Limited Memory Resources Research Articles

A fast algorithm for sparse support vector machines for mobile computing applications

This research presents a fast algorithm for projected support vector machines (PSVM) by selecting a basis vector set (BVS) for the kernel-induced feature space, the training points are projected onto the subspace spanned by the selected BVS. A standard linear support vector machine (SVM) is then produced in the subspace with the projected training points. As the dimension of the subspace is determined by the size of the selected basis vector set, the size of the produced SVM expansion can be specified. A two-stage algorithm is derived which selects and refines the basis vector set achieving a locally optimal model. The model expansion coefficients and bias are updated recursively for increase and decrease in the basis set and support vector set. The condition for a point to be classed as outside the current basis vector and selected as a new basis vector is derived and embedded in the recursive procedure. This guarantees the linear independence of the produced basis set. The proposed algorithm is tested and compared with an existing sparse primal SVM (SpSVM) and a standard SVM (LibSVM) on seven public benchmark classification problems. Our new algorithm is designed for use in the application area of human activity recognition using smart devices and embedded sensors where their sometimes limited memory and processing resources must be exploited to the full and the more robust and accurate the classification the more satisfied the user. Experimental results demonstrate the effectiveness and efficiency of the proposed algorithm. This work builds upon a previously published algorithm specifically created for activity recognition within mobile applications for the EU Haptimap project [1]. The algorithms detailed in this paper are more memory and resource efficient making them suitable for use with bigger data sets and more easily trained SVMs.

Internal inductance of a conductor of rectangular cross-section using the proper generalized decomposition

Purpose Rectangular conductors play an important role in planar transmission line structures, multiconductor transmission lines, in power transmission and distribution systems, LCL filters, transformers, industrial busbars, MEMs devices, among many others. The precise determination of the inductance of such conductors is necessary for their design and optimization, but no explicit solution for the AC resistance and internal inductances per-unit length of a linear conductor with a rectangular cross-section has been found, so numerical methods must be used. The purpose of this paper is to introduce the use of a novel numerical technique, the proper generalized decomposition (PGD), for the calculation of DC and AC internal inductances of rectangular conductors. Design/methodology/approach The PGD approach is used to obtain numerically the internal inductance of a conductor with circular cross-section and with rectangular cross-section, both under DC and AC conditions, using a separated representation of the magnetic vector potential in a 2D domain. The results are compared with the analytical and approximate expressions available in the technical literature, with an excellent concordance. Findings The PGD uses simple one-dimensional meshes, one per dimension, so the use of computational resources is very low, and the simulation speed is very high. Besides, the application of the PGD to conductors with rectangular cross-section is particularly advantageous, because rectangular shapes can be represented with a very few number of independent terms, which makes the code very simple and compact. Finally, a key advantage of the PGD is that some parameters of the numerical model can be considered as additional dimensions. In this paper, the frequency has been considered as an additional dimension, and the internal inductance of a rectangular conductor has been computed for the whole range of frequencies desired using a single numerical simulation. Research limitations/implications The proposed approach may be applied to the optimization of electrical conductors used in power systems, to solve EMC problems, to the evaluation of partial inductances of wires, etc. Nevertheless, it cannot be applied, as presented in this work, to 3D complex shapes, as, for example, an arrangement of layers of helically stranded wires. Originality/value The PGD is a promising new numerical procedure that has been applied successfully in different fields. In this paper, this novel technique is applied to find the DC and AC internal inductance of a conductor with rectangular cross-section, using very dense and large one-dimensional meshes. The proposed method requires very limited memory resources, is very fast, can be programmed using a very simple code, and gives the value of the AC inductance for a complete range of frequencies in a single simulation. The proposed approach can be extended to arbitrary conductor shapes and complex multiconductor lines to further exploit the advantages of the PGD.

Queueing Analysis of Continuous Queries for Uncertain Data Streams Over Sliding Windows

With the rapid development of data collection methods and their practical applications, the management of uncertain data streams has drawn wide attention in both academia and industry. System capacity planning and Quality of service (QoS) metrics are two very important problems for data stream management systems (DSMSs) to process streams efficiently due to unpredictable input characteristics and limited memory resource in the system. Motivated by this, in this paper, we explore an effective approach to estimate the memory requirement, data loss ratio, and tuple latency of continuous queries for uncertain data streams over sliding windows in a DSMS. More specifically, we propose a queueing model to address these problems in this paper. We study the average number of tuples, average tuple latency in the queue, and the distribution of the number of tuples and tuple latency in the queue under the Poisson arrival of input data streams in our queueing model. Furthermore, we also determine the maximum capacity of the queueing system based on the data loss ratio. The solutions for the above problems are very important to help researchers design, manage, and optimize a DSMS, including allocating buffer needed for a queue and admitting a continuous uncertain query to the system without violation of the pre-specified QoS requirements.

A new hash algorithm exploiting triple-state bucket directory for flash storage devices

Recently, the flash storage device becomes more important to the mobile consumer electronics that supports various functionalities because of its portability and reliability. The capacity of the flash storage device is becoming larger and its price is going down due to the mass production of the Triple-Level Cell (TLC) flash memory. Since TLC flash memories are very cheap, they are widely deployed in the mobile flash storage device although its durability is poor. In addition, it is difficult to handle the large-scale flash memories with limited memory resources in the mobile flash storage device. Therefore, TLC-based flash storage device requires to internally optimize the write and erase operations with small memory resources. This paper proposes a new hash-based mapping algorithm, which reduces the redundant write and erase operations and uses only a few memory resources, for the mobile flash storage device. The proposed hash algorithm also guarantees uniform performance regardless of various workload patterns. Through various experiments with related works, this paper shows the superiority of the proposed hash algorithm.

A Data-Driven Point Cloud Simplification Framework for City-Scale Image-Based Localization.

City-scale 3D point clouds reconstructed via structure-from-motion from a large collection of Internet images are widely used in the image-based localization task to estimate a 6-DOF camera pose of a query image. Due to prohibitive memory footprint of city-scale point clouds, image-based localization is difficult to be implemented on devices with limited memory resources. Point cloud simplification aims to select a subset of points to achieve a comparable localization performance using the original point cloud. In this paper, we propose a data-driven point cloud simplification framework by taking it as a weighted K-Cover problem, which mainly includes two complementary parts. First, a utility-based parameter determination method is proposed to select a reasonable parameter K for K-Cover-based approaches by evaluating the potential of a point cloud for establishing sufficient 2D-3D feature correspondences. Second, we formulate the 3D point cloud simplification problem as a weighted K-Cover problem, and propose an adaptive exponential weight function based on the visibility probability of 3D points. The experimental results on three popular datasets demonstrate that the proposed point cloud simplification framework outperforms the state-of-the-art methods for the image-based localization application with a well predicted parameter in the K-Cover problem.

A Hybrid Secure Scheme for Wireless Sensor Networks against Timing Attacks Using Continuous-Time Markov Chain and Queueing Model.

Wireless sensor networks (WSNs) have recently gained popularity for a wide spectrum of applications. Monitoring tasks can be performed in various environments. This may be beneficial in many scenarios, but it certainly exhibits new challenges in terms of security due to increased data transmission over the wireless channel with potentially unknown threats. Among possible security issues are timing attacks, which are not prevented by traditional cryptographic security. Moreover, the limited energy and memory resources prohibit the use of complex security mechanisms in such systems. Therefore, balancing between security and the associated energy consumption becomes a crucial challenge. This paper proposes a secure scheme for WSNs while maintaining the requirement of the security-performance tradeoff. In order to proceed to a quantitative treatment of this problem, a hybrid continuous-time Markov chain (CTMC) and queueing model are put forward, and the tradeoff analysis of the security and performance attributes is carried out. By extending and transforming this model, the mean time to security attributes failure is evaluated. Through tradeoff analysis, we show that our scheme can enhance the security of WSNs, and the optimal rekeying rate of the performance and security tradeoff can be obtained.

HEADA: a low cost RFID authentication technique using homomorphic encryption for key generation

AbstractThe limited computational and memory resources available in the Radio Frequency Identification (RFID) tags constitute the essential challenge to find a technique that satisfies high security requirements. In this paper, security and privacy requirements for an RFID authentication system are defined. Although some of the conventional cryptographic operations provide these requirements partially, they are not considered as suitable solution for RFID applications because operations cost is high especially for RFID tags. An unconventional use of homomorphic encryption is proposed to provide low‐cost security and privacy in this research. HEADA is proposed as a novel authentication technique, which consists of deployment and authentication process stages. The homomorphic encryption is used solely in the generation of keys during deployment stage. These keys are used by RFID tags to generate anonymous authentication keys during the authentication process using only integer addition operations. Moreover, some of the conventional approaches have to use a brute–force search in the server side to identify a tag. Unlike these techniques, HEADA enables the server to identify a tag only by a binary search. It is shown that HEADA is the only technique to satisfy all security and privacy requirements using low‐cost operations in both tag and server sides. Copyright © 2016 John Wiley & Sons, Ltd.

Active binaural localization of multiple sound sources

Sound source localization serves as a significant capability of autonomous robots that conduct missions such as search and rescue, and target tracking in challenging environments. However, localization of multiple sound sources and static sound source tracking in self-motion are both difficult tasks, especially when the number of sound sources or reflections increase. This study presents two robotic hearing approaches based on a human perception model (Wallach, 1939) that combines interaural time difference (ITD) and head turn motion data to locate sound sources. The first method uses a fitting-based approach to recognize the changing trends of the cross-correlation function of binaural inputs. The effectiveness of this method was validated using data collected from a two-microphone array rotating in a non-anechoic environment, and the experiments reveal its ability to separate and localize up to three sound sources of the same spectral content (white noise) at different azimuth and elevation angles. The second method uses an extended Kalman filter (EKF) that estimates the orientation of a sound source by fusing the robot’s self-motion and ITD data to reduce the localization errors recursively. This method requires limited memory resources and is able to keep tracking the relative position change of a number of static sources when the robot moves. In the experiments, up to three sources can be tracked simultaneously with a two-microphone array.

Resource aware and reliable data reporting algorithm for object tracking in WSNs

Object tracking have become one of the major applications of Wireless Sensor Networks (WSNs) due to its wide real-life applications such as wildlife animal monitoring and military area intrusion detection. Many recent articles have been dedicated to localization of objects; however, few of these ar ticles were concentrated on the reliability of network data reporting along with objects localization. In this work, an efficient data reporting method is proposed for object tracking in WSNs. Energy is considered as one of the most critical resources for WSN. Data transmission from the nodes to the sink along with the minimum energy path could be one of the solutions to minimize the overall network energy consumption. However, this might lead to unbalanced energy among sensor nodes resulting in, energy hole problem. Moreover, the reliable data transmission is an essential aspect that should be considered when designing a WSN for object tracking application, where the loss of data packets will affect the accuracy of the tracking and location estimation of a mobile object. Furthermore, due to the limited memory resources of sensor nodes, full utilization of such resources with less buffer overflow remains as a one of main consideration when a WSN application is designed. Consequently, this paper aims to achieve both minimum energy consumption in reporting operation and balanced energy consumption among sensor nodes for WSN lifetime extension. In addition, data reliability is considered in our model where, the sensed data can reach the sink node in a more reliable way. Finally, buffer space is considered in to reduce the packet loss and energy consumption due to the retransmission of the same packets. This work first formulates the problem as 0/1 Integer Linear Programming (ILP) problem, and proposes SWARM intelligence to solve the optimization problem. Through simulation, the performance of proposed method to report information about the detected objects to the sink is compared with the previous work such as LR-based object tracking algorithm, EBRP, ACO, TADR, SEB, and CLR-Routing.

Large-Scale Simulations of Plastic Neural Networks on Neuromorphic Hardware.

SpiNNaker is a digital, neuromorphic architecture designed for simulating large-scale spiking neural networks at speeds close to biological real-time. Rather than using bespoke analog or digital hardware, the basic computational unit of a SpiNNaker system is a general-purpose ARM processor, allowing it to be programmed to simulate a wide variety of neuron and synapse models. This flexibility is particularly valuable in the study of biological plasticity phenomena. A recently proposed learning rule based on the Bayesian Confidence Propagation Neural Network (BCPNN) paradigm offers a generic framework for modeling the interaction of different plasticity mechanisms using spiking neurons. However, it can be computationally expensive to simulate large networks with BCPNN learning since it requires multiple state variables for each synapse, each of which needs to be updated every simulation time-step. We discuss the trade-offs in efficiency and accuracy involved in developing an event-based BCPNN implementation for SpiNNaker based on an analytical solution to the BCPNN equations, and detail the steps taken to fit this within the limited computational and memory resources of the SpiNNaker architecture. We demonstrate this learning rule by learning temporal sequences of neural activity within a recurrent attractor network which we simulate at scales of up to 2.0 × 104 neurons and 5.1 × 107 plastic synapses: the largest plastic neural network ever to be simulated on neuromorphic hardware. We also run a comparable simulation on a Cray XC-30 supercomputer system and find that, if it is to match the run-time of our SpiNNaker simulation, the super computer system uses approximately 45× more power. This suggests that cheaper, more power efficient neuromorphic systems are becoming useful discovery tools in the study of plasticity in large-scale brain models.

Sensory processing patterns predict the integration of information held in visual working memory.

Given the limited resources of visual working memory, multiple items may be remembered as an averaged group or ensemble. As a result, local information may be ill-defined, but these ensemble representations provide accurate diagnostics of the natural world by combining gist information with item-level information held in visual working memory. Some neurodevelopmental disorders are characterized by sensory processing profiles that predispose individuals to avoid or seek-out sensory stimulation, fundamentally altering their perceptual experience. Here, we report such processing styles will affect the computation of ensemble statistics in the general population. We identified stable adult sensory processing patterns to demonstrate that individuals with low sensory thresholds who show a greater proclivity to engage in active response strategies to prevent sensory overstimulation are less likely to integrate mean size information across a set of similar items and are therefore more likely to be biased away from the mean size representation of an ensemble display. We therefore propose the study of ensemble processing should extend beyond the statistics of the display, and should also consider the statistics of the observer. (PsycINFO Database Record

A Fuzzy Logic-Based Clustering Algorithm for WSN to Extend the Network Lifetime

Wireless sensor network (WSN) brings a new paradigm of real-time embedded systems with limited computation, communication, memory, and energy resources that are being used for huge range of applications where the traditional infrastructure-based network is mostly infeasible. The sensor nodes are densely deployed in a hostile environment to monitor, detect, and analyze the physical phenomenon and consume considerable amount of energy while transmitting the information. It is impractical and sometimes impossible to replace the battery and to maintain longer network life time. So, there is a limitation on the lifetime of the battery power and energy conservation is a challenging issue. Appropriate cluster head (CH) election is one such issue, which can reduce the energy consumption dramatically. Low energy adaptive clustering hierarchy (LEACH) is the most famous hierarchical routing protocol, where the CH is elected in rotation basis based on a probabilistic threshold value and only CHs are allowed to send the information to the base station (BS). But in this approach, a super-CH (SCH) is elected among the CHs who can only send the information to the mobile BS by choosing suitable fuzzy descriptors, such as remaining battery power, mobility of BS, and centrality of the clusters. Fuzzy inference engine (Mamdani's rule) is used to elect the chance to be the SCH. The results have been derived from NS-2 simulator and show that the proposed protocol performs better than the LEACH protocol in terms of the first node dies, half node alive, better stability, and better lifetime.

Lightweight identity-based authentication protocol for wireless sensor networks

Wireless sensor network WSN is often deployed in hostile environments, which make it vulnerable to attacks. WSN comprises a large number of sensor nodes with different hardware abilities and functions. Due to the limited memory resources and energy constraints, complex security algorithms cannot be used in WSN. Hence, it is necessary to balance between security requirements and energy consumption. In this paper, we propose a lightweight public-key-based authentication protocol for WSNs. Our protocol uses identity-based encryption in order to lighten the energy consumption from the public-key certificate management. We have developed two variations of the proposed solution depending on the manner and when an attack should be detected. We perform an overall evaluation of our approach through simulations. The results indicate out performance of our approach in terms of energy consumption while providing effective security.

Swarm intelligence based reliable and energy balance routing algorithm for wireless sensor network

Energy is an extremely crucial resource for Wireless Sensor Networks (WSNs). Many routing techniques have been proposed for finding the minimum energy routing paths with a view to extend the network lifetime. However, this might lead to unbalanced distribution of energy among sensor nodes resulting in, energy hole problem. Therefore, designing energy-balanced routing technique is a challenge area of research in WSN. Moreover, dynamic and harsh environments pose great challenges in the reliability of WSN. To achieve reliable wireless communication within WSN, it is essential to have reliable routing protocol. Furthermore, due to the limited memory resources of sensor nodes, full utilization of such resources with less buffer overflow remains as a one of main consideration when designing a routing protocol for WSN. Consequently, this paper proposes a routing scheme that uses SWARM intelligence to achieve both minimum energy consumption and balanced energy consumption among sensor nodes for WSN lifetime extension. In addition, data reliability is considered in our model where, the sensed data can reach the sink node in a more reliable way. Finally, buffer space is considered to reduce the packet loss and energy consumption due to the retransmission of the same packets. Through simulation, the performance of proposed algorithm is compared with the previous work such as EBRP, ACO, TADR, SEB, and CLR-Routing.

Block-Matching Correlation Motion Estimation for Frame-Rate up-Conversion

This paper proposes a novel motion estimation algorithm that combines recursive block-matching and customized phase plane correlation. In comparison with alternative approaches the proposed solution shows better subjective impressions and objective measurements, such as Peak Signal to Noise Ratio and Universal Image Quality Index. Computational complexity of the algorithm is constrained and suitable for real-time implementations on platforms with limited memory and processing resources. The proof of concept is the first real-time implementation of frame-rate up-conversion on a mobile platform (Intel Merrifield), which is capable of processing 1080p video at 30 frames per second while clocked at 200 MHz.

Development of the reconfiguration acceleration method in the dynamically reconfigurable computing systems

The problem of performance improvement of reconfigurable computing systems, including solving the problem of reconfiguration overhead reduction was considered. A new reconfiguration acceleration method and hardware for its implementation, which allow to minimize the reconfiguration time overhead were proposed. Analytical expressions that formalize the reconfiguration acceleration method, justify the virtually complete removal of unproductive reconfiguration time by reducing the communication component of time that provides an intensive reconfiguration acceleration were obtained. The formalization shows that the volume of the removed unproductive time component is linearly dependent on the number of repetitive tasks. Thus, using the proposed method is the most efficient in the algorithms that contain a large number of similar tasks. The proposed hardware for schedule management of placement and support of configurations of functional units, based on the multi-level memory, provide effective support of the reconfiguration overhead reduction method and allow to reduce computational complexity of reconfiguration control algorithms and solve the problem of limited resources of internal memory of the FPGA. The designed emulator of the reconfigurable computing system and software model of the reconfiguration acceleration method enable the real-time control simulation of reconfigurable resources. The developed software is a handy tool to study the temporal characteristics of the reconfigurable computing system. Simulation of the reconfiguration acceleration method for computing algorithms with the multilevel structure, which is experimentally confirmed by theoretical research was performed.

An Efficient Multi Parent Hierarchical Routing Protocol for WSN’s

Wireless sensor networks (WSNs) nodes are commonly designed to work with limited resources of memory, energy and processing. The routing protocol is one of the key components of WSNs and its features impact network performance significantly. We present an efficient Multi-Parent Hierarchical (MPH) routing protocol for wireless sensor networks; its main goal is to achieve reliable delivery of data in a single sink scenario while keeping low overhead, reduced latency and low energy consumption. The main features of MPH are self-configuration, hierarchical topology, persistence according to link quality, and source routing from sink to nodes. Network performance simulations of the MPH routing protocol are carried out and compared with popular protocol, AODV. Results show that for the single sink scenario, the MPH protocol has an energy saving of 35% against AODV. MPH has 27% less overhead compared with AODV. And MPH presents a 10% increase in packet delivery compared with AODV. Finally, we present a real WSN built based on the MPH protocol, which works satisfactorily, providing an experimental demonstration of the capabilities of the protocol.Â

Evidence for Optimal Integration of Visual Feature Representations across Saccades.

This study examines the consequences of saccadic eye movements for the internal representation of visual objects. A saccade shifts the image of a stable visual object from one part of the retina to another. We show that visual representations are built up over these different views of the same object, by combining information obtained before and after each saccade. The weights given to presaccadic and postsaccadic information are determined by the relative reliability of each input. This provides evidence that the visual system combines inputs over time in a statistically optimal way.

An interference account of the missing-VP effect.

Sentences with doubly center-embedded relative clauses in which a verb phrase (VP) is missing are sometimes perceived as grammatical, thus giving rise to an illusion of grammaticality. In this paper, we provide a new account of why missing-VP sentences, which are both complex and ungrammatical, lead to an illusion of grammaticality, the so-called missing-VP effect. We propose that the missing-VP effect in particular, and processing difficulties with multiply center-embedded clauses more generally, are best understood as resulting from interference during cue-based retrieval. When processing a sentence with double center-embedding, a retrieval error due to interference can cause the verb of an embedded clause to be erroneously attached into a higher clause. This can lead to an illusion of grammaticality in the case of missing-VP sentences and to processing complexity in the case of complete sentences with double center-embedding. Evidence for an interference account of the missing-VP effect comes from experiments that have investigated the missing-VP effect in German using a speeded grammaticality judgments procedure. We review this evidence and then present two new experiments that show that the missing-VP effect can be found in German also with less restricting procedures. One experiment was a questionnaire study which required grammaticality judgments from participants without imposing any time constraints. The second experiment used a self-paced reading procedure and did not require any judgments. Both experiments confirm the prior findings of missing-VP effects in German and also show that the missing-VP effect is subject to a primacy effect as known from the memory literature. Based on this evidence, we argue that an account of missing-VP effects in terms of interference during cue-based retrieval is superior to accounts in terms of limited memory resources or in terms of experience with embedded structures.

Failing to get the gist of what's being said: background noise impairs higher-order cognitive processing

A dynamic interplay is known to exist between auditory processing and human cognition. For example, prior investigations of speech-in-noise have revealed there is more to learning than just listening: Even if all words within a spoken list are correctly heard in noise, later memory for those words is typically impoverished. These investigations supported a view that there is a “gap” between the intelligibility of speech and memory for that speech. Here, the notion was that this gap between speech intelligibility and memorability is a function of the extent to which the spoken message seizes limited immediate memory resources (e.g., Kjellberg et al., 2008). Accordingly, the more difficult the processing of the spoken message, the less resources are available for elaboration, storage, and recall of that spoken material. However, it was not previously known how increasing that difficulty affected the memory processing of semantically rich spoken material. This investigation showed that noise impairs higher levels of cognitive analysis. A variant of the Deese-Roediger-McDermott procedure that encourages semantic elaborative processes was deployed. On each trial, participants listened to a 36-item list comprising 12 words blocked by each of 3 different themes. Each of those 12 words (e.g., bed, tired, snore…) was associated with a “critical” lure theme word that was not presented (e.g., sleep). Word lists were either presented without noise or at a signal-to-noise ratio of 5 decibels upon an A-weighting. Noise reduced false recall of the critical words, and decreased the semantic clustering of recall. Theoretical and practical implications are discussed.