Random signals have been used as test signals in various fields. Among the various types of random signals, a white and normally-distributed random signal is most frequently used. A normal random signal is usually generated on the basis of the central limit theorem. This theorem is approximately realized by adding n binary random numbers with a weighted adder. The purpose of this paper is to find out the optimum sets of n weights of the weighted adder which can generate a best-approximated normal random signal on condition that n is constant. First the characteristic function of a distribution obtained by the adder is represented as the function of the n weights. And the low order cumulants of the distribution are led from the characteristic function. These cumulants are best-approximated to those of a normal distribution, when all weights have the same value. The uniformity of the weights, however, make an obtained distribution discrete. Therefore the effect of the dispersion of the weights is taken into consideration. The characteristic function obtained from dispersed weights can be expressed by the nominal value of the weights and the variance of the dispersion. The optimum variances of the weights are determined for several n's from minimization of difference between the two characteristic functions of an obtained distribution and a normal one. This result shows that economic rough resistors should be used as the input resistances of the weighted adder.
A practical Josephson junction array voltage standard system which is available to calibrate 1.018V directry has been satisfactorily working at the manfacturing company. The uncertainty of the system is estimated as 0.01 ppm, as well as the national voltage standard system of Electrotechnical Laboratory. Calibration results between the practical Josephson voltage standard and the national voltage standard agree within 0.02 ppm.
Carrier Sense Multiple Access with Collision Detection (CSMA/CD) protocol is one of remarkable access protocol on Local Area Network (LAN). The recent interest in the application of LAN is the interconnection of LANs. Then the performance of the system of a CSMA/CD LAN interconnected by bridge with finite buffer is investigated. We present an analysis of slotted nonpersistent CSMA/CD LAN for terminals and one bridge using Tobagi's method. Terminals and bridge generate a packet with different access rate respectively. This analysis is a practical analysis by describing all status of backlogged packets in terminals and bridge. We have compared analytical results with simulation results. The points of consideration are throughput-delay characteristics and probability of occupancy of bridge. We evaluate the buffer capacity of bridge from the results of the rejection probability of voice packets and data packets. The analytical results is very close to the simulation results.
Recently, electric power systems are facing increasing uncertainties in demand, energy price and environmental constraints in future. In such a situation, a plan for expansion of the system capacity must be made robust against the aforesaid uncertainties. Large-scale power plants require long time for construction, so that, the decision of construction must be made under much uncertainty in future. On the other hand, small-scale plants require less time for construction, i.e., the decision can be made under less uncertainty in spite of their relatively higher costs. The present paper is concerned with an optimal combination of the large-scale and the small-scale plants having the aforesaid characteristics under uncertainty of the demand in future. First, the demand growth in future is described by scenarios of demand growth branching like a tree. Then, the optimization problem of the system expansion is formulated into a stochastic linear-programming problem. An optimal solution of the problem is obtained by using the scenario-aggregation algorithm proposed by Rockafellar and Wets. The simulation results, yielded by using a parallel computation on Transputers, show that there is a possibility of constructing the small-scale power plants to cope with the uncertainty even if they are more expensive than the large-scale plants.
The backpropagation through time, a learning algorithm for recurrent networks, has a drawback of being unable to provide structural information, such as locations of feedback loops in a network. I have previously proposed a structural learning algorithm with forgetting of link weights for hierarchical networks. By eliminating unnecessary links due to forgetting, the structural learning can generate a skeletal structure reflecting inherent regularity in training patterns. Proposed here is a general structural learning algorithm with forgetting for recurrent networks by combining the above two algorithms. This algorithm can generate a skeletal structure for recurrent networks based on input and output sequences with no prior structural information. Four kinds of Jordan networks are selected as typical examples of recurrent networks including feedback and self loops. The proposed algorithm can, in most cases, rediscover the original Jordan network structure solely from input and output sequences. This well demonstrates the effectiveness of the proposed algorithm in rediscovering the original network structure for recurrent networks.