Exploring cache performance in multithreaded processors

Multithreading is a well known technique to hide latency in a non-blocking cache architecture. By switching execution from one thread to another, the CPU can perform useful work, while waiting for pending requests to be processed by the main memory. In this paper we examine the effects of varying the associativity and block size on cache performance in a reduced locality of reference environment, due to multithreading. We find that for associativity equal to the number of threads, the cache produces very low miss rate even for small sizes. Also by taking into account the increase in cycle time due to larger cache size or associativity we find that the optimum cache configuration for best processor performance is 16Kbytes direct mapped. Finally, with a constant main memory bandwidth, increasing the block size to more than 32 bytes, reduces the miss rate, but degrades processor performance.     

Accelerated collection of sensor data by mobility-enabled topology ranks

We study the problem of fast and energy-efficient data collection of sensory data using a mobile sink, in wireless sensor networks in which both the sensors and the sink move. Motivated by relevant applications, we focus on dynamic sensory mobility and heterogeneous sensor placement. Our approach basically suggests to exploit the sensor motion to adaptively propagate information based on local conditions (such as high placement concentrations), so that the sink gradually “learns” the network and accordingly optimizes its motion. Compared to relevant solutions in the state of the art (such as the blind random walk, biased walks, and even optimized deterministic sink mobility), our method significantly reduces latency (the improvement ranges from 40% for uniform placements, to 800% for heterogeneous ones), while also improving the success rate and keeping the energy dissipation at very satisfactory levels.     

COVERAGE: detecting and reacting to worm epidemics using cooperation and validation

Cooperative defensive systems communicate and cooperate in their response to worm attacks, but determine the presence of a worm attack solely on local information. Distributed worm detection and immunization systems track suspicious behavior at multiple cooperating nodes to determine whether a worm attack is in progress. Earlier work has shown that cooperative systems can respond quickly to day-zero worms, while distributed detection systems allow detectors to be more conservative (i.e., paranoid) about potential attacks because they manage false alarms efficiently. In this paper we present our investigation into the complex tradeoffs in such systems between communication costs, computation overhead, accuracy of the local tests, estimation of viral virulence, and the fraction of the network infected before the attack crests. We evaluate the effectiveness of different system configurations in various simulations. Our experiments show that distributed algorithms are better able to balance effectiveness against worms and viruses with reduced cost in computation and communication when faced with false alarms. Furthermore, cooperative, distributed systems seem more robust against malicious participants in the immunization system than earlier cooperative but non-distributed approaches.     

Design and Analysis of an Efficient Communication Strategy for Hierarchical and Highly Changing Ad-hoc Mobile Networks

In this work we introduce two practical and interesting models of ad-hoc mobile networks: (a) hierarchical ad-hoc networks, comprised of dense subnetworks of mobile users interconnected by a very fast yet limited backbone infrastructure, (b) highly changing ad-hoc networks, where the deployment area changes in a highly dynamic way and is unknown to the protocol. In such networks, we study the problem of basic communication, i.e., sending messages from a sender node to a receiver node. For highly changing networks, we investigate an efficient communication protocol exploiting the coordinated motion of a small part of an ad-hoc mobile network (the runners support) to achieve fast communication. This protocol instead of using a fixed sized support for the whole duration of the protocol, employs a support of some initial (small) size which adapts (given some time which can be made fast enough) to the actual levels of traffic and the (unknown and possibly rapidly changing) network area, by changing its size in order to converge to an optimal size, thus satisfying certain Quality of Service criteria. Using random walks theory, we show that such an adaptive approach is, for this class of ad-hoc mobile networks, significantly more efficient than a simple non-adaptive implementation of the basic runners support idea, introduced in [9,10]. For hierarchical ad-hoc networks, we establish communication by using a runners support in each lower level of the hierarchy (i.e., in each dense subnetwork), while the fast backbone provides interconnections at the upper level (i.e., between the various subnetworks). We analyze the time efficiency of this hierarchical approach. This analysis indicates that the hierarchical implementation of the support approach significantly outperforms a simple implementation of it in hierarchical ad-hoc networks. Finally, we discuss a possible combination of the two approaches above (the hierarchical and the adaptive ones) that can be useful in ad-hoc networks that are both hierarchical and highly changing. Indeed, in such cases the hierarchical nature of these networks further supports the possibility of adaptation.     

Probing the dynamics of nanoparticle growth in a flame using synchrotron radiation

Flame synthesis is one of the most versatile and promising technologies for large-scale production of nanoscale materials. Pyrolysis has recently been shown to be a useful route for the production of single-walled nanotubes, quantum dots and a wide variety of nanostructured ceramic oxides for catalysis and electrochemical applications. An understanding of the mechanisms of nanostructural growth in flames has been hampered by a lack of direct observations of particle growth, owing to high temperatures (2,000 K), rapid kinetics (submillisecond scale), dilute growth conditions (10(-6) volume fraction) and optical emission of synthetic flames. Here we report the first successful in situ study of nanoparticle growth in a flame using synchrotron X-ray scattering. The results indicate that simple growth models, first derived for colloidal synthesis, can be used to facilitate our understanding of flame synthesis. Further, the results indicate the feasibility of studies of nanometre-scale aerosols of toxicological and environmental concern.     

Nanostructured titania powders by hydrothermal processing and spray drying

Sol-gel precipitates of titania were prepared from organometallic precursors, followed by hydrothermal crystallization and spray drying into spherical, micron-size nanostructured granules that were characterized by X-ray diffraction, nitrogen adsorption, and microscopy. A systematic investigation of the effect of hydrothermal processing conditions on the physical properties of the granules was carried out. When the hydrothermal processing time was increased, product granules exhibited larger anatase crystallite size and improved thermal stability in pore structure and phase composition. Increasing the hydrothermal processing temperature retarded crystallite growth and increased the anatase-to-rutile transformation temperature.     

Direction-based adaptive data propagation for heterogeneous sensor mobility

We consider sensor networks where the sensor nodes are attached on entities that move in a highly dynamic, heterogeneous manner. To capture this mobility diversity we introduce a new network parameter, the direction-aware mobility level, which measures how fast and close each mobile node is expected to get to the data destination (the sink). We then provide local, distributed data dissemination protocols that adaptively exploit the node mobility to improve performance. In particular, “high” mobility is used as a low cost replacement for data dissemination (due to the ferrying of data), while in the case of “low” mobility either (a) data propagation redundancy is increased (when highly mobile neighbors exist) or (b) long-distance data transmissions are used (when the entire neighborhood is of low mobility) to accelerate data dissemination toward the sink. An extensive performance comparison to relevant methods from the state of the art demonstrates significant improvements, i.e. latency is reduced by even four times while keeping energy dissipation and delivery success at very satisfactory levels.     

The SAINT observatory subsystem: an open-source intelligence tool for uncovering cybersecurity threats

International Journal of Information Security - Data from Online Social Networks, search engines, and the World Wide Web are forms of unstructured knowledge that are not regularly used in...     

Robust face recognition using 2D and 3D data: Pose and illumination compensation

The paper addresses the problem of face recognition under varying pose and illumination. Robustness to appearance variations is achieved not only by using a combination of a 2D color and a 3D image of the face, but mainly by using face geometry information to cope with pose and illumination variations that inhibit the performance of 2D face recognition. A face normalization approach is proposed, which unlike state-of-the-art techniques is computationally efficient and does not require an extended training set. Experimental results on a large data set show that template-based face recognition performance is significantly benefited from the application of the proposed normalization algorithms prior to classification.     

The Impact of Network Structure on the Stability of Greedy Protocols

A packet-switching network is stable if the number of packets in the network remains bounded at all times. A very natural question that arises in the context of stability properties of such networks is how network structure precisely affects these properties. In this work we embark on a systematic study of this question in the context of Adversarial Queueing Theory, which assumes that packets are adversarially injected into the network. We consider size, diameter, maximum vertex degree, minimum number of disjoint paths that cover all edges of the network and network subgraphs as crucial structural parameters of the network, and we present a comprehensive collection of structural results, in the form of stability and instability bounds on injection rate of the adversary for various greedy protocols: —Increasing the size of a network may result in dropping its instability bound. This is shown through a novel, yet simple and natural, combinatorial construction of a size-parameterized network on which certain compositions of greedy protocols are running. The convergence of the drop to 0.5 is found to be fast with and proportional to the increase in size. —Maintaining the size of a network small may already suffice to drop its instability bound to a substantially low value. This is shown through a construction of a FIFO network with size 22, which becomes unstable at rate 0.704. This represents the current state-of-the-art trade-off between network size and instability bound. —The diameter, maximum vertex degree and minimum number of edge-disjoint paths that cover a network may be used as control parameters for the stability bound of the network. This is shown through an improved analysis of the stability bound of any arbitrary FIFO network, which takes these parameters into account. —How much can network subgraphs that are forbidden for stability affect the instability bound? Through improved combinatorial constructions of networks and executions, we improve the state-of-the-art instability bound induced by certain known forbidden subgraphs on networks running a certain greedy protocol. —Our results shed more light and contribute significantly to a finer understanding of the impact of structural parameters on stability and instability properties of networks.