Shunji Abe

A traffic control method for service quality assurance in an ATM network

The authors describe an ATM traffic control method to avoid both long-term and short-term congestion. Call admission control and policing based on user-specified traffic parameters have mainly been studied as ways to avoid long-term congestion. However, these control methods are not practical since it is difficult for users to specify accurate values of traffic parameters that include the effects of cell collision controllers and multiplexers in the customer premises network. In order to solve the problem, they propose an intelligent call admission control method. This method consists of a function to determine users traffic parameters by monitoring the cell traffic flow of each service and a function to statistically predict long-term congestion by using Gaussian and Poisson distributions. They show that congestion can be predicted by deriving the peak, average, and variation. They also describe how to derive traffic parameters and determine their values. To avoid short-term congestion, we propose reactive control, which drops cells with lower cell loss priority when congestion occurs, and we show the buffer construction and buffer control algorithm used to achieve this. Computer simulations prove that their control strategy is effective for cell toss sensitive services such as VBR (variable bit rate) video service. >

Traffic Analysis of Delayed Delivery Services Considering the Correlation Between Circuit Switching and Message Switching

To improve the ratio of successful calls, a new switching system permitting callers to store messages when the called party is busy or does not answer can be constructed by adding a message storing function to the conventional circuit switching system. The mutual traffic influences exerted by this combination of circuit and message switching systems have not yet been clarified. This paper considers a two-stage loss system traffic model consisting of a circuit switching system that has a message switching system connected to it. The traffic characteristics of this model are analyzed using the nonindependent state transition probability, which depends upon the interacting states of the circuit and message switching system. A method of solving the state equations by determining the state transition probabilities of the message switching system from the state probabilities of the circuit switching system is proposed. Using this method, the mutual traffic influence of the circuit and message switching systems is evaluated.