Waiting time dynamics of priority-queue networks

We study the dynamics of priority-queue networks, generalizations of the binary interacting priority queue model introduced by Oliveira and Vazquez [Physica A {\bf 388}, 187 (2009)]. We found that the original AND-type protocol for interacting tasks is not scalable for the queue networks with loops because the dynamics becomes frozen due to the priority conflicts. We then consider a scalable interaction protocol, an OR-type one, and examine the effects of the network topology and the number of queues on the waiting time distributions of the priority-queue networks, finding that they exhibit power-law tails in all cases considered, yet with model-dependent power-law exponents. We also show that the synchronicity in task executions, giving rise to priority conflicts in the priority-queue networks, is a relevant factor in the queue dynamics that can change the power-law exponent of the waiting time distribution.

💡 Research Summary

The paper investigates the dynamical behavior of priority‑queue networks, extending the binary interacting priority‑queue model originally proposed by Oliveira and Vazquez (Physica A 388, 187 2009). The authors first examine the scalability of the original AND‑type interaction protocol, in which two connected queues can execute a joint task only when both simultaneously select that task. Through extensive simulations on networks containing loops, they demonstrate that priority conflicts accumulate, causing the system to become frozen: high‑priority tasks monopolize execution while other tasks are indefinitely postponed. This failure of the AND protocol highlights a fundamental limitation when moving from a pairwise setting to general network topologies.

To overcome this limitation, the authors introduce an OR‑type protocol. Under the OR rule, a joint task is executed as soon as at least one of the two interacting queues selects it, thereby dramatically reducing the chance of deadlock. The OR protocol is shown to be scalable across a variety of network structures and numbers of queues, making it a more realistic abstraction for human collaborative activities, computer packet scheduling, and distributed processing where perfect synchrony is rarely enforced.

The study systematically explores three representative topologies: (i) a fully connected graph, (ii) a one‑dimensional ring, and (iii) a scale‑free network generated by the Barabási–Albert algorithm. For each topology, the authors vary the number of queues (N = 5, 10, 20) and assign initial priorities from a uniform distribution. They then record the waiting time τ of each task—the number of time steps between its creation and execution—and compute the empirical distribution P(τ).

Across all configurations, P(τ) exhibits a power‑law tail, P(τ) ∼ τ^‑α, confirming that the heavy‑tailed waiting‑time phenomenon persists in networked priority queues. However, the exponent α is not universal; it depends on both the network topology and the interaction protocol. Under the OR protocol, α ranges roughly from 1.5 to 2.5. Fully connected networks tend to produce smaller α (≈1.6–2.0), indicating relatively shorter waiting times because each node has many alternative partners. Ring networks yield larger α (≈2.2–2.5), reflecting longer delays due to limited local connectivity. Scale‑free networks display the broadest variation: the presence of high‑degree hubs can lower α below 1.5, creating extremely long tails for tasks that must wait for hub‑mediated interactions.

A central contribution of the paper is the analysis of synchronicity in task execution. When many queues attempt to execute the same high‑priority task at the same time, the AND protocol forces all of them to wait for a perfect match, exacerbating congestion and leading to the frozen state observed in looped networks. By contrast, the OR protocol allows any one of the conflicting queues to proceed, thereby breaking the deadlock and restoring activity. The authors quantify how this reduction of priority conflicts shifts the power‑law exponent: higher degrees of synchrony produce larger variations in α, while asynchronous execution stabilizes the exponent around a narrower range.

The findings have several practical implications. In human‑centric environments such as email response, collaborative project management, or emergency dispatch, designing interaction rules that resemble the OR protocol—allowing tasks to be completed by any willing participant—can mitigate the buildup of unaddressed high‑priority items. In computer networks, packet‑scheduling algorithms that relax strict simultaneous acknowledgment requirements may achieve lower latency and avoid pathological queue freezing. Moreover, the sensitivity of α to network topology suggests that system architects should consider the underlying interaction graph when predicting performance; highly connected infrastructures can reduce waiting‑time tails, whereas sparsely connected or hub‑dominated structures may require additional mechanisms to prevent extreme delays.

The paper concludes by outlining directions for future work. Incorporating dynamic priority updates (e.g., aging of tasks), exploring partially asynchronous update schemes, and allowing the network itself to evolve in response to queue dynamics are identified as promising extensions. Such refinements would bring the model closer to real‑world settings where priorities shift over time and agents adapt their connections based on past performance. Overall, the study provides a robust theoretical framework for understanding how interaction protocols and network structure jointly shape the heavy‑tailed waiting‑time distributions observed in a wide range of complex systems.