A fractional queueing-fluid model with N-policy vacations and power-law tail behavior

Lakshmi, M., Mathavavisakan, N. Micheal
OPSEARCH · Springer Nature · 2026-10-23 · DOI: 10.1007/s12597-026-01302-y

Аннотация

In this paper, we develop a comprehensive analytical framework for a fractional-order fluid queueing system with priority-based retrial arrivals, multiple exponential vacations, N -policy server activation and Poisson-driven catastrophic resets. This model integrates Caputo fractional derivatives of order $$\alpha \in (0,1)$$ with respect to the buffer level to capture memory-dependent fluid accumulation, while incorporating a two-class priority discipline ( $$\text {Class 1} \succ \text {Class 2} \succ \text {Retrial}$$ ) and orbit-based retrial mechanisms. Further, using the supplementary variable method combined with Laplace-generating function techniques, we derive the steady-state Laplace transform of the buffer content distribution, along with explicit expressions for the mean buffer occupancy, variance, tail asymptotics and transient dynamics governed by Mittag-Leffler relaxation kernels. Moreover, the key theoretical contributions of this work include: (i) a closed-form characterization of the power-law tail decay $$\mathbb {P}\{C > w\} \sim w^{-\alpha }$$ induced by fractional spatial memory; (ii) a rigorous sensitivity analysis demonstrating that increasing $$\alpha $$ strictly reduces the expected buffer content and convergence time; (iii) optimization formulas for the N -policy threshold $$N^*$$ and buffer capacity $$K^*$$ , balancing holding costs against overflow penalties under fractional-catastrophic dynamics and (iv) a unified analytical framework that recovers classical queueing-fluid models as special limiting cases. In addition, extensive numerical experiments, supported by figures and summary tables generated using MATLAB R2024a, validate the theoretical findings and reveal that fractional memory amplifies tail risks, prolongs transient convergence and necessitates larger safety margins for quality-of-service guarantees. Finally, the proposed model provides practical insights for the design and operation of energy-aware cloud computing centers, self-similar communication networks, resilient manufacturing systems and patient-flow management in healthcare operations, where long-range dependence and sudden disruptions coexist.

Читать у издателя →

Похожие статьи