The Design of Genetically Optimized Self-Organizing Neural Networks with Polynomial and Fuzzy Polynomial Neurons

In this study, we introduce and investigate a class of neural architectures of self-organizing neural networks (SONNs) that is based on a genetically optimized multilayer perceptron with polynomial neurons (PNs) or fuzzy polynomial neurons (FPNs), develop a comprehensive design methodology involving mechanisms of genetic optimization, and carry out a series of numeric experiments. We distinguish between two kinds of SONN architectures: (a) PN-based and (b) FPN-based SONNs. The augmented genetically optimized SONN (gSONN) results in a structurally optimized structure and comes with a higher level of flexibility in comparison to the one encountered in the conventional SONN. The genetic algorithm (GA)-based design procedure being applied at each layer of SONN leads to the selection of preferred nodes (PNs or FPNs) with specific local characteristics (such as the number of input variables, the order of the polynomial, and a collection of the specific subset of input variables) available within the network.     

Technical foundations for noninvasive assessment of changes in the width of the subarachnoid space with near-infrared transillumination-backscattering sounding (NIR-TBSS)

This paper presents technical foundations for a new technique of near-infrared transillumination-backscattering sounding, which is designed to enable noninvasive detection and monitoring of changes in the width of the subarachnoid space (SAS) and magnitude of cerebrovascular pulsation in humans. The key novelty of the technique is elimination of influence of blood flow in the scalp on the signals received from two infrared sensors-proximal and distal. A dedicated digital algorithm is used to estimate on line the ratio of the powers of received signals, referred to as two-sensor distal-to-proximal received power quotient, TQ (t). The propagation duct for NIR radiation reaching the distal sensor is the SAS filled with translucent cerebrospinal fluid. Information on slow fluctuations of the average width of the SAS is contained in the slow-variable part of the TQ (t), called the subcardiac component, and in TQ itself. Variations in frequency and magnitude of faster oscillations of the width of that space around the baseline value, dependent on cerebrovascular pulsation, are reflected in instantaneous frequency and envelope of the fast-variable component. Frequency and magnitude of the cerebrovascular pulsation depend on the action of the heart, so this fast-variable component is referred to as the cardiac component.     

An application of scanning electron microscopy combined with roentgen microanalysis (SEM-EDS) in canine urolithiasis

Urolithiasis is a common diagnostic and therapeutic problem in small-animal veterinary practice. The traditional diagnostic approach usually consists of clinical, radiological and ultrasonographic examination of the patient. The main diagnostic material is still urine sediment, ignoring the fact that presence of crystalluria is not always of pathological significance. In order to establish the most effective therapeutic and preventative strategies, especially in the case of multicomponent stone, it is crucial to define the exact elemental composition of the given stone including crystallization nidus chemical contents. In the course of the research, the usefulness of scanning electron microscopy combined with X-ray-dispersive spectrometry in analysis of canine mixed and compound stones was investigated. The obtained results indicated that the tested method allows one to trace the dynamics of the crystallization process, including crystallization nucleus detection, and concurrently and quantitatively assess the elemental composition of the given urinary concrement. Moreover, the conducted research showed epidemiological data of urolithiasis occurrence in a population of dogs coming from the southern part of Poland.     

Hybrid DPWM implementation using coarse and fine programmable ADLL

In the paper we propose a novel architecture and implementation of 11-bit Digital Pulse Width Modulator (DPWM) circuit based on previously known building blocks. Linearized Class-AD Double-sided (LADD) algorithm has been used to calculate the DPWM signals of the 11-bit resolution hybrid DPWM for a Class-AD digital audio amplifier. Noise-shaping process is used to support high fidelity with practical values of time resolution. The proposed DPWM circuit is composed of 8-bit counter and Analog Delay Locked Loop (ADLL) using 4-bit tapped delay line. A dual ADLL employing coarse and fine programmable delay element is used to adjust the delay time of delay line and lock it to required time. The coarse- as well as fine-delay lines are implemented as a cascade of variable-delay elements based on shunt capacitor delay element or single-ended Schmitt trigger. The proposed 11-bit DPWM circuit, at a switching frequency of 352.8 kHz and clock generator frequency of 90.3 MHz allows us to attain SNR of 120 dB and THD of the output signal less than 0.1% within the audio baseband and modulation index M=0.95. Basic verification of circuit manufacturability and simulation results (Monte Carlo analysis) for real CMOS process are presented.     

Microvaristors in thick-film and LTCC circuits

ZnO-based varistors protect electronic circuits against overvoltage. High temperature from the range of 1150-1300 °C is required for proper sintering of such material. Varistor inks with lower firing temperature are needed for application in thick-film and LTCC technology. ZnO-based thick-film composition was prepared and varistors were fabricated on alumina and LTCC substrate. Different topologies (capacitor-like or planar), electrode metallurgies (PdAg, Au or Pt-based) and firing profiles (850 °C or 950 °C) were used. Samples microstructure was investigated. Varistor I-V characteristics, long-term stability and durability to high voltage pulses were examined. Satisfactory results were achieved, because nonlinearity coefficient α up to 23 was obtained for capacitor-like varistors with Pt terminations on LTCC substrates, long-term thermally aged (150 h at 250 °C) varistors had slightly smaller nonlinearity coefficient and characteristic voltage, V_1 mA and components subjected to series of high voltage pulse (1000 pulses with 10 mA amplitude and 5 ms duration each) exhibited almost the same electrical parameters.     

A multiobjective design of a patient and anaesthetist-friendly neuromuscular blockade controller

During surgeries (especially in long ones), patients are subject to a substantial amount of drug dosage necessary to achieve the required neuromuscular blockade level. This paper aims at the development of a fuzzy controller that satisfies two important goals: 1) an optimization of the amount of drug (atracurium) required to induce an adequate level of relaxation and 2) a concomitant ability to explain the undertaken control decision at the level of natural language. For instance, statements of the form "Since the difference between the target and the current blockade level is near zero, a small quantity of drug infusion is currently being applied", where "near zero" and "small" are linguistic terms that are represented as fuzzy sets. In this sense, we can regard this controller as a construct that is human friendly and highly interpretable (transparent). To address the two objectives outlined above, we consider the use of a multiobjective evolutionary optimization. How the quality of the control action and the controller interpretability are formalized and captured in this optimization framework is presented. The effectiveness of the approach is demonstrated through a comprehensive suite of experiments involving 100 simulated patients (used for training) and 500 patients (forming the test set), validating the approach for application in the operating theater.     

Scheduling Algorithms and Throughput Maximization for the Downlink of Packet-Data Cellular Systems with Multiple Antennas at the Base Station

The capacity-achieving coding scheme for the multiple-input multiple-output (MIMO) broadcast channel is dirty-paper coding. With this type of transmission scheme the optimal number of active users that receive data and the optimal power allocation strategy are highly dependent on the structure of the channel matrix and on the total transmit power available. In the context of packet-data access with adaptive transmission where mobile users are equipped with a single receive antenna and the base station has multiple transmit antennas, we study the optimal number of active users and the optimal power allocation. In the particular case of two transmit antennas, we prove that the optimal number of active users can be a non-monotonic function of the total transmit power. Thus not only the number of users that should optimally be served simultaneously depends on the user channel vectors but also on the power available at the base station transmitter. The expected complexity of optimal scheduling algorithms is thus very high. Yet we then prove that at most as many users as the number of transmit antennas are allocated a large amount of power asymptotically in the high-power region in order to achieve the sum-capacity. Simulations confirm that constraining the number of active users to be no more than the number of transmit antennas incurs only a marginal loss in spectral efficiency. Based on these observations, we propose low-complexity scheduling algorithms with sub-optimal transmission schemes that can approach the sum-capacity of the MIMO broadcast channel by taking advantage of multiuser diversity. The suitability of known antenna selection algorithms is also demonstrated. We consider the cases of complete and partial channel knowledge at the transmitter. We provide simulation results to illustrate our conclusions.     

Scalability and performance of an agent‐based network management middleware

Rapid growth of computer network sizes and uses necessitate analysis of network application middleware in terms of its scalability as well as performance. In this paper we analyze a distributed network management middleware based on agents that can be dispatched to locations where they can execute close to the managed nodes. The described middleware operates between the network protocol layer and the application layer and uses standard TCP protocol and SNMP probes to interface the network. By aggregating requests from many users into a single agent, our system allows multiple managers to probe problem areas with minimal management traffic overhead. We discuss and quantify the benefits of the described middleware by implementing real‐time network managers using our system. The main result of this paper is a comparison of scalability and efficiency of our agent‐based management middleware and traditional SNMP‐based data collection. To this end, we measured traffic in both real and simulated networks. In the latter case, we designed, used and described here a method of separating simulated application flow into separate subflows to simplify design of simulations. Copyright © 2004 John Wiley & Sons, Ltd.     

Interrelations between initial pore structure, morphology and distribution of accumulated deposits, and lifetimes of hydrodemetallisation catalysts

The functioning of commercial HDM catalysts, experiencing conditions of industrial use in residual feedstocks upgrading reactors is analysed, in order to derive the essential physical and chemical processes that determine their deactivation. A mathematical model of this evolution is presented as an attempt to integrate this analysis into computer simulations for improved reliability of lifetimes predictions. A method to obtain accelerated aging data on HDM catalysts at the bench scale is presented, with examples of results. A method is given to combine such results with EPMA analysis of used catalyst pellets in order to determine the Ultimate Storage Capacity intrinsic to the catalyst's pore structure. Selected results of an High Resolution Electron Microscopy study of microtome thin foiled used HDM catalyst pellets are presented: they show for the first time the nature, morphology and structure of metal sulfides deposits at the micropore level, and support the “discrete” mode of deposition theory. Model predictions and experimental deactivation results are compared, and the limitations of the model discussed. Unsolved questions are listed for future research.