There have been many investigations of mutual synchronization of oscillators. In this article, N oscillators with the same natural frequencies mutually coupled by one resistor are analyzed. In this system, various synchronization phenomena can be observed because the system tends to minimize the current through the coupling resistor. When the nonlinear characteristics are third-power, we can observe N-phase oscillation, and this system can take (N 1)! phase states. When the nonlinear characteristics are fifth-power, we can observe (N 1),(N 2)3 and 2-phase oscillations as well as N-phase oscillations and we can get much more phase states from this system than that of the system with third-power nonlinear characteristics. Because of their coupling structure and huge number of steady states of the system, our system would be a structural element of cellular neural networks. In this study, it is confirmed that our systems can stably take huge number of phase states by theoretical analysis, computer calculations and circuit experiments.

In this paper, we propose the adaptive cutoff priority for a multiple server finite-capacity buffer system with two kinds of arrival streams to reduce loss probabilities of both kinds of arrivals. The number of reserved" servers for high priority customers is adaptively decreased by using the adaptive cutoff priority, when the number of waiting customers of low priority is increased. We first derive the state probabilities which are the function of both the number of customers in the queue and the number of customers in service, so that it is possible to present general expressions for the loss probability and the average queueing time. Next, the numerical calculation is carried out and it is found that not only the loss probability of each arrival but slso the total loss probability are reduced by using more than two thresholds and selecting the treshold values appropriately. It is shown that the computer simulated results agree with the theoretical results.