A Capabilities-Aware Framework for Using Computational Accelerators in Data-Intensive Computing

Multicore computational accelerators such as GPUs are now commodity components for highperformance computing at scale. While such accelerators have been studied in some detail as stand-alone computational engines, their integration in large-scale distributed systems raises new challenges and trade-offs. In this paper, we present an exploration of resource management alternatives for building asymmetric accelerator-based distributed systems. We present these alternatives in the context of a capabilities-aware framework for data-intensive computing, which uses an enhanced implementation of the MapReduce programming model for accelerator-based clusters, compared to the state of the art. The framework can transparently utilize heterogeneous accelerators for deriving high performance with low programming effort. Our work is the first to compare heterogeneous types of accelerators, GPUs and a Cell processors, in the same environment and the first to explore the trade-offs between compute-efficient and control-efficient accelerators on data-intensive systems. Our investigation shows that our framework scales well with the number of different compute nodes. Furthermore, it runs simultaneously on two different types of accelerators, successfully adapts to the resource capabilities, and performs 26.9% better on average than a static execution approach.

Evaluating the impact of network density, hidden nodes and capture effect for throughput guarantee in multi-hop wireless networks

To optimize the performance of wireless networks, one needs to consider the impact of key factors such as interference from hidden nodes, the capture effect, the network density and network conditions (saturated versus non-saturated). In this research, our goal is to quantify the impact of these factors and to propose effective mechanisms and algorithms for throughput guarantees in multi-hop wireless networks. For this purpose, we have developed a model that takes into account all these key factors, based on which an admission control algorithm and an end-to-end available bandwidth estimation algorithm are proposed. Given the necessary network information and traffic demands as inputs, these algorithms are able to provide predictive control via an iterative approach. Evaluations using analytical comparison with simulations as well as existing research show that the proposed model and algorithms are accurate and effective.

Spontaneous Attention to Faces in Asperger Sndrome using Ecologically Valid Static Stimuli

Previous eye tracking research on the allocation of attention to social information by individuals with autism spectrum disorders is equivocal and may be in part a consequence of variation in stimuli used between studies. The current study explored attention allocation to faces, and within faces,
by individuals with Asperger syndrome using a range of static stimuli where faces were either viewed in isolation or viewed in the context of a social scene. Results showed that faces were viewed typically by the individuals with Asperger syndrome when presented in isolation, but attention to the eyes was significantly diminished in comparison to age and IQ-matched typical viewers when faces were viewed as part of social scenes. We show that when using static stimuli, there is evidence of atypicality for individuals with Asperger syndrome depending on the extent of social context. Our findings shed light on the previous explanations of gaze behaviour that have emphasised the role of movement in atypicalities of social attention in autism spectrum disorders and highlight the importance of consideration of the realistic portrayal of social information for future studies.

Gender Classification via Lips: Static and Dynamic Features

Automatic gender classification has many security and commercial applications. Various modalities have been investigated for gender classification with face-based classification being the most popular. In some real-world scenarios the face may be partially occluded. In these circumstances a classification based on individual parts of the face known as local features must be adopted. We investigate gender classification using lip movements. We show for the first time that important gender specific information can be obtained from the way in which a person moves their lips during speech. Furthermore our study indicates that the lip dynamics during speech provide greater gender discriminative information than simply lip appearance. We also show that the lip dynamics and appearance contain complementary gender information such that a model which captures both traits gives the highest overall classification result. We use Discrete Cosine Transform based features and Gaussian Mixture Modelling to model lip appearance and dynamics and employ the XM2VTS database for our experiments. Our experiments show that a model which captures lip dynamics along with appearance can improve gender classification rates by between 16-21% compared to models of only lip appearance.

An Experimental and Numerical Study of the Static and Fatigue Performance of a Composite Adhesive Repair

Experimental static and fatigue tension-tension tests were carried out on 5HS/RTM6 composite intact coupons and coupons incorporating adhesively-bonded (FM300-2) stepped flush joints. The results show that the adhesive joint, which is widely used in repairs, significantly reduces the static strength as well as the fatigue life of the composite. Both, the static and the fatigue failure of the ‘repaired’ coupons occur at the adhesive joint and involve crack initiation and propagation. The latter is modelled using interface finite elements based on the decohezive zone approach. The material degradation in the interface constitutive law is described by a damage variable, which can evolve due to the applied loads as well as the number of fatigue cycles. The fatigue formulation, based on a published model, is adapted to fit the framework of the pseudotransient formulation that is used as a numerical tool to overcome convergence difficulties. The fatigue model requires three material parameters. Numerical tests show that a single set of these parameters can be used to recover, very accurately, the experimental S-N relationship. Sensitivity studies show that the results are not mesh dependent.

Resolution theory, and static and frequency-dependent cross-talk in piezoresponse force microscopy

Probing the functionality of materials locally by means of scanning probe microscopy (SPM) requires a reliable framework for identifying the target signal and separating it from the effects of surface morphology and instrument non-idealities, e.g. instrumental and topographical cross-talk. Here we develop a linear resolution theory framework in order to describe the cross-talk effects, and apply it for elucidation of frequency-dependent cross-talk mechanisms in piezoresponse force microscopy. The use of a band excitation method allows electromechanical/electrical and mechanical/topographic signals to be unambiguously separated. The applicability of a functional fit approach and multivariate statistical analysis methods for identification of data in band excitation SPM is explored.

Fatigue Life Assessment and Static Testing of Structural GFRP Tubes Based on Coupon Tests

Fiber-reinforced polymer (FRP) hollow tubes are used in structural applications, such as utility poles and pipelines. Concrete-filled FRP tubes (CFFTs) are also used as piles and bridge piers. Applications such as poles and marine piles are typically governed by cyclic bending. In this paper, the fatigue behavior of glass-FRP filament-wound tubes is studied using coupons cut from the tubes. Several coupon configurations were first examined in 24 tension and five compression monotonic loading tests. Fatigue tests were then conducted on 81 coupons to examine several parameters; namely, loading frequency as well as maximum-to-ultimate (max ult) and minimum-to-maximum (min max) stress ratios, including tension tension and tension compression, to simulate reversed bending. The study demonstrated the sensitivity of test results and failure mode to coupon configuration. The presence of compression loads reduced fatigue life, while increasing load frequency increased fatigue life. Stiffness degradation behavior was also established. To achieve at least one million cycles, it is recommended to limit (max ult) to 0.25. Models were used to simulate stiffness degradation and fatigue life curve of the tube. Fatigue life predictions of large CFFT beams showed good correlation with experimental results. © 2008 ASCE.

Invisible Targets, Strengthened Morale: Static Camouflage as a ‘Weapon of the Weak’

In the natural world, camouflage is habitually deployed by 'vulnerable' creatures to deceive predators. Such protective strategies have been culturally, socially and technologically translated into human societies, whereby camouflage has been used to mask intentions, actions, feelings and valuable objects or spaces. Through the material presence of such techniques, everyday spaces can become inscribed as places of sanctuary. Focusing on British civil camouflage work of the 1930s and 1940s, this paper explores the historical, cultural and political connotations of camouflage and how the attainment of invisibility, as a 'weapon of the weak', can both physically and affectively protect urban populations. © 2012 Copyright Taylor and Francis Group, LLC.