Power-Capped DVFS and Thread Allocation with ANN Models on Modern NUMA Systems

Abstract—Power capping is an essential function for efficient power budgeting and cost management on modern server systems. Contemporary server processors operate under power caps by using dynamic voltage and frequency scaling (DVFS). However, these processors are often deployed in non-uniform memory
access (NUMA) architectures, where thread allocation between cores may significantly affect performance and power consumption. This paper proposes a method which maximizes performance under power caps on NUMA systems by dynamically optimizing two knobs: DVFS and thread allocation. The method selects the optimal combination of the two knobs with models based on artificial neural network (ANN) that captures the nonlinear effect of thread allocation on performance. We implement
the proposed method as a runtime system and evaluate it with twelve multithreaded benchmarks on a real AMD Opteron based NUMA system. The evaluation results show that our method outperforms a naive technique optimizing only DVFS by up to
67.1%, under a power cap.

Voltage over-scaling:A cross-layer design perspective for energy efficient systems

Today's multi-media electronic era is driven by the increasing demand for small multifunctional devices able to support diverse services. Unfortunately, the high levels of transistor integration and performance required by such devices lead to an unprecedented increase of on-chip power that significantly limits the battery lifetime and even poses reliability concerns. Several techniques have been developed to address the power increase, but voltage over-scaling (VOS) is considered to be one of the most effective ones due to the quadratic dependence of voltage on dynamic power consumption. However, VOS may not always be applicable since it increases the delay in all paths of a system and may limit high performance required by today's complex applications. In addition, application of VOS is further complicated since it increases the variations in transistor characteristics imposed by their tiny size which can lead to large delay and leakage variations, making it difficult to meet delay and power budgets. This paper presents a review of various cross-layer design options that can provide solutions for dynamic voltage over-scaling and can potentially assist in meeting the strict power budgets and yield/quality requirements of future systems. © 2011 IEEE.

A computational intelligent approach to multi-factor analysis of violent crime information systems

Various scientific studies have explored the causes of violent behaviour from different perspectives, with psychological tests, in particular, applied to the analysis of crime factors. The relationship between bi-factors has also been extensively studied including the link between age and crime. In reality, many factors interact to contribute to criminal behaviour and as such there is a need to have a greater level of insight into its complex nature. In this article we analyse violent crime information systems containing data on psychological, environmental and genetic factors. Our approach combines elements of rough set theory with fuzzy logic and particle swarm optimisation to yield an algorithm and methodology that can effectively extract multi-knowledge from information systems. The experimental results show that our approach outperforms alternative genetic algorithm and dynamic reduct-based techniques for reduct identification and has the added advantage of identifying multiple reducts and hence multi-knowledge (rules). Identified rules are consistent with classical statistical analysis of violent crime data and also reveal new insights into the interaction between several factors. As such, the results are helpful in improving our understanding of the factors contributing to violent crime and in highlighting the existence of hidden and intangible relationships between crime factors.

Dynamic identification of building structures: new strategies:Meccanica dei materiali e delle strutture

In this paper the evolution of a time domain dynamic identification technique based on a statistical moment approach is presented. This technique can be used in the case of structures under base random excitations in the linear state and in the non linear one. By applying Itoˆ stochastic calculus, special algebraic equations can be obtained depending on the statistical moments of the response of the system to be identified. Such equations can be used for the dynamic identification of the mechanical parameters and of the input. The above equations, differently from many techniques in the literature, show the possibility of obtaining the identification of the dissipation characteristics independently from the input. Through the paper the first formulation of this technique, applicable to non linear systems, based on the use of a restricted class of the potential models, is presented. Further a second formulation of the technique in object, applicable to each kind of linear systems and based on the use of a class of linear models, characterized by a mass proportional damping matrix, is described.

Development of a Bridge Weigh-in-Motion Sensor: Performance comparison using fibre optic and electric resistance strain sensor systems

This paper addresses the problems of effective in situ measurement of the real-time strain for bridge weigh in motion in reinforced concrete bridge structures through the use of optical fiber sensor systems. By undertaking a series of tests, coupled with dynamic loading, the performance of fiber Bragg grating-based sensor systems with various amplification techniques were investigated. In recent years, structural health monitoring (SHM) systems have been developed to monitor bridge deterioration, to assess load levels and hence extend bridge life and safety. Conventional SHM systems, based on measuring strain, can be used to improve knowledge of the bridge's capacity to resist loads but generally give no information on the causes of any increase in stresses. Therefore, it is necessary to find accurate sensors capable of capturing peak strains under dynamic load and suitable methods for attaching these strain sensors to existing and new bridge structures. Additionally, it is important to ensure accurate strain transfer between concrete and steel, adhesives layer, and strain sensor. The results show the benefits in the use of optical fiber networks under these circumstances and their ability to deliver data when conventional sensors cannot capture accurate strains and/or peak strains.

A self-learning teaching-learning based optimization for dynamic economic/environmental dispatch considering multiple plug-in electric vehicle loads

Economic and environmental load dispatch aims to determine the amount of electricity generated from power plants to meet load demand while minimizing fossil fuel costs and air pollution emissions subject to operational and licensing requirements. These two scheduling problems are commonly formulated with non-smooth cost functions respectively considering various effects and constraints, such as the valve point effect, power balance and ramp rate limits. The expected increase in plug-in electric vehicles is likely to see a significant impact on the power system due to high charging power consumption and significant uncertainty in charging times. In this paper, multiple electric vehicle charging profiles are comparatively integrated into a 24-hour load demand in an economic and environment dispatch model. Self-learning teaching-learning based optimization (TLBO) is employed to solve the non-convex non-linear dispatch problems. Numerical results on well-known benchmark functions, as well as test systems with different scales of generation units show the significance of the new scheduling method.

Computational scheduling methods for integrating plug-in electric vehicles with power systems: a review

Traditional internal combustion engine vehicles are a major contributor to global greenhouse gas emissions and other air pollutants, such as particulate matter and nitrogen oxides. If the tail pipe point emissions could be managed centrally without reducing the commercial and personal user functionalities, then one of the most attractive solutions for achieving a significant reduction of emissions in the transport sector would be the mass deployment of electric vehicles. Though electric vehicle sales are still hindered by battery performance, cost and a few other technological bottlenecks, focused commercialisation and support from government policies are encouraging large scale electric vehicle adoptions. The mass proliferation of plug-in electric vehicles is likely to bring a significant additional electric load onto the grid creating a highly complex operational problem for power system operators. Electric vehicle batteries also have the ability to act as energy storage points on the distribution system. This double charge and storage impact of many uncontrollable small kW loads, as consumers will want maximum flexibility, on a distribution system which was originally not designed for such operations has the potential to be detrimental to grid balancing. Intelligent scheduling methods if established correctly could smoothly integrate electric vehicles onto the grid. Intelligent scheduling methods will help to avoid cycling of large combustion plants, using expensive fossil fuel peaking plant, match renewable generation to electric vehicle charging and not overload the distribution system causing a reduction in power quality. In this paper, a state-of-the-art review of scheduling methods to integrate plug-in electric vehicles are reviewed, examined and categorised based on their computational techniques. Thus, in addition to various existing approaches covering analytical scheduling, conventional optimisation methods (e.g. linear, non-linear mixed integer programming and dynamic programming), and game theory, meta-heuristic algorithms including genetic algorithm and particle swarm optimisation, are all comprehensively surveyed, offering a systematic reference for grid scheduling considering intelligent electric vehicle integration.

Differential evolution combined with clonal selection for dynamic economic dispatch

Dynamic economic load dispatch (DELD) is one of the most important steps in power system operation. Various optimisation algorithms for solving the problem have been developed; however, due to the non-convex characteristics and large dimensionality of the problem, it is necessary to explore new methods to further improve the dispatch results and minimise the costs. This article proposes a hybrid differential evolution (DE) algorithm, namely clonal selection-based differential evolution (CSDE), to solve the problem. CSDE is an artificial intelligence technique that can be applied to complex optimisation problems which are for example nonlinear, large scale, non-convex and discontinuous. This hybrid algorithm combines the clonal selection algorithm (CSA) as the local search technique to update the best individual in the population, which enhances the diversity of the solutions and prevents premature convergence in DE. Furthermore, we investigate four mutation operations which are used in CSA as the hyper-mutation operations. Finally, an efficient solution repair method is designed for DELD to satisfy the complicated equality and inequality constraints of the power system to guarantee the feasibility of the solutions. Two benchmark power systems are used to evaluate the performance of the proposed method. The experimental results show that the proposed CSDE/best/1 approach significantly outperforms nine other variants of CSDE and DE, as well as most other published methods, in terms of the quality of the solution and the convergence characteristics.

Approximate Computing with Unreliable Dynamic Memories

Embedded memories account for a large fraction of the overall silicon area and power consumption in modern SoC(s). While embedded memories are typically realized with SRAM, alternative solutions, such as embedded dynamic memories (eDRAM), can provide higher density and/or reduced power consumption. One major challenge that impedes the widespread adoption of eDRAM is that they require frequent refreshes potentially reducing the availability of the memory in periods of high activity and also consuming significant amount of power due to such frequent refreshes. Reducing the refresh rate while on one hand can reduce the power overhead, if not performed in a timely manner, can cause some cells to lose their content potentially resulting in memory errors. In this paper, we consider extending the refresh period of gain-cell based dynamic memories beyond the worst-case point of failure, assuming that the resulting errors can be tolerated when the use-cases are in the domain of inherently error-resilient applications. For example, we observe that for various data mining applications, a large number of memory failures can be accepted with tolerable imprecision in output quality. In particular, our results indicate that by allowing as many as 177 errors in a 16 kB memory, the maximum loss in output quality is 11%. We use this failure limit to study the impact of relaxing reliability constraints on memory availability and retention power for different technologies.

Multiloop robust H∞ control design based on the dynamic PLS approach to chemical processes

In this paper, a multiloop robust control strategy is proposed based on H∞ control and a partial least squares (PLS) model (H∞_PLS) for multivariable chemical processes. It is developed especially for multivariable systems in ill-conditioned plants and non-square systems. The advantage of PLS is to extract the strongest relationship between the input and the output variables in the reduced space of the latent variable model rather than in the original space of the highly dimensional variables. Without conventional decouplers, the dynamic PLS framework automatically decomposes the MIMO process into multiple single-loop systems in the PLS subspace so that the controller design can be simplified. Since plant/model mismatch is almost inevitable in practical applications, to enhance the robustness of this control system, the controllers based on the H∞ mixed sensitivity problem are designed in the PLS latent subspace. The feasibility and the effectiveness of the proposed approach are illustrated by the simulation results of a distillation column and a mixing tank process. Comparisons between H∞_PLS control and conventional individual control (either H∞ control or PLS control only) are also made

Loss-of-Main Monitoring and Detection for Distributed Generations Using Dynamic Principal Component Analysis

In this paper, our previous work on Principal Component Analysis (PCA) based fault detection method is extended to the dynamic monitoring and detection of loss-of-main in power systems using wide-area synchrophasor measurements. In the previous work, a static PCA model was built and verified to be capable of detecting and extracting system faulty events; however the false alarm rate is high. To address this problem, this paper uses a well-known ‘time lag shift’ method to include dynamic behavior of the PCA model based on the synchronized measurements from Phasor Measurement Units (PMU), which is named as the Dynamic Principal Component Analysis (DPCA). Compared with the static PCA approach as well as the traditional passive mechanisms of loss-of-main detection, the proposed DPCA procedure describes how the synchrophasors are linearly
auto- and cross-correlated, based on conducting the singular value decomposition on the augmented time lagged synchrophasor matrix. Similar to the static PCA method, two statistics, namely T2 and Q with confidence limits are calculated to form intuitive charts for engineers or operators to monitor the loss-of-main situation in real time. The effectiveness of the proposed methodology is evaluated on the loss-of-main monitoring of a real system, where the historic data are recorded from PMUs installed in several locations in the UK/Ireland power system.

An efficient harmony search with new pitch adjustment for dynamic economic dispatch

A simple yet efficient harmony search (HS) method with a new pitch adjustment rule (NPAHS) is proposed for dynamic economic dispatch (DED) of electrical power systems, a large-scale non-linear real time optimization problem imposed by a number of complex constraints. The new pitch adjustment rule is based on the perturbation information and the mean value of the harmony memory, which is simple to implement and helps to enhance solution quality and convergence speed. A new constraint handling technique is also developed to effectively handle various constraints in the DED problem, and the violation of ramp rate limits between the first and last scheduling intervals that is often ignored by existing approaches for DED problems is effectively eliminated. To validate the effectiveness, the NPAHS is first tested on 10 popular benchmark functions with 100 dimensions, in comparison with four HS variants and five state-of-the-art evolutionary algorithms. Then, NPAHS is used to solve three 24-h DED systems with 5, 15 and 54 units, which consider the valve point effects, transmission loss, emission and prohibited operating zones. Simulation results on all these systems show the scalability and superiority of the proposed NPAHS on various large scale problems.

Dynamic Modeling and Interaction of Hybrid Natural Gas and Electricity Supply System in Microgrid

Natural gas (NG) network and electric network are becoming tightly integrated by microturbines in the microgrid. Interactions between these two networks are not well captured by the traditional microturbine (MT) models. To address this issue, two improved models for single-shaft MT and split-shaft MT are proposed in this paper. In addition, dynamic models of the hybrid natural gas and electricity system (HGES) are developed for the analysis of their interactions. Dynamic behaviors of natural gas in pipes are described by partial differential equations (PDEs), while the electric network is described by differential algebraic equations (DAEs). So the overall network is a typical two-time scale dynamic system. Numerical studies indicate that the two-time scale algorithm is faster and can capture the interactions between the two networks. The results also show the HGES with a single-shaft MT is a weakly coupled system in which disturbances in the two networks mainly influence the dc link voltage of the MT, while the split-shaft MT is a strongly coupled system where the impact of an event will affect both networks.