Showing papers in "IEEE Transactions on Emerging Topics in Computing in 2014"
TL;DR: Concepts and algorithms related to clustering, a concise survey of existing (clustering) algorithms as well as a comparison, both from a theoretical and an empirical perspective are introduced.
Abstract: Clustering algorithms have emerged as an alternative powerful meta-learning tool to accurately analyze the massive volume of data generated by modern applications. In particular, their main goal is to categorize data into clusters such that objects are grouped in the same cluster when they are similar according to specific metrics. There is a vast body of knowledge in the area of clustering and there has been attempts to analyze and categorize them for a larger number of applications. However, one of the major issues in using clustering algorithms for big data that causes confusion amongst practitioners is the lack of consensus in the definition of their properties as well as a lack of formal categorization. With the intention of alleviating these problems, this paper introduces concepts and algorithms related to clustering, a concise survey of existing (clustering) algorithms as well as providing a comparison, both from a theoretical and an empirical perspective. From a theoretical perspective, we developed a categorizing framework based on the main properties pointed out in previous studies. Empirically, we conducted extensive experiments where we compared the most representative algorithm from each of the categories using a large number of real (big) data sets. The effectiveness of the candidate clustering algorithms is measured through a number of internal and external validity metrics, stability, runtime, and scalability tests. In addition, we highlighted the set of clustering algorithms that are the best performing for big data.
833 citations
TL;DR: Novel robust and low-overhead physical unclonable function (PUF) authentication and key exchange protocols that are resilient against reverse-engineering attacks are proposed and evaluated and confirmed by hardware implementation.
Abstract: This paper proposes novel robust and low-overhead physical unclonable function (PUF) authentication and key exchange protocols that are resilient against reverse-engineering attacks. The protocols are executed between a party with access to a physical PUF (prover) and a trusted party who has access to the PUF compact model (verifier). The proposed protocols do not follow the classic paradigm of exposing the full PUF responses or a transformation of them. Instead, random subsets of the PUF response strings are sent to the verifier so the exact position of the subset is obfuscated for the third-party channel observers. Authentication of the responses at the verifier side is done by matching the substring to the available full response string; the index of the matching point is the actual obfuscated secret (or key) and not the response substring itself. We perform a thorough analysis of resiliency of the protocols against various adversarial acts, including machine learning and statistical attacks. The attack analysis guides us in tuning the parameters of the protocol for an efficient and secure implementation. The low overhead and practicality of the protocols are evaluated and confirmed by hardware implementation.
160 citations
TL;DR: Theoretical analyzes and experimental results demonstrate that the proposed unified tensor model and IHOSVD method are efficient for big data representation and dimensionality reduction.
Abstract: Variety and veracity are two distinct characteristics of large-scale and heterogeneous data It has been a great challenge to efficiently represent and process big data with a unified scheme In this paper, a unified tensor model is proposed to represent the unstructured, semistructured, and structured data With tensor extension operator, various types of data are represented as subtensors and then are merged to a unified tensor In order to extract the core tensor which is small but contains valuable information, an incremental high order singular value decomposition (IHOSVD) method is presented By recursively applying the incremental matrix decomposition algorithm, IHOSVD is able to update the orthogonal bases and compute the new core tensor Analyzes in terms of time complexity, memory usage, and approximation accuracy of the proposed method are provided in this paper A case study illustrates that approximate data reconstructed from the core set containing 18% elements can guarantee 93% accuracy in general Theoretical analyzes and experimental results demonstrate that the proposed unified tensor model and IHOSVD method are efficient for big data representation and dimensionality reduction
155 citations
TL;DR: A whole model for generating the association relation between multimedia resources using semantic link network model is proposed, which shows the proposed method can measure the semantic relatedness between Flickr images accurately and robustly.
Abstract: Recent research shows that multimedia resources in the wild are growing at a staggering rate. The rapid increase number of multimedia resources has brought an urgent need to develop intelligent methods to organize and process them. In this paper, the semantic link network model is used for organizing multimedia resources. A whole model for generating the association relation between multimedia resources using semantic link network model is proposed. The definitions, modules, and mechanisms of the semantic link network are used in the proposed method. The integration between the semantic link network and multimedia resources provides a new prospect for organizing them with their semantics. The tags and the surrounding texts of multimedia resources are used to measure their semantic association. The hierarchical semantic of multimedia resources is defined by their annotated tags and surrounding texts. The semantics of tags and surrounding texts are different in the proposed framework. The modules of semantic link network model are implemented to measure association relations. A real data set including 100 thousand images with social tags from Flickr is used in our experiments. Two evaluation methods, including clustering and retrieval, are performed, which shows the proposed method can measure the semantic relatedness between Flickr images accurately and robustly.
147 citations
TL;DR: This paper presents a set of algorithms for the reverse engineering of digital circuits starting from an unstructured netlist and resulting in a high-level netlist with components such as register files, counters, adders, and subtractors, and demonstrates that they are scalable to real designs.
Abstract: Integrated circuits (ICs) are now designed and fabricated in a globalized multivendor environment making them vulnerable to malicious design changes, the insertion of hardware Trojans/malware, and intellectual property (IP) theft. Algorithmic reverse engineering of digital circuits can mitigate these concerns by enabling analysts to detect malicious hardware, verify the integrity of ICs, and detect IP violations. In this paper, we present a set of algorithms for the reverse engineering of digital circuits starting from an unstructured netlist and resulting in a high-level netlist with components such as register files, counters, adders, and subtractors. Our techniques require no manual intervention and experiments show that they determine the functionality of ${>}{45\%}$ and up to 93% of the gates in each of the test circuits that we examine. We also demonstrate that our algorithms are scalable to real designs by experimenting with a very large, highly-optimized system-on-chip (SOC) design with over 375000 combinational elements. Our inference algorithms cover 68% of the gates in this SOC. We also demonstrate that our algorithms are effective in aiding a human analyst to detect hardware Trojans in an unstructured netlist.
128 citations
TL;DR: The proposed PUF has state of the art PUF characteristics with a good ratio of PUF response variability to response length and is not sensitive to the locking phenomenon, which challenges the use of ring oscillators for the design of both PUF and TRNG.
Abstract: This paper presents a new silicon physical unclonable function (PUF) based on a transient effect ring oscillator (TERO). The proposed PUF has state of the art PUF characteristics with a good ratio of PUF response variability to response length. Unlike RO-PUF, it is not sensitive to the locking phenomenon, which challenges the use of ring oscillators for the design of both PUF and TRNG. The novel architecture using differential structures guarantees high stability of the TERO-PUF. The area of the TERO-PUF is relatively high, but is still comparable with other PUF designs. However, since the same piece of hardware can be used for both PUF and random number generation, the proposed principle offers an interesting low area mixed solution.
123 citations
TL;DR: This paper proposes a new mobile sink routing and data gathering method through network clustering based on modified expectation-maximization technique and derives an optimal number of clusters to minimize the energy consumption.
Abstract: Recently, the big data emerged as a hot topic because of the tremendous growth of the information and communication technology. One of the highly anticipated key contributors of the big data in the future networks is the distributed wireless sensor networks (WSNs). Although the data generated by an individual sensor may not appear to be significant, the overall data generated across numerous sensors in the densely distributed WSNs can produce a significant portion of the big data. Energy-efficient big data gathering in the densely distributed sensor networks is, therefore, a challenging research area. One of the most effective solutions to address this challenge is to utilize the sink node's mobility to facilitate the data gathering. While this technique can reduce energy consumption of the sensor nodes, the use of mobile sink presents additional challenges such as determining the sink node's trajectory and cluster formation prior to data collection. In this paper, we propose a new mobile sink routing and data gathering method through network clustering based on modified expectation-maximization technique. In addition, we derive an optimal number of clusters to minimize the energy consumption. The effectiveness of our proposal is verified through numerical results.
121 citations
TL;DR: This paper proposes a 2-D Markov chain and derives the average task completion time in closed-form and proposes an efficient solution to linearize it, which is validated by extensive simulation-based studies.
Abstract: As we have known, cloud networking provides the possibility of orchestrating all resources towards different optimisation goals. For data transferring between the storage units and the processing units in big batch data (e.g., credit billing data) processing, SDN enables the programmers to customize the data routing as needed. Communication cost of large volume data transferring is non-ignorable and shall be carefully addressed in the consideration of cost efficiency.
112 citations
TL;DR: This work proposes bi-clustering and fusion (BiFu)-a newly-fashioned scheme for the cold-start problem based on the BiFu techniques under a cloud computing setting and shows that BiFu significantly alleviates theColdstart problem in terms of accuracy and scalability.
Abstract: Social recommender systems leverage collaborative filtering (CF) to serve users with content that is of potential interesting to active users. A wide spectrum of CF schemes has been proposed. However, most of them cannot deal with the cold-start problem that denotes a situation that social media sites fail to draw recommendation for new items, users or both. In addition, they regard that all ratings equally contribute to the social media recommendation. This supposition is against the fact that low-level ratings contribute little to suggesting items that are likely to be of interest of users. To this end, we propose bi-clustering and fusion (BiFu)-a newly-fashioned scheme for the cold-start problem based on the BiFu techniques under a cloud computing setting. To identify the rating sources for recommendation, it introduces the concepts of popular items and frequent raters. To reduce the dimensionality of the rating matrix, BiFu leverages the bi-clustering technique. To overcome the data sparsity and rating diversity, it employs the smoothing and fusion technique. Finally, BiFu recommends social media contents from both item and user clusters. Experimental results show that BiFu significantly alleviates the cold-start problem in terms of accuracy and scalability.
103 citations
TL;DR: This paper proposes an efficient energy management framework to control the duty cycles of physical sensors under quality-of-information (QoI) expectations in a multitask-oriented environment and introduces the novel concept of QoI-aware sensor-to-task relevancy.
Abstract: Considering physical sensors with certain sensing capabilities in an Internet-of-Things (IoTs) sensory environment, in this paper, we propose an efficient energy management framework to control the duty cycles of these sensors under quality-of-information (QoI) expectations in a multitask-oriented environment. Contrary to past research efforts, our proposal is transparent and compatible both with the underlying low-layer protocols and diverse applications, and preserving energy-efficiency in the long run without sacrificing the QoI levels attained. In particular, we first introduce the novel concept of QoI-aware sensor-to-task relevancy to explicitly consider the sensing capabilities offered by a sensor to the IoT sensory environments, and QoI requirements required by a task. Second, we propose a novel concept of the critical covering set of any given task in selecting the sensors to service a task over time. Third, energy management decision is made dynamically at runtime, to reach the optimum for long-term application arrivals and departures under the constraint of their service delay. We show a case study to utilize sensors to perform environmental monitoring with a complete set of performance analysis. We further consider the signal propagation and processing latency into the proposal, and provide a thorough analysis on its impact on average measured delay probability.
97 citations
TL;DR: MVCWalker is presented, an innovative method that stands on the shoulders of random walk with restart (RWR) for recommending collaborators to scholars and shows that incorporating the above factors into random walk model can improve the precision, recall rate, and coverage rate of academic collaboration recommendations.
Abstract: In academia, scientific research achievements would be inconceivable without academic collaboration and cooperation among researchers. Previous studies have discovered that productive scholars tend to be more collaborative. However, it is often difficult and time-consuming for researchers to find the most valuable collaborators (MVCs) from a large volume of big scholarly data. In this paper, we present MVCWalker, an innovative method that stands on the shoulders of random walk with restart (RWR) for recommending collaborators to scholars. Three academic factors, i.e., coauthor order, latest collaboration time, and times of collaboration, are exploited to define link importance in academic social networks for the sake of recommendation quality. We conducted extensive experiments on DBLP data set in order to compare MVCWalker to the basic model of RWR and the common neighbor-based model friend of friends in various aspects, including, e.g., the impact of critical parameters and academic factors. Our experimental results show that incorporating the above factors into random walk model can improve the precision, recall rate, and coverage rate of academic collaboration recommendations.
TL;DR: Cognitive acoustic is advocated as a promising technique to develop an environment-friendly UAN with high spectrum utilization and underwater cognitive acoustic networks (UCANs) also pose grand challenges due to the unique features of underwater channel and acoustic systems.
Abstract: In oceans, both the natural acoustic systems (such as marine mammals) and artificial acoustic systems [like underwater acoustic networks (UANs) and sonar users] use acoustic signal for communication, echolocation, sensing, and detection. This makes the channel spectrum heavily shared by various underwater acoustic systems. Nevertheless, the precious spectrum resource is still underutilized temporally and spatially in underwater environments. To efficiently utilize the spectrum while avoiding harmful interference with other acoustic systems, a smart UAN should be aware of the surrounding environment and reconfigure their operation parameters. Unfortunately, existing UAN designs have mainly focused on the single network scenario, and very few studies have considered the presence of nearby acoustic activities. In this paper, we advocate cognitive acoustic as a promising technique to develop an environment-friendly UAN with high spectrum utilization. However, underwater cognitive acoustic networks (UCANs) also pose grand challenges due to the unique features of underwater channel and acoustic systems. In this paper, we comprehensively investigate these unique characteristics and their impact on the UCAN design. Finally, possible solutions to tackle such challenges are advocated.
TL;DR: This paper proposes efficient power aware routing (EPAR), a newPower aware routing protocol that increases the network lifetime of MANET and reduces for more than 20% the total energy consumption and decreases the mean delay, especially for high load networks, while achieving a good packet delivery ratio.
Abstract: Notice of Violation of IEEE Publication Principles "Designing Energy Routing Protocol with Power Consumption Optimization in MANET," by Shivashankar, H.N. Suresh, G. Varaprasad, and G. Jayanthi, in the IEEE Transactions on Emerging Topics in Computing, Vol.2, No.2, June 2014, pp.192-197 After careful and considered review, it has been determined that the above paper is in violation of IEEE's Publication Principles. One of the authors, Golla Varaprasad, falsified reviewer names and provided them as recommended reviewers for the manuscript. As technology rapidly increases, diverse sensing and mobility capabilities have become readily available to devices and, consequently, mobile ad hoc networks (MANETs) are being deployed to perform a number of important tasks. In MANET, power aware is important challenge issue to improve the communication energy efficiency at individual nodes. We propose efficient power aware routing (EPAR), a new power aware routing protocol that increases the network lifetime of MANET. In contrast to conventional power aware algorithms, EPAR identifies the capacity of a node not just by its residual battery power, but also by the expected energy spent in reliably forwarding data packets over a specific link. Using a mini-max formulation, EPAR selects the path that has the largest packet capacity at the smallest residual packet transmission capacity. This protocol must be able to handle high mobility of the nodes that often cause changes in the network topology. This paper evaluates three ad hoc network routing protocols (EPAR, MTPR, and DSR) in different network scales, taking into consideration the power consumption. Indeed, our proposed scheme reduces for more than 20% the total energy consumption and decreases the mean delay, especially for high load networks, while achieving a good packet delivery ratio.
TL;DR: This paper aims to explore new challenges and opportunities of AR, by also presenting the software framework that is being developed in the EASE-R3 project by exploiting reconfigurable AR procedures and tele-assistance to overcome some of the limitations of current solutions.
Abstract: Augmented reality (AR) is a well-known technology that can be exploited to provide mass-market users an effective and customizable support in a large spectrum of personal applications, by overlapping computer-generated hints to the real world. Mobile devices, such as smartphones and tablets, are playing a key role in the exponential growth of this kind of solutions. Nonetheless, there exists some application domains that just started to take advantage from the AR systems. Maintenance, repair, and assembly have been considered as strategic fields for the application of the AR technology from the 1990s, but often only specialists using ad hoc hardware were involved in limited experimental tests. Nowadays, AR-based maintenance and repair procedures are available also for end-users on consumer electronics devices. This paper aims to explore new challenges and opportunities of this technology, by also presenting the software framework that is being developed in the EASE- $\text{R}^{3}$ project by exploiting reconfigurable AR procedures and tele-assistance to overcome some of the limitations of current solutions.
TL;DR: A semi-online EDF-based scheduling algorithm theoretically optimal (i.e., processing and energy costs neglected) that relies on the notions of energy demand and slack energy, which are different from the well known notions of processorDemand and slack time.
Abstract: In this paper, we study a scheduling problem, in which every job is associated with a release time, deadline, required computation time, and required energy. We focus on an important special case where the jobs execute on a uniprocessor system that is supplied by a renewable energy source and use a rechargeable storage unit with limited capacity. Earliest deadline first (EDF) is a class one online algorithm in the classical real-time scheduling theory where energy constraints are not considered. We propose a semi-online EDF-based scheduling algorithm theoretically optimal (i.e., processing and energy costs neglected). This algorithm relies on the notions of energy demand and slack energy, which are different from the well known notions of processor demand and slack time. We provide an exact feasibility test. There are no restrictions on this new scheduler: each job can be one instance of a periodic, aperiodic, or sporadic task with deadline.
TL;DR: A detailed survey on the state-of-the-art in scan-based side-channel attacks on symmetric and public-key cryptographic hardware implementations, both in the absence and presence of advanced DfT structures, such as test compression and X-masking, which may make the attack difficult.
Abstract: Cryptographic circuits need to be protected against side-channel attacks, which target their physical attributes while the cryptographic algorithm is in execution. There can be various side-channels, such as power, timing, electromagnetic radiation, fault response, and so on. One such important side-channel is the design-for-testability (DfT) infrastructure present for effective and timely testing of VLSI circuits. The attacker can extract secret information stored on the chip by scanning out test responses against some chosen plaintext inputs. The purpose of this paper is to first present a detailed survey on the state-of-the-art in scan-based side-channel attacks on symmetric and public-key cryptographic hardware implementations, both in the absence and presence of advanced DfT structures, such as test compression and X-masking, which may make the attack difficult. Then, the existing scan attack countermeasures are evaluated for determining their security against known scan attacks. In addition, JTAG vulnerability and security countermeasures are also analyzed as part of the external test interface. A comparative area-timing-security analysis of existing countermeasures at various abstraction levels is presented in order to help an embedded security designer make an informed choice for his intended application.
TL;DR: This paper presents the first systematic work to design a unified recommendation system for both the regular and carpooling services, called CallCab, based on a data-driven approach, and designs a reciprocal price mechanism to facilitate the taxicab carpooled implementation in the real world.
Abstract: Carpooling taxicab services hold the promise of providing additional transportation supply, especially in the extreme weather or rush hour when regular taxicab services are insufficient. Although many recommendation systems about regular taxicab services have been proposed recently, little research, if any, has been done to assist passengers to find a successful taxicab ride with carpooling. In this paper, we present the first systematic work to design a unified recommendation system for both the regular and carpooling services, called CallCab, based on a data-driven approach. In response to a passenger’s real-time request, CallCab aims to recommend either: 1) a vacant taxicab for a regular service with no detour or 2) an occupied taxicab heading to the similar direction for a carpooling service with the minimum detour, yet without assuming any knowledge of destinations of passengers already in taxicabs. To analyze these unknown destinations of occupied taxicabs, CallCab generates and refines taxicab trip distributions based on GPS data sets and context information collected in the existing taxicab infrastructure. To improve CallCab’s efficiency to process such a big data set, we augment the efficient MapReduce model with a Measure phase tailored for our CallCab. Finally, we design a reciprocal price mechanism to facilitate the taxicab carpooling implementation in the real world. We evaluate CallCab with a real-world data set of 14000 taxicabs, and results show that compared with the ground truth, CallCab reduces 60% of the total mileage to deliver all passengers and 41% of passenger’s waiting time. Our price mechanism reduces 23% of passengers’ fares and increases 28% of drivers’ profits simultaneously.
TL;DR: This paper presents the first comprehensive analysis of the impact of failures on energy consumption in a real-world large-scale cloud system (comprising over 12 500 servers), including the study of failure and energy trends of the spatial and temporal environmental characteristics.
Abstract: Cloud computing providers are under great pressure to reduce operational costs through improved energy utilization while provisioning dependable service to customers; it is therefore extremely important to understand and quantify the explicit impact of failures within a system in terms of energy costs This paper presents the first comprehensive analysis of the impact of failures on energy consumption in a real-world large-scale cloud system (comprising over 12 500 servers), including the study of failure and energy trends of the spatial and temporal environmental characteristics Our results show that 88% of task failure events occur in lower priority tasks producing 13% of total energy waste, and 1% of failure events occur in higher priority tasks due to server failures producing 8% of total energy waste These results highlight an unintuitive but significant impact on energy consumption due to failures, providing a strong foundation for research into dependable energy-aware cloud computing
TL;DR: Experimental results show that the evolutionary approach compared with existing methods, such as Monte Carlo and Blind Pick, can achieve higher overall average scheduling performance in real-world applications with dynamic workloads and an optimal computing budget allocating method that smartly allocates computing cycles to the most promising schedules.
Abstract: Scheduling of dynamic and multitasking workloads for big-data analytics is a challenging issue, as it requires a significant amount of parameter sweeping and iterations. Therefore, real-time scheduling becomes essential to increase the throughput of many-task computing. The difficulty lies in obtaining a series of optimal yet responsive schedules. In dynamic scenarios, such as virtual clusters in cloud, scheduling must be processed fast enough to keep pace with the unpredictable fluctuations in the workloads to optimize the overall system performance. In this paper, ordinal optimization using rough models and fast simulation is introduced to obtain suboptimal solutions in a much shorter timeframe. While the scheduling solution for each period may not be the best, ordinal optimization can be processed fast in an iterative and evolutionary way to capture the details of big-data workload dynamism. Experimental results show that our evolutionary approach compared with existing methods, such as Monte Carlo and Blind Pick, can achieve higher overall average scheduling performance, such as throughput, in real-world applications with dynamic workloads. Furthermore, performance improvement is seen by implementing an optimal computing budget allocating method that smartly allocates computing cycles to the most promising schedules.
TL;DR: This paper addresses the problem of energy-aware heterogeneous data allocation and task scheduling on heterogeneous multiprocessor systems for real-time applications and proposes an optimal approach, i.e., an integer linear programming method, to solve this problem.
Abstract: In this paper, we address the problem of energy-aware heterogeneous data allocation and task scheduling on heterogeneous multiprocessor systems for real-time applications. In a heterogeneous distributed shared-memory multiprocessor system, an important problem is how to assign processors to real-time application tasks, allocate data to local memories, and generate an efficient schedule in such a way that a time constraint can be met and the total system energy consumption can be minimized. We propose an optimal approach, i.e., an integer linear programming method, to solve this problem. As the problem has been conclusively shown to be computationally very complicated, we also present two heuristic algorithms, i.e., task assignment considering data allocation (TAC-DA) and task ratio greedy scheduling (TRGS), to generate near-optimal solutions for real-time applications in polynomial time. We evaluate the performance of our algorithms by comparing them with a greedy algorithm that is commonly used to solve heterogeneous task scheduling problems. Based on our extensive simulation study, we observe that our algorithms exhibit excellent performance. We conducted experimental performance evaluation on two heterogeneous multiprocessor systems. The average reduction rates of the total energy consumption of the TAC-DA and TRGS algorithms to that of the greedy algorithm are 13.72% and 15.76%, respectively, on the first system, and 19.76% and 24.67%, respectively, on the second system. To the best of our knowledge, this is the first study to solve the problem of task scheduling incorporated with data allocation and energy consumption on heterogeneous distributed shared-memory multiprocessor systems.
TL;DR: A clustering-based collaborative filtering approach is proposed in this paper, which aims at recruiting similar services in the same clusters to recommend services collaboratively, to reduce the online execution time of collaborative filtering.
Abstract: Spurred by service computing and cloud computing, an increasing number of services are emerging on the Internet. As a result, service-relevant data become too big to be effectively processed by traditional approaches. In view of this challenge, a clustering-based collaborative filtering approach is proposed in this paper, which aims at recruiting similar services in the same clusters to recommend services collaboratively. Technically, this approach is enacted around two stages. In the first stage, the available services are divided into small-scale clusters, in logic, for further processing. At the second stage, a collaborative filtering algorithm is imposed on one of the clusters. Since the number of the services in a cluster is much less than the total number of the services available on the web, it is expected to reduce the online execution time of collaborative filtering. At last, several experiments are conducted to verify the availability of the approach, on a real data set of 6225 mashup services collected from ProgrammableWeb.
TL;DR: Experimental results demonstrate that the achieved demanded throughput is improved 10.7% for the most stringent ToD requirement, while the residual battery energy of the RLTDPM is improved 7.4% compared with an existing DPM algorithm for EHWSN with image sensing purpose.
Abstract: In this paper, a reinforcement learning-based throughput on demand (ToD) provisioning dynamic power management method (RLTDPM) is proposed for sustaining perpetual operation and satisfying the ToD requirements for today's energy harvesting wireless sensor node (EHWSN). The RLTDPM monitors the environmental state of the EHWS and adjusts their operational duty cycle under criteria of energy neutrality to meet the demanded throughput. Outcomes of these observation-adjustment interactions are then evaluated by feedback/reward that represents how well the ToD requests are met; subsequently, the observation-adjustment-evaluation process, so-called reinforcement learning, continues. After the learning process, the RLTDPM is able to autonomously adjust the duty cycle for satisfying the ToD requirement, and in doing so, sustain the perpetual operation of the EHWSN. Simulations of the proposed RLTDPM on a wireless sensor node powered by a battery and solar cell for image sensing tasks were performed. Experimental results demonstrate that the achieved demanded throughput is improved 10.7% for the most stringent ToD requirement, while the residual battery energy of the RLTDPM is improved 7.4% compared with an existing DPM algorithm for EHWSN with image sensing purpose.
TL;DR: This work proposes to incorporate trojan toleration into MPSoC platforms by revising the task scheduling step of theMPSoC design process, and imposes a set of security-driven diversity constraints into the scheduling process, enabling the system to detect the presence of malicious modifications or to mute their effects during application execution.
Abstract: Multiprocessor system-on-chip (MPSoC) platforms face some of the most demanding security concerns, as they process, store, and communicate sensitive information using third-party intellectual property (3PIP) cores. The complexity of MPSoC makes it expensive and time consuming to fully analyze and test during the design stage. This has given rise to the trend of outsourcing design and fabrication of 3PIP components, that may not be trustworthy. To protect MPSoCs against malicious modifications, we impose a set of security-driven diversity constraints into the task scheduling step of the MPSoC design process, enabling the system to detect the presence of malicious modifications or to mute their effects during application execution. We pose the security-constrained MPSoC task scheduling as a multidimensional optimization problem, and propose a set of heuristics to ensure that the introduced security constraints can be fulfilled with a minimum impact on the other design goals such as performance and hardware. Experimental results show that without any extra cores, security constraints can be fulfilled within four vendors and 81% overhead in schedule length.
TL;DR: The results obtained demonstrate that LIME classifiers can significantly increase the accuracy of classifications and function as integral parts of other ensemble meta classifiers at higher tiers.
Abstract: This paper introduces and investigates large iterative multitier ensemble (LIME) classifiers specifically tailored for big data. These classifiers are very large, but are quite easy to generate and use. They can be so large that it makes sense to use them only for big data. They are generated automatically as a result of several iterations in applying ensemble meta classifiers. They incorporate diverse ensemble meta classifiers into several tiers simultaneously and combine them into one automatically generated iterative system so that many ensemble meta classifiers function as integral parts of other ensemble meta classifiers at higher tiers. In this paper, we carry out a comprehensive investigation of the performance of LIME classifiers for a problem concerning security of big data. Our experiments compare LIME classifiers with various base classifiers and standard ordinary ensemble meta classifiers. The results obtained demonstrate that LIME classifiers can significantly increase the accuracy of classifications. LIME classifiers performed better than the base classifiers and standard ensemble meta classifiers.
TL;DR: This paper demonstrates zero overhead malicious modifications on both high-performance and embedded microprocessors that enable privilege escalation through execution of an instruction stream that excites the necessary conditions to make the modification appear.
Abstract: The wide deployment of general purpose and embedded microprocessors has emphasized the need for defenses against cyber-attacks. Due to the globalized supply chain, however, there are several stages where a processor can be maliciously modified. The most promising stage, and the hardest during which to inject the hardware trojan, is the fabrication stage. As modern microprocessor chips are characterized by very dense, billion-transistor designs, such attacks must be very carefully crafted. In this paper, we demonstrate zero overhead malicious modifications on both high-performance and embedded microprocessors. These hardware trojans enable privilege escalation through execution of an instruction stream that excites the necessary conditions to make the modification appear. The minimal footprint, however, comes at the cost of a small window of attack opportunities. Experimental results show that malicious users can gain escalated privileges within a few million clock cycles. In addition, no system crashes were reported during normal operation, rendering the modifications transparent to the end user.
TL;DR: A wavelet transform is used to represent remote sensing big data that are large scale in the space domain, correlated in the spectral domain, and continuous in the time domain and it is found that the scale features of different textures for the big data set are obviously reflected in the probability density function and GMM parameters of the wavelet coefficients.
Abstract: Since it is difficult to deal with big data using traditional models and algorithms, predicting and estimating the characteristics of big data is very important. Remote sensing big data consist of many large-scale images that are extremely complex in terms of their structural, spectral, and textual features. Based on multiresolution analysis theory, most of the natural images are sparse and have obvious clustering and persistence characters when they are transformed into another domain by a group of basic special functions. In this paper, we use a wavelet transform to represent remote sensing big data that are large scale in the space domain, correlated in the spectral domain, and continuous in the time domain. We decompose the big data set into approximate multiscale detail coefficients based on a wavelet transform. In order to determine whether the density function of wavelet coefficients in a big data set are peaky at zero and have a heavy tailed shape, a two-component Gaussian mixture model (GMM) is employed. For the first time, we use the expectation-maximization likelihood method to estimate the model parameters for the remote sensing big data set in the wavelet domain. The variance of the GMM with changing of bands, time, and scale are comprehensively analyzed. The statistical characteristics of different textures are also compared. We find that the cluster characteristics of the wavelet coefficients are still obvious in the remote sensing big data set for different bands and different scales. However, it is not always precise when we model the long-term sequence data set using the GMM. We also found that the scale features of different textures for the big data set are obviously reflected in the probability density function and GMM parameters of the wavelet coefficients.
TL;DR: This paper develops an algorithm that leverages intentional post-silicon aging to tune the inter- and intra-chip signatures variation and proposes a novel lightweight (low overhead) strong PUF based on the timings of a classic processor architecture.
Abstract: A strong physically unclonable function (PUF) is a circuit structure that extracts an exponential number of unique chip signatures from a bounded number of circuit components. The strong PUF unique signatures can enable a variety of low-overhead security and intellectual property protection protocols applicable to several computing platforms. This paper proposes a novel lightweight (low overhead) strong PUF based on the timings of a classic processor architecture. A small amount of circuitry is added to the processor for on-the-fly extraction of the unique timing signatures. To achieve desirable strong PUF properties, we develop an algorithm that leverages intentional post-silicon aging to tune the inter- and intra-chip signatures variation. Our evaluation results show that the new PUF meets the desirable inter- and intra-chip strong PUF characteristics, whereas its overhead is much lower than the existing strong PUFs. For the processors implemented in 45 nm technology, the average inter-chip Hamming distance for 32-bit responses is increased by 16.1% after applying our post-silicon tuning method; the aging algorithm also decreases the average intra-chip Hamming distance by 98.1% (for 32-bit responses).
TL;DR: This paper proposes two online ensemble learning methods for workload prediction, which address the issues that arise specifically in large-scale server systems, viz., extensive nonstationarity of server workloads, and massive online streaming data.
Abstract: Increasing energy costs of large-scale server systems have led to a demand for innovative methods for optimizing resource utilization in these systems. Such methods aim to reduce server energy consumption, cooling requirements, carbon footprint, and so on, thereby leading to improved holistic sustainability of the overall server infrastructure. At the core of many of these methods lie reliable workload-prediction techniques that guide in identifying servers, time intervals, and other parameters that are needed for building sustainability solutions based on techniques like virtualization and server consolidation for server systems. Many workload prediction methods have been proposed in the recent paper, but unfortunately they do not deal adequately with the issues that arise specifically in large-scale server systems, viz., extensive nonstationarity of server workloads, and massive online streaming data. In this paper, we fill this gap by proposing two online ensemble learning methods for workload prediction, which address these issues in large-scale server systems. The proposed algorithms are motivated from the weighted majority and simulatable experts approaches, which we extend and adapt to the large-scale workload prediction problem. We demonstrate the effectiveness of our algorithms using real and synthetic data sets, and show that using the proposed algorithms, the workloads of 91% of servers in a real data center can be predicted with accuracy > 89%, whereas using baseline approaches, the workloads of only 13%–24% of the servers can be predicted with similar accuracy.
TL;DR: Simulation results show that the performance of the distributed algorithms, specifically the DPGS algorithm, is very competitive with that of the centralized algorithm while providing the advantage of naturally adapting to time-varying communication patterns of users.
Abstract: We study the problem of assigning users to servers with an emphasis on the distributed algorithmic solutions. Typical online social network applications, such as Facebook and Twitter, are built on top of an infrastructure of servers that provides the services on behalf of the users. For a given communication pattern among users, the loads of the servers depend critically on how the users are assigned to the servers. A good assignment will reduce the overall load of the system while balancing the loads among the servers. Unfortunately, this optimal assignment problem is NP-hard. Therefore, we investigate three heuristic algorithms for solving the user server assignment problem: 1) the centralized simulated annealing (CSA) algorithm; 2) the distributed simulated annealing (DSA) algorithm; and 3) the distributed perturbed greedy search (DPGS). The CSA algorithm produces good solution in the fastest time, however it relies on timely accurate global system information, and is practical only for small and static systems. In contrast, the two distributed algorithms, DSA and DPGS, exploit local information at each server during their search for the optimal assignment, and thus can scale well with the number of users and servers as well as adapting to the system dynamics. Simulation results show that the performance of the distributed algorithms, specifically the DPGS algorithm, is very competitive with that of the centralized algorithm while providing the advantage of naturally adapting to time-varying communication patterns of users.
TL;DR: This case study uses a case study of energy waste on computer workstations to motivate the incorporation of humans into the control loops and develops a human-in-the-loop control that can put workst stations into sleep by early detection of distraction.
Abstract: Although current cyber physical systems (CPSs) act as the bridge between humans and environment, their implementation mostly assumes humans as an external component to the control loops. We use a case study of energy waste on computer workstations to motivate the incorporation of humans into the control loops. The benefits include better response accuracy and timeliness of the CPS systems. However, incorporating humans into tight control loops remains a challenge as it requires understanding complex human behavior. In our case study, we collect empirical data to understand human behavior regarding distractions in computer usage and develop a human-in-the-loop control that can put workstations into sleep by early detection of distraction. Our control loop implements strategies such as an adaptive timeout interval, multilevel sensing, and addressing background processing. Evaluation on multiple subjects show an accuracy of 97.28% in detecting distractions, which cuts the energy waste of computers by 80.19%.