Network users and designers are frequently concerned about the fairness of service achieved either in the local network or in internetworking. The authors compile the basic issues, notions, and definitions of fairness in both the internetworking and intranetworking contexts from the available literature. The optimization schemes that ensure fairness and the criteria used for such optimization, as available in the internetworking literature, are discussed. In the intranetworking environment, delay and throughput are often used as the fairness parameters. Often the approach to fairness entirely depends on the protocol implementation, and hence fairness criteria should be uniquely identified for each protocol. >

https://ieeexplore.ieee.org/iel2/642/6695/00268128.pdf

Access to a network channel: a survey into the unfairness problem

A flow control algorithm for virtual circuits is proposed which takes into account the fact that in a packet switched network, different users can have different requirements for throughput and average delay. The algorithm is based on a new definition of power based on multiple objective optimization theory. Problems of approximation and convergence are discussed, and a numerical example is given. >

An approach to user requirements-oriented flow control

Conventionally, the problem of capacity-flow assignment in packet switching computer networks has been considered mainly for the purpose of improving the network performance and the effective usage of the resources in the networks. The problem of the fair allocation of the network resources to each user (fairness) has been investigated little. Because one of the most important purposes of the computer network is that a plural number of users can use the network resources which are arranged dispersively in the network, we must make sufficient study of fairness in the problem of the flow assignment of the packet and the problem of the distribution of the line capacity. However, no clear definition of the fair allocation has been proposed to dare. In this paper, we adopt the “power” (function) as the performance measure and discuss the problem of the network capacity-flow assignment in which maximized network. power ensures effective usage of networks, and equalized power for each user realizes the fairness. The results of this study provide a new guideline to solve the problem of distributing the resources effectively and fairly. The method in this paper can be discussed generally without limitation of the network topology. However, since the concrete numerical calculation is so complicated, we take loop networks for the object of the analysis.

Optimum capacity-flow assignment problems with respect to fairness in loop networks

Learning routing is introduced to large-scale communication network to compensate the lack of a priori information on the network behavior. Learning routing consists of learning automaton as a learning controller placed at each network node. This learning automaton makes decision on routes for packets which are the basic unit of data to be transmitted through the network. Using learning automaton we propose two routing methods, LRI and LRI, which are basically reinforcement methods for the learning automaton. The effectiveness of these proposed methods is shown by means of simulation on a concrete network topology.

Learning routing applied to large scale packet–switched communication networks

The receiving behavior of a window mechanism, an important function of flow control in packet switching computer networks, is modeled by a 2-stage tandem queueing system with total capacity limitation. In queueing analysis, it is difficult to obtain analytical solutions for such a limited queueing system because the local balancing equations do not hold. This paper proposes two approximate analytical methods utilizing the state transition rate diagram structure: simple analysis method and heuristic approximation method. the former gives the upper and lower bounds of the steady-state probabilities of the given system, the latter obtains the approximate analytical solutions directly. the properties of the two methods, applicability and the relation between the two methods, are clarified from numerical and simulated results and it is shown that the two analysis methods are effective approximation methods. Throughput, average system size and average waiting time are calculated and it is shown that the two approximation methods are very useful for evaluating the performance measures of the window mechanism.

Tadao Nakamura

Papers

Optimum capacity-flow assignment problems with respect to fairness in loop networks

Learning routing applied to large scale packet–switched communication networks

Analysis of receiving behavior of a window flow control mechanism in packet switching networks