This paper uses stochastic fluid models (SFMs) for control and optimization (rather than performance analysis) of communication networks, focusing on problems of buffer control. We derive gradient estimators for packet loss and workload related performance metrics with respect to threshold parameters. These estimators are shown to be unbiased and directly observable from a sample path without any knowledge of underlying stochastic characteristics, including traffic and processing rates (i.e., they are nonparametric). This renders them computable in online environments and easily implementable for network management and control. We further demonstrate their use in buffer control problems where our SFM-based estimators are evaluated based on data from an actual system. Index Terms-Communication network, perturbation analysis, stochastic fluid network.
This paper considers the problem of computing the schedule of modes in an autonomous switched dynamical system, that minimizes a cost functional defined on the trajectory of the system's continuous state variable. It proposes an algorithm that modifies a finite but unbounded number of modes at each iteration, whose computational workload at the various iterations appears to be independent on the number of modes being changed. The algorithm is based on descent directions defined by Gâteaux differentials of the performance function with respect to variations in mode-sequences, and its convergence to (local) minima is established in the framework of optimality functions and minimizing sequences, devised by Polak for infinite-dimensional optimization problems.
This paper develops an abstract framework for Infinitesimal Perturbation Analysis (IPA) in the setting of stochastic flow models, and it applies it to several problems arising in the study of flow control in single-server fluid-flow queues. The framework is based on a switched-mode hybrid-system paradigm, and especially on the interplay between its discrete-event dynamics and continuous-time dynamics. It is quite general, and most of the formulas obtained to-date for IPA on single-server queues can be derived from it as simple corollaries. Additional new results can be derived as well, and the paper demonstrates it by considering a queue with loss-rate-based flow control. The main contribution of the paper is in the proposed framework and its apparent broad scope. Its possible extension to a general class of fluid-flow queueing networks appears likely, and will be pointed out as a direction for future research.Index Terms-Fluid-flow queues, infinitesimal perturbation analysis (IPA), stochastic hybrid systems.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.