methodological tools for both deterministic and probabilistic analysis of temporal network characteristics. The TVG framework thus provides a common language for future research, facilitating cross‑disciplinary comparison, algorithm design, and theoretical advances in the study of dynamic systems.