Distributed memory systems for simulating artificial neural networks

In executing tasks involving intelligent information processing, the human brain performs better than the digital computer. The human brain derives its power from a large number [O(10¹¹)] of neurons which are interconnected by a dense interconnection network [O(10⁵) connections per neuron]. Artificial neural network (ANN) paradigms adopt the structure of the brain to try to emulate the intelligent information processing methods of the brain. ANN techniques are being employed to solve problems in areas such as pattern recognition, and robotic processing. Simulation of ANNs involves implementation of large number of neurons and a massive interconnection network. In this paper, we discuss various simulation models of ANNs and their implementation on distributed memory systems. Our investigations reveal that communication-efficient networks of distributed memory systems perform better than other topologies in implementing ANNs.     

Mössbauer spectroscopic studies of Fe-substituted YIG

Compositional dependence of hyperfine parameters, determined through Mössbauer spectral analysis has been studied for Y_3−xFe_5+xO₁₂ (x = 0.0, 0.1, 0.3 and 0.5) garnet system at 300 K. The Mössbauer spectra have been fitted with three sextets in the ferrimagnetic state corresponding to Fe³⁺ ions at tetrahedral (d), octahedral (a) and dodecahedral (c) sites of the crystal structure. It is observed that isormershift, quadrupole shift and hyperfine field of d-site show no significant variation with Fe³⁺ concentration. The change in hyperfine fields of a- and c-sites with composition (x) has been explained on the basis of strength of exchange integrals, change in isomershift can be understood due to s-electron charge distribution and asymmetric displacement of oxygen ions surrounding the a- and c-sites seems to be responsible for observable quadrupole shift. The magneton number values obtained from magnetization and Mossbauer data are in agreement to those calculated using Neel's three sublattice model of ferrimagnetism.     

An integrated modeling approach to predict flooding on urban basin.

Correct prediction of flood extents in urban catchments has become a challenging issue. The traditional urban drainage models that consider only the sewerage-network are able to simulate the drainage system correctly until there is no overflow from the network inlet or manhole. When such overflows exist due to insufficient drainage capacity of downstream pipes or channels, it becomes difficult to reproduce the actual flood extents using these traditional one-phase simulation techniques. On the other hand, the traditional 2D models that simulate the surface flooding resulting from rainfall and/or levee break do not consider the sewerage network. As a result, the correct flooding situation is rarely addressed from those available traditional 1D and 2D models. This paper presents an integrated model that simultaneously simulates the sewerage network, river network and 2D mesh network to get correct flood extents. The model has been successfully applied into the Tenpaku basin (Nagoya, Japan), which experienced severe flooding with a maximum flood depth more than 1.5 m on September 11, 2000 when heavy rainfall, 580 mm in 28 hrs (return period > 100 yr), occurred over the catchments. Close agreements between the simulated flood depths and observed data ensure that the present integrated modeling approach is able to reproduce the urban flooding situation accurately, which rarely can be obtained through the traditional 1D and 2D modeling approaches.     

A genetic‐based neuro‐fuzzy controller for blind equalization of time‐varying channels

This paper presents a neuro‐fuzzy network (NFN) where all its parameters can be tuned simultaneously using genetic algorithms (GAs). The approach combines the merits of fuzzy logic theory, neural networks and GAs. The proposed NFN does not require a priori knowledge about the system and eliminates the need for complicated design steps such as manual tuning of input–output membership functions, and selection of fuzzy rule base. Although, only conventional GAs have been used, convergence results are very encouraging. A well‐known numerical example derived from literature is used to evaluate and compare the performance of the network with other equalizing approaches. Simulation results show that the proposed neuro‐fuzzy controller, all parameters of which have been tuned simultaneously using GAs, offers advantages over existing equalizers and has improved performance. From the perspective of application and implementation, this paper is very interesting as it provides a new method for performing blind equalization. The main contribution of this paper is the use of learning algorithms to train a feed‐forward neural network for M‐ary QAM and PSK signals. This paper also provides a platform for researchers of the area for further development. Copyright © 2008 John Wiley & Sons, Ltd.     

Degradation of Stiffness of Cemented Calcareous Soil in Cyclic Triaxial Tests

The deformation characteristics of artificially cemented calcareous soil subjected to undrained cyclic triaxial loading are investigated at different confining pressure and cyclic stress levels. The influence of cementation on the shear stiffness is investigated by comparing the behavior of cemented and uncemented soils with similar initial conditions. It is observed that the deviator stress and the deviatoric strain at yield reduced with increasing number of cycles for cemented sand due to progressive degradation of bond, which results in significant decrease in stiffness. On the other hand, a strain-hardening effect is observed in uncemented sand and this results in increasing yield stress and strain with progressive number of cycles. A linear relationship between degradation index and number of cycles is observed for cemented sand. This relationship has been synthesized in the form of an empirical equation by modifying a previously proposed equation for cohesive soils. This empirical equation was further used to evaluate the fatigue life of soils by adopting a failure criterion.     

Plastic Limit, Liquid Limit and Undrained Shear Strength of Soil—Reappraisal

The concept that plasticity index of soils can be defined as a range of water contents producing a 100-fold variation in undrained shear strength has been experimentally verified with the help of a large number of tests on soils of diverse nature. This has led to the redefinition of the plastic limit as the water content at which undrained shear strength is around 170 kN/m2. Undrained shear strength of a soil at the liquid limit can be considered to be around 1.7 kN/m2. Accordingly, both the liquid limit and the plastic limit have been determined in the present work by a single consistent method, i.e., the Swedish fall cone method. The undrained shear strength-water content relationship has been found to be log-linear for a wide range of water contents beginning from lower than the plastic limit to higher than the liquid limit. This resulted in the formulation of an expression for predicting undrained shear strength of a remolded soil at any water content based solely on its liquid limit and plastic limit.     

下一代SONET/SDH对网络部署的冲击

面对电信业的低迷,运营商和设备制造商都捏紧了钱袋,想方设法少花钱多办事。这就意味着,必须改进现有传输网(见图1),使话音和数据传输效率更高。从网络部署的角度来看,运营商需要:用一些有效的方法,将多协议数据转换到SONET/SDH信道上;用一种灵活的带宽适应机制,使SONET/SDH信道大小与在其中传输的数据的需求相一致;用一种机制,灵活并以理想的动态方式改变整个SONET/SDH网络的信道带宽。这就要求集成电路供应商提供有助于在SONET/SDH上有效传输数据的芯片。虽然现在有多种SONET/SDH信息包(PoS)产品,但是它们的转换效率都太…     

The sliding-window packet switch: a new class of packet switch architecture with plural memory modules and decentralized control

Shared-memory based packet switches are known to provide the best possible throughput performance for bursty data traffic in high-speed packet networks and internets compared with other buffering strategies under conditions of identical memory resources deployed in the switch. However, scaling of shared-memory packet switches to a larger size has been restricted mainly due to the physical limitations imposed by the memory-access speed and the centralized control for switching functions in shared-memory switches. A new scalable architecture for a shared-memory packet switch, called the sliding-window (SW) switch, is proposed to overcome these limitations. The SW switch introduces a new class of switching architecture, where physically separate multiple memory modules are logically shared among all the ports of the switch, and the control is decentralized. The SW switch alleviates the bottleneck caused by the centralized control of switching functions in large shared-memory switches. Decentralized switching functions enable the SW switch to operate in a pipeline fashion to enhance scalability and switching capacity compared with that of previously known classes of shared-memory switch architecture.     

Transparent ultra-long-haul DWDM networks with "broadcast-and-select" OADM/OXC architecture

We describe an experimental realization of ultra-long-haul (ULH) networks with dynamically reconfigurable transparent optical add-drop multiplexers (OADMs) and optical cross-connects (OXCs). A simple new approach to dispersion management in ULH dense-wavelength-division-multiplexing (DWDM) transparent optical networks is proposed and implemented, which enables excellent transmission performance while avoiding dispersion compensation on a connection-by-connection basis. We demonstrate "broadcast-and-select" node architectures that take full advantage of this method. Our implementation of signal leveling ensures minimum variations of path-averaged power among the wavelength-division-multiplexing (WDM) channels between the dynamic gain-equalizing nodes and results in uniform nonlinear and spontaneous-emission penalties across the WDM spectrum. We achieve 80/spl times/10.7-Gb/s DWDM networking over 4160 km (52 spans/spl times/80 km each) of all-Raman-amplified symmetric dispersion-managed fiber and 13 concatenated OADMs or 320/spl times/320 wavelength-port OXCs with 320-km node spacing. The WDM channels use 50-GHz grid in C band and the simple nonreturn-to-zero (NRZ) modulation format. The measured Q values exhibit more than a 1.8-dB margin over the forward-error correction threshold for 10/sup -15/ bit-error-rate operation. We compare these results with point-to-point transmission of 80/spl times/10-Gb/s NRZ WDM signals over 4160 km without OADM/OXC and provide detailed characterization of penalties due to optical signal-to-noise-ratio degradation, filter concatenation, and crosstalk.