Search

Humans can easily parse and find answers to complex queries such as "What was the capital of the country of the discoverer of the element which has atomic number 1?" by breaking them up into small pieces, querying these appropriately, and assembling a final answer. However, contemporary search engines lack such capability and fail to handle even slightly complex queries. Search engines process queries by identifying keywords and searching against them in knowledge bases or indexed web pages. The results are, therefore, dependent on the keywords and how well the search engine handles them. In our work, we propose a three-step approach called parsing, tree generation, and querying (PTGQ) for effective searching of larger and more expressive queries of potentially unbounded complexity. PTGQ parses a complex query and constructs a query tree where each node represents a simple query. It then processes the complex query by recursively querying a back-end search engine, going over the corresponding query tree in postorder. Using PTGQ makes sure that the search engine always handles a simpler query containing very few keywords. Results demonstrate that PTGQ can handle queries of much higher complexity than standalone search engines.

A problem currently faced by autonomous systems, learning agents, multi-agent systems and other cognitive systems is the extreme susceptibility to bad information and malicious misdirection. In this paper, we propose a strategy to identify trusted agents in a multi-agent system, so that a system can eliminate the knowledge provided by malicious or unknown agents. The strategy is covered in three phases by each agent; firstly sharing information with other agents in the system, secondly calculating the conclusion based on the majority of information received from others and lastly identifying the trusted, malicious and unknown agents in the system. The theoretical analysis shows that the trusted agent strategy (TAS) is capable of identifying malicious or compromised agents if a system has less than 25% of malicious agents. This is further improved in the advanced-TAS which can handle up to 40% malicious agents. The experimental results show that trusted and malicious agents can be identified with high accuracy.

Virtualization of resources on the cloud offers a scalable means of consuming services beyond the capabilities of small systems. In a cloud that offers infrastructure such as processor, memory, hard disk, etc., a coalition of virtual machines formed by grouping two or more may be needed. Economical management of cloud resources needs allocation strategies with minimum wastage, while configuring services ahead of actual requests. We propose a resource allocation mechanism for machines on the cloud, based on the principles of coalition formation and the uncertainty principle of game theory. We compare the results of applying this mechanism with existing resource allocation methods that have been deployed on the cloud. We also show that this method of resource allocation by coalition-formation of the machines on the cloud leads not only to better resource utilization but also higher request satisfaction.

Vaccine development is a laborious and time-consuming process and can benefit from statistical machine learning techniques, which can produce general outcomes based on the patterns observed in the limited available empirical data. In this paper, we show how limited gene expression data from a small sample of subjects can be used to predict the outcomes of malaria vaccine. In addition, we also draw inferences from the gene expression data, with over 22000 columns (or features), by visualizing the data, and reduce the data dimensions based on this inference for efficient model training. Our methods are general and reliable and can be extended to vaccines developed against any pathogen. Given the gene expression data from a sample of subjects administered with a novel vaccine, our methods can be used to test the outcome of that vaccine, without the need for empirical observations on a larger population. By carefully tuning the available data and the machine learning models, we are able to achieve greater than 98% accuracy, with sensitivity and specificity of 0.93 and 1 respectively, in predicting the outcomes of the malaria vaccine in developing immunogenicity against the malaria pathogen.

Due to its widespread practical applications to image database management, semantic image retrieval has received a lot of attention in past decades. In particular, relevance feedback-based methods have been a popular approach toward bridging the gap between low-level features and high-level semantic concepts. Nature-inspired algorithms have shown a lot of potential to solve problems which are complex and require discovering of patterns under changing environment. However, lack of sound mathematical foundations has been considered a drawback toward better analysis of these algorithms. In this chapter, we propose a novel general topological model for semantic image retrieval using relevance feedback. We use point-set topology to develop mathematical constructs for modeling the semantic retrieval. In particular, we develop an image retrieval algorithm based on the ant sleeping model and extend the topological model to analyze it. Through experiments we show that our algorithm performs well in indexing an image database for relevance feedback. With our indexing procedure, the average response time to access image results from a storage device is lower when compared to vector quantization techniques. We also evaluate our algorithm, theoretically and empirically, against PicSOM (a CBIR system based on relevance feedback). Our ASM-based technique shows a very efficient retrieval performance using relevance feedback.

High sparsity and the problem of overspecialization are challenges faced by collaborative filtering (CF) algorithms in recommender systems. In this paper, we design an approach that efficiently tackles the above problems. We address the first issue of high sparsity in CF by modifying the popular parallel seeding technique proposed by Bahmani et al. for use in a spherical setting, for the clustering of users. Experimental evaluations on highly sparse real world datasets demonstrate the better performance of our algorithm than existing Spherical K-Means algorithms. Contrary to the common belief that users are only interested in items similar to those of their previous liking, it has been well established that including serendipitous recommendations improves their satisfaction. Thus, to tackle the second problem of overspecialization, we effectuate serendipity in movie recommender systems with an end-to-end algorithm, Serendipitous Clustering for Collaborative Filtering (SC-CF) that considers diversity, unexpectedness and relevance. SC-CF takes advantage of carefully generated user profiles to then assign users to a "serendipitous cluster." The overall clustering process leverages these user profiles and improves recommendation quality through serendipity. A series of experiments on Serendipity 2018 (part of Movielens) dataset built on real user feedback has shown that SC-CF outperforms the existing popular recommendation methods like K-Means and deep learning based CF.

As military, academic, and commercial computing systems evolve from autonomous entities that deliver computing products into network centric enterprise systems that deliver computing as a service, opportunities emerge to consolidate computing resources, software, and information through cloud computing. Along with these opportunities come challenges, particularly to service providers and operations centers that struggle to monitor and manage quality of service (QoS) for these services in order to meet customer service commitments. Traditional approaches fall short in addressing these challenges because they examine QoS from a limited perspective rather than from a system-of-systems (SoS) perspective applicable to a net-centric enterprise system in which any user from any location can share computing resources at any time. This paper presents a SoS approach to enable QoS monitoring, management, and response for enterprise systems that deliver computing as a service through a cloud computing environment. A concrete example is provided for application of this new SoS approach to a real-world scenario (viz., distributed denial of service). Simulated results confirm the efficacy of the approach.

Neural Machine Translation (NMT) is a new technique for machine translation that has led to remarkable improvements compared to rule-based and statistical machine translation (SMT) techniques, by overcoming many of the weaknesses in the conventional techniques. We study and apply NMT techniques to create a system with multiple models which we then apply for six Indian language pairs. We compare the performances of our NMT models with our system using automatic evaluation metrics such as UNK Count, METEOR, F-Measure, and BLEU. We find that NMT techniques are very effective for machine translations of Indian language pairs. We then demonstrate that we can achieve good accuracy even using a shallow network; on comparing the performance of Google Translate on our test dataset, our best model outperformed Google Translate by a margin of 17 BLEU points on Urdu-Hindi, 29 BLEU points on Punjabi-Hindi, and 30 BLEU points on Gujarati-Hindi translations.

The cost of electricity contributes significantly to the operating expense incurred in hosting cloud services. It is necessary to consider this cost while charging the consumers for their service utilization. In this work, we arrive at a metering mechanism for cloud services, in which the price of a cloud service tracks the variable input cost of electricity from a smart grid. The power-aware cloud metering developed here is a dynamic pricing and billing model where tariff for a cloud service is varied in accordance with the input electricity cost. We arrive at a model for power consumption of virtual machines hosted on the cloud infrastructure. This power consumption model is used in calculating the cost of operation of the service. A cloud instance leased by a consumer is billed based on the cost of operation obtained, and its resource utilization. Experimental results validate the approach presented.

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.

The rapid spread of fake news is a serious problem calling for AI solutions. We employ a deep learning based automated detector through a three level hierarchical attention network (3HAN) for fast, accurate detection of fake news. 3HAN has three levels, one each for words, sentences, and the headline, and constructs a news vector: an effective representation of an input news article, by processing an article in an hierarchical bottom-up manner. The headline is known to be a distinguishing feature of fake news, and furthermore, relatively few words and sentences in an article are more important than the rest. 3HAN gives a differential importance to parts of an article, on account of its three layers of attention. By experiments on a large real-world data set, we observe the effectiveness of 3HAN with an accuracy of 96.77%. Unlike some other deep learning models, 3HAN provides an understandable output through the attention weights given to different parts of an article, which can be visualized through a heatmap to enable further manual fact checking.

We present EEQuest, an application that extracts events from text using natural language processing (NLP) and supervised machine-learning techniques, and provides a system to query events extracted from a text corpus. We provide a use case for the application wherein we extract business-related events from news articles. The extracted events are then categorized based on the business organization/company that they are related to. Finally, the events are added to a knowledge base using which a query system is built. The system can be used to display events related to a particular organization or a group of organizations. Although we are using the system to extract business-related events, the event extraction mechanism can be used in a more general sense with any available textual data, to extract any kind of events that have a structure that can answer the question: Who did what, when and where?

A unique and efficient system model is proposed for sensor networks to create sensor-smart highways. The approach first considers various cell geometries encountered with 2-D roads and then extends the same to 3-D roads. Mathematical relations describing each type of cell are derived, and the analyses indicate the numbers and positions of sensor nodes needed in each cell of a certain type. We then propose algorithms for computing the deployment positions of sensor nodes on roads by describing them as consisting of cells of different geometries. We also validate the approach by computing a sensor deployment using real geospatial data.

—With the growth of cloud computing, large scale data centers have become common in the computing industry, and there has been a significant increase in energy consumption at these data centers, which thus becomes a key issue to address. As most of the time a data center remains underutilized, a significant amount of energy can be conserved by migrating virtual machines (VM) running on underutilized machines to other machines and hibernating such underutilized machines. This paper aims to design such a strategy for energy-efficient cloud data centers. It makes use of historical traffic data from data centers and uses a service request prediction model which enables the identification of the number of active servers required at a given moment, thus making possible the hibernation of underutilized servers. The simulation results indicate that this approach brings about a significant amount of energy conservation.

Astronomical datasets are typically very large, and manually classifying the data in them is effectively impossible. We use machine learning algorithms to provide classifications (as stars, quasars and galaxies) for more than one billion objects given photometrically in the Third Data Release of the Sloan Digital Sky Survey (SDSS-III). We have used k NN, SVM and random forest algorithms in a distributed environment over the cloud to classify 1,183,850,913 unclassified photometric objects present in the SDSSIII catalog. This catalog contains photometric data for all objects viewed through a telescope and spectroscopic data for a small part of these. Although it is possible to classify all the objects using spectroscopic data, it is impractical to obtain such data for each one of them. To classify such a big dataset on a single machine would be impractically slow, so we have used the Spark cluster computing framework to implement a distributed computing environment over the cloud. We found that writing results (dozens of gigabytes) to the cloud storage is very slow while using k NN. Though writing the results with SVM is faster as it is done in parallel, its accuracy is only around 87%, due to lack of a kernel implementation of it in Spark. We then used the random forest algorithm to classify the entire set of 1,183,850,913 objects with an accuracy of 94% in about 17 hours of processing time. The result set is significant as even collecting spectroscopic data for these many objects would take decades, and our classifications can help astronomers and astrophysicists carry out further studies.

DoD, government agency, and commercial information technology systems are rapidly evolving from isolated enclaves to fully interoperable, net-centric, cloud-based enterprise systems where the vast amount of available data on which to make decisions is overwhelming end-users (e.g., mission commanders, operators and analysts). Many of these data are not useful for the mission at hand and must be extracted from the mission-relevant information before end-users can make decisions. These end-users seek a solution that allows them to observe relevant information in a timely manner and then to use this information to appropriately respond through information exploitation. This paper proposes a new and innovative solution to the above challenges that enables end-users to make timely mission decisions using a Cloud-based Mission Observation, Response, and Exploitation (CMORE) System that includes a mobile situational awareness (SA) console to identify information and to determine the correct response to mission events through information exploitation over the cloud.

Apache Hadoop is an open source framework for processing massive amount of data in a distributed environment. Hadoop services use a polling mechanism for event notifications. In this paper, we propose a distributed notification system for Hadoop based on the Publish-Subscribe model. Such a notification system can be used for message-passing among Hadoop services. It can also be used to chain multiple MapReduce jobs based on events occuring in a Hadoop cluster. This results in reduced cluster load and network bandwidth requirement. We have used two popular Publish-Subscribe-based messaging systems—Apache ActiveMQ and Apache Kafka—for implementation. Lastly, we have executed performance tests on both these messaging systems to monitor time taken for message delivery and reception.

Predictive workload analytics for server systems has been the focus of recent research in energy-aware computing, with many algorithmic and architectural techniques proposed to analyze, predict, and optimize server workloads in order to build energy-aware IT systems. These techniques, though effective in optimizing server workloads, often ignore the green-ness aspect 'in' the technique itself and incur heavy computational costs in their operations. In this paper, we propose a meta-learning based architecture for building server workload prediction mechanism using which the computational cost of holistic predictive workload analytics can be optimized, and green-ness 'in' the analytics can be achieved. We also present an algorithmic optimization of the proposed meta-learning architecture for handling concept drift in server workloads, thereby also achieving improved greenness 'by' the analytics. Our experiments show that the proposed meta-learning based architecture substantially reduces the total computational cost of workload prediction mechanism, with a minor decrease of 0.5--1.3% in the accuracy, and the proposed algorithmic optimization significantly improves accuracy of the workload prediction mechanism in concept drift scenarios by 8.1%.

Growing scale of server infrastructure in large datacenters has led to an increased need for effective server workload prediction mechanisms. Two main challenges faced in server workload prediction task are lack of large-scale training data and changes in the underlying distribution of server workloads in events like change in dominant applications of servers or change in allocation of servers, etc. In this work, we propose an online server workload prediction approach based on ensemble learning which addresses these issues. We evaluate the proposed approach using real dataset of an enterprise data center and a synthetic dataset. Experimental results reveal that the proposed approach achieves accuracy of 87.8% on real dataset and 88.8% on synthetic dataset.

We present a virtualized setup of a Hadoop cluster that provides greater computing capacity with lesser resources, since a virtualized cluster requires fewer physical machines. The master node of the cluster is set up on a physical machine, and slave nodes are set up on virtual machines (VMs) that may be on a common physical machine. Hadoop configured VM images are created by cloning of VMs, which facilitates fast addition and deletion of nodes in the cluster without much overhead. Also, we have configured the Hadoop virtualized cluster to use capacity scheduler instead of the default FIFO scheduler. The capacity scheduler schedules tasks based on the availability of RAM and virtual memory (VMEM) in slave nodes before allocating any job. So instead of queuing up the jobs, they are efficiently allocated on the VMs based on the memory available. Various configuration parameters of Hadoop are analyzed and the virtualized cluster is fine-tuned to ensure best performance and maximum scalability.

Data streams are vast sources of information. In such streams only certain episodic points called events are interesting. The number of events occurring in a data stream can be very large, and a given user would not be interested in all of them. Thus, there is a need to selectively disseminate information about the occurrence of events to users who express interest in them. Publish/Subscribe is a distributed paradigm of messaging in which events published by publishers are automatically sent to subscribers who have expressed interest in them. In this work, we apply the publish/subscribe paradigm to notify a user when an event of interest occurs. Furthermore, user subscriptions may often be vague and non-specific. To capture this uncertainty, we use fuzzy set theory in the modeling of the subscriptions. The use of fuzzy sets gives a user more expressive power in subscriptions and gives better notifications of events.

As military, academic, and commercial computing systems evolve from autonomous entities that deliver computing products into network centric enterprise systems that deliver computing as a service, opportunities emerge to consolidate computing resources, software and information through cloud computing. Along with these opportunities come challenges, especially to service providers and operations centers that struggle to monitor and manage Quality of Service (QoS) for these services in order to meet customer service commitments. Traditional approaches fall short in addressing these challenges because they examine QoS from a limited perspective rather than from a System of Systems (SoS) perspective applicable to a net-centric enterprise system in which any user from any location can share computing resources at any time. This paper presents a SoS approach to provide QoS monitoring, management, and response for enterprise systems that delivers computing as a service through a cloud computing environment. Concrete examples are provided that identify the key components of this SoS and their application to real-world scenarios. Simulated results demonstrate the effectiveness of the approach for a representative scenario.

Even though the adoption of cloud computing and virtualization have improved resource utilization to a great extent, the continued traditional approach of resource allocation in production environments has introduced the problem of over-provisioning of resources for enterprise-class applications hosted in cloud systems. In this paper, we address the problem and propose ways to minimize over-provisioning of IT resources in multi-tier cloud systems by adopting an innovative approach of application performance monitoring and resource allocation at individual tier levels, on the basis of criticality of the business services and availability of the resources at one's disposal.

Our approach to deduplication and compression in cloud computing aims at reduction in storage space and bandwidth usage during file transfers. The design depends on multiple metadata structures for deduplication. Only a copy of the duplicate files is retained while others are deleted. The existence of duplicate files is determined from the metadata. The files are clustered into bins depending on their size. They are then segmented, deduplicated, compressed and stored. Binning restricts the number of segments and their sizes so that it is optimum for each file size. When the user requests a file, compressed segments of the file are sent over the network along with the file-to-segment mapping. These are the uncompressed and combined to create a complete file, hence minimizing bandwidth requirements.

The publish/subscribe paradigm enables creation of loosely coupled dynamic systems with asynchronous communication models but at the cost of unpredictable system behavior. Hence quality of service guarantees (QoS) are to be provided to bring in predictability in system behavior. There is a cost associated with providing the QoS guarantees and this cost needs to be reduced. In a publish/subscribe system, published information has to be delivered to a set of subscribers who have expressed interest in the publications. Optimal routing of information from producers of information to the consumers of information is critical to the performance of the publish/subscribe system. This paper proposes a strategy for routing of publications in a publish/subscribe system by modeling it as a multi commodity transportation problem. The proposed approach minimizes the cost of routing publications and also maintains the QoS guarantees.

XML files are ideal for describing semi structured data. The advent of native XML databases has enhanced storage of a large number of XML files preserving the hierarchical format. The challenge is to serve and access XML data from geographically distributed native XML databases. In this paper, we discuss the modus operandi for insertion and querying of data in a location transparent manner in decentralised networks. We present the design for a cluster based architecture where query processingF is achieved using distributed hash tables. A bulk insertion method has been employed for insertion of XML data. The proposed design was implemented on eXist-db, a native XML database engine. Experiments were conducted and the effectiveness of the proposed design was evaluated.

Numerous automated anomaly detection and application performance modeling and management tools are available to detect and diagnose faulty application behavior. However, these tools have limited utility in ‘on-demand’ virtual computing infrastructures because of the increased tendencies for the applications in virtual machines to migrate across un-comparable hosts in virtualized environments and the unusually long latency associated with the training phase. The relocation of the application subsequent to the training phase renders the already collected data meaningless and the tools need to re-initiate the learning process on the new host afresh. Further, data on several metrics need to be correlated and analyzed in real time to infer application behavior. The multivariate nature of this problem makes detection and diagnosis of faults in real time all the more challenging as any suggested approach must be scalable. In this paper, we provide an overview of a system architecture for detecting and diagnosing anomalous application behaviors even as applications migrate from one host to another and discuss a scalable approach based on Hotelling's T2 statistic and MYT decomposition. We show that unlike existing methods, the computations in the proposed fault detection and diagnosis method is parallelizable and hence scalable.

A virtual machine (VM), during its lifecycle, can be scheduled for execution at geographically disparate cloud locations depending upon the cost of computation and the load at these locations. However, trans-locating a live VM across high-latency low-bandwidth wide area networks (WAN) within 'reasonable' time is nearly impossible due to the large size of the VM image. In this paper, we deal with this problem by combining VM scheduling strategies with VM replication strategies. In particular, we propose to replicate a VM image selectively across different cloud sites, choose a replica of the VM image to be the primary copy and propagate the incremental changes at the primary copy to all the remaining replicas of the VM image. The replica placement strategies are based on factors that influence long-term costs such as the average per-unit cost of storage and the average per-unit cost of computation at different cloud sites besides the 'end-user' latency requirements associated with the VMs. We propose to compensate the additional storage requirements due to replication by exploring commonality that naturally exists amongst different VM images using de-duplication techniques. A key issue that naturally arises in this integrated replication and scheduling context for minimizing migration latencies associated with live migration of VMs across WAN is the design of a good replica placement algorithm that minimizes additional storage requirements. In this paper we address this issue as part for our integrated replication and scheduling architecture, called Cloud Spider. We discuss the trade-offs involved in the design of a replica placement algorithm and propose an algorithm that factors in deduplication ratios amongst pairs of VM images while deciding on the question of replica placement of the VM images. Preliminary experiments show extremely promising results.

Migrating virtual machines (VM) at run time within acceptable time limits is significantly challenging due to large size of the virtual disk files. In this paper, we propose our framework called cloudspider to overcome this challenge by combining replication and scheduling of VM images into a common framework. In particular, we replicate VM images apriori onto a few chosen cloud sites and select one of the replicas of the VM image to be the primary copy. Based on dynamically changing cost parameters at these cloud sites, the scheduler can appropriately select a different replica of a VM image to be the primary copy. The updates at the primary copy are incrementally propagated to the remaining copies. In order to bring down the cost of storage requirement due to the additional replicas our work proposes the use of deduplication techniques. In this paper we address the question of replica placement of VM images when disparate VM images have varying degrees of commonality and latency requirements. The replication algorithm has been implemented and tested using an open-source cloud simulator called CloudSim and the initial results are quite promising.

Reducing the power consumption and operational cost of IT servers is of great concern today. With the growth of the Internet and online services, the number of data centers is increasing day by day. Servers for many cloud applications and other large providers are spread globally. Energy costs across the globe vary dynamically. Servers operate at varied energy costs based on their location and time of use. The load of a server varies based on its geographical location and the time of operation. This paper focuses on exploiting the dynamic nature of electrical power pricing, so that a cost saving is obtained by geo-location of requests to servers operating at lower costs at particular times. There exist patterns of load that are similar for different types of servers. Scheduling decisions are made considering both loads and operating costs of the servers into account, i. e., requests are scheduled to run on servers operating at low cost that also have low expected load. In order to meet the business requirements of an application, scheduling decisions for requests that have stringent SLA considerations or high server affinity, are made by assigning high priority for these requests. Geo-location of requests is done for low priority requests.

Diagnosing cause of system failure in data centers that house large interconnected complex computer systems is a herculean task. This is because different monitoring tools for network, storage, server, facilities and application provide useful information regarding the health of the communication systems, the storage arrays, the physical machines, the environmental factors and the applications within a data center respectively in only a piece-meal manner. The existing tools fail to provide a comprehensive view of the complete set of operations within a data-center. In the absence of integrated monitoring and management tools, a data center administrator has to manually shuffle through and analyze data from various logs generated by the disparate monitoring tools on occurrence of a fault for identifying the root cause. In this paper we propose an approach for integrated data center health monitoring and management framework on top of the existing monitoring tools. The integrated framework leverages complex event processing techniques to process massive streams of events from these tools in (near) real time and enables automatic reuse of the existing monitoring tools in a non-intrusive manner.

Shrisha Rao’s major academic interest is in the area of safety in distributed com-puter systems – what constitutes a safe distributed system, and how can we createdistributed systems that are safe? The motivation behind this is easy to realize; theworld of computing is increasingly distributed, with resources that are geographi-cally spread, often with no central authority or assurances of performance. Tasks andservices that a single person uses are increasingly likely to depend upon invocationsof resources that are distributed. The safety of persons or property may also dependupon the proper functioning of diverse, uncorrelated computing elements. In hisspare time, he volunteers for the American Red Cross – he is a certified instructorin First Aid and CPR, and is also trained in various disaster relief tasks.

Server Clusters have become integral components to operate Internet Services and Scientific computation. With the growing use of cluster systems in web servers, file distribution and database transactions, power conservation and efficiency have been identified as critical issues in the design of server cluster systems. The server cluster electricity costs are of the order of $Billion annually. To reduce power consumption, a master in a server cluster should hibernate the idle servers following the pattern observed from the historical data. This technique saves energy that would have been consumed by the idle servers if not hibernated. This project focuses on finding patterns from historical data, decision logic to send a server to hibernation. The project source can be found at : http://sourceforge.net/projects/energy-awaremod/ © 2010 This material is available under the Creative Commons Attribution-Noncommercial-Share Alike License. See http://creativecommons.org/licenses/by-nc-sa/3.0/ for details

We construct energy optimization policies for a server cluster, using statistical data analyses and demand prediction methodologies, with an aim to reduce the power consumption of the server cluster. In doing so, we monitor and analyze the historical time-series utilization data of a server cluster. Based on the analyses results, we develop predictions about utilization of servers for future time periods. Using this predictive analysis, we formulate energy optimization rules or policies for the server cluster. These policies are then evaluated to determine if they result in energy savings in the server cluster. High-level implementation of this entire mechanism is provided and a strategy for inclusion of this mechanism in existing data center automation products is discussed.

Citation matching is the problem of finding which citation occurs in a given textual corpus. Most existing citation matching work is done on scientific literature. The goal of this paper is to present methods for performing citation matching on Sanskrit texts. Exact matching and approximate matching are the two methods for performing citation matching. The exact matching method checks for exact occurrence of the citation with respect to the textual corpus. Approximate matching is a fuzzy string-matching method which computes a similarity score between an individual line of the textual corpus and the citation. The Smith-Waterman-Gotoh algorithm for local alignment, which is generally used in bioinformatics, is used here for calculating the similarity score. This similarity score is a measure of the closeness between the text and the citation. The exact- and approximate-matching methods are evaluated and compared. The methods presented can be easily applied to corpora in other Indic languages like Kannada, Tamil, etc. The approximate-matching method can in particular be used in the compilation of critical editions and plagiarism detection in a literary work.

The concept of Mandatory Access Controls (MAC) enforces a security policy on users. Linux Security Modules (LSM) provides for development of such frameworks. Lothlorien is an exploratory framework using LSM and the POSIX 1.e draft specifications'1 Extended Attributes (EA) to implement MAC on Linux. Lothlorien makes use of system call hooks provided by LSM to place checks on the access to system resources. The entire system is divided logically into different zones where the resources and users of the systems are distributed depending on their security context similar to a real organisation. Along with the already available Discretionary Access Control (DAC) on Linux, Lothlorien intends to achieve TCSEC B1 level of security. A policy will define the rules of access. Separating all the entities of the system allows fine-grained access control.

This paper presents a formal method based on standard concepts from mathematical logic, for the analysis of system properties, particularly safety. Informally, a safety proposition is a statement, “This system is safe,” and a safety predicate is a statement, “This system is safe under such-and-such conditions.” The standard syntactic rules for predicate logic apply with safety predicates, though their semantics now derive from the context of systems theory and systems engineering. The main insight is that the description of a safety predicate by its quantifier, variable, range, and term can be used as a safety specification for a system on a large scale as well, and that substitutions of expressions for variables in a safety predicate can be used as a tool of refinement.

We present a model and architecture that enhance anomaly based intrusion detection system(IDS) with threat-awareness capability. Anomaly based network IDS, profile network traffic to arrive base-line based on which it identifies anomalous events. However, due to dynamic changes in the threat level of a network, only a subset of these identified events are relevant to the network at any given instance. Hence, we introduce the notion of Threat-Awareness for anomaly based network IDS that periodically learns the changing threats in a network and enhance the capability of traditional anomaly based IDS to obtain network specific useful alarms. In this paper, we present a Threat-Aware Anomaly-Based IDS model for obtaining network-specific useful alarms. We also present our architecture and discuss its internal functions. Finally, we present our experiments based on various threat scenarios and the results obtained proves the efficiency of our model.

We present a model and architecture that enhances the traditional signature based intrusion detection engine with threat-awareness capability. Signature based network intrusion detection systems use a set of signatures S to evaluate captured network traffic for detecting intrusions. However, due to dynamic changes in the threat level of a network, only a subset of signatures s of S is relevant to the network at any given instance. Hence, we introduce a component called a dynamic threat profiler that periodically learns the changing nature of threats in a network. We model and prove the efficacy of the threat-aware signature based intrusion detection approach for obtaining network-specific useful alarms. We also present our architecture and discuss its internal functions. Finally, we present our experiments based on various threat scenarios and the results obtained, which show that network-specific useful alarms formed 95 percent of the alarms that were generated using our model, when compared with the traditional signature-based detection engine where useful alarms form only 30 percent of the generated alarms.

In this paper we discuss our implementation of a virtual Beowulf cluster, and the results of experiments using which we have built a predictive model. Estimating a cluster's performance is difficult, and investing in a real cluster without any idea of the performance that will be delivered is not recommended. We aim to suggest a way to address this problem by using a virtual Beowulf cluster. A virtual Beowulf cluster is a virtualized setup to simulate and understand the working of a real Beowulf cluster, but without extensive investments in hardware. This virtual setup is used to build a predictive model.