The behavior of line concentrators in a data communication system has been analyzed for different channel allocation strategies. The users (terminals) are divided into classes according to their transmission rate requirements. The different bit rate classes (BRC) can be separated by dividing the network into subnets or, alternatively, they can share one concentrator which forms outgoing channels from a common quantity of channels of a basic speed, or by splitting a high speed channel into two or more channels with lower transmission rate. The equations of state and analytical expressions for the time congestion factors have been developed for the different system models. The numerical results show that a joint concentrator arrangement requires less channel capacity than a separated concentrator system for a fixed value of the congestion probability. The results of this investigation may serve as background material for planning future data networks.

An analysis of the discrete-time queuing problem: binomial input, constant message length, infinite queue length, one server, is carried out. The state probabilities are calculated using the method of imbedded Markov chains, and the results are applied to the problem of determining characteristic features in the performance of a traffic concentrator. The results are compared to the direct numerical solution of the equations of state for a finite buffer queue concentrator.