Rajib K. Das

Bio: Rajib K. Das is an academic researcher from University of Calcutta. The author has contributed to research in topics: Cloud computing & Computer science. The author has an hindex of 7, co-authored 40 publications receiving 137 citations. Previous affiliations of Rajib K. Das include Indian Statistical Institute & Tezpur University.

Energy and quality of service-aware virtual machine consolidation in a cloud data center

Abstract: The large-scale virtualized Cloud data centers consume huge amount of electrical energy leading to high operational costs and emission of greenhouse gases. Virtual machine (VM) consolidation has been found to be a promising approach to improve resource utilization and reduce energy consumption of the data center. However, aggressive consolidation of VMs tends to increase the number of VM migrations and leads to over-utilization of hosts. This in turn affects the quality of service (QoS) of the applications running in the VMs. Thus, reduction in energy consumption and at the same time ensuring proper QoS to the Cloud users are one of the major challenges among the researchers. In this paper, we have proposed an energy efficient and QoS-aware VM consolidation technique in order to address this problem. We have used Markov chain-based prediction approach to identify the over-utilized and under-utilized hosts in the data center. We have also proposed an efficient VM selection and placement policy based on linear weighted sum approach to migrate the VMs from over-utilized and under-utilized hosts considering both energy and QoS. Extensive simulations using real-world traces and comparison with state-of-art strategies show that our VM consolidation approach substantially reduces energy consumption within a data center while delivering suitable QoS.

Unsupervised Learning of Morphology for Building Lexicon for a Highly Inflectional Language

Abstract: Words play a crucial role in aspects of natural language understanding such as syntactic and semantic processing. Usually, a natural language understanding system either already knows the words that appear in the text, or is able to automatically learn relevant information about a word upon encountering it. Usually, a capable system---human or machine, knows a subset of the entire vocabulary of a language and morphological rules to determine attributes of words not seen before. Developing a knowledge base of legal words and morphological rules is an important task in computational linguistics. In this paper, we describe initial experiments following an approach based on unsupervised learning of morphology from a text corpus, especially developed for this purpose. It is a method for conveniently creating a dictionary and a morphology rule base, and is, especially suitable for highly inflectional languages like Assamese. Assamese is a major Indian language of the Indic branch of the Indo-European family of languages. It is used by around 15 million people.

Acquisition of Morphology of an Indic Language from Text Corpus

Abstract: This article describes an approach to unsupervised learning of morphology from an unannotated corpus for a highly inflectional Indo-European language called Assamese spoken by about 30 million people. Although Assamese is one of Indias national languages, it utterly lacks computational linguistic resources. There exists no prior computational work on this language spoken widely in northeast India. The work presented is pioneering in this respect. In this article, we discuss salient issues in Assamese morphology where the presence of a large number of suffixal determiners, sandhi, samas, and the propensity to use suffix sequences make approximately 50% of the words used in written and spoken text inflected. We implement methods proposed by Gaussier and Goldsmith on acquisition of morphological knowledge, and obtain F-measure performance below 60%. This motivates us to present a method more suitable for handling suffix sequences, enabling us to increase the F-measure performance of morphology acquisition to almost 70%. We describe how we build a morphological dictionary for Assamese from the text corpus. Using the morphological knowledge acquired and the morphological dictionary, we are able to process small chunks of data at a time as well as a large corpus. We achieve approximately 85% precision and recall during the analysis of small chunks of coherent text.

Heuristic-Based Resource Reservation Strategies for Public Cloud

Abstract: Cloud service providers (CSPs) adapt different pricing models for their offered services. Some of the models are suitable for short term requirement while others may be suitable for the cloud service user's (CSU) long term requirement. For example, reservation-based pricing model is appropriate for a CSU's long term demand for resources. Finding the optimal amount of resources to be reserved in advance, to minimize the total cost, needs sufficient research effort. Various algorithms were discussed in the last couple of years to solve the resource reservation problem but most of them are based on integer programming problem (IPP) which is NP in nature. In this paper, we derive some heuristic-based polynomial time algorithms to find some near optimal solution to this problem. We show that the cost for CSU using our approach is comparable to the solution obtained using optimal IPP.

Energy and Makespan Aware Scheduling of Deadline Sensitive Tasks in the Cloud Environment

Abstract: Cloud computing enables the execution of various applications submitted by the users in the virtualized Cloud environment. However, the Cloud infrastructure consumes a significant amount of electrical energy to provide services to its users that have a detrimental effect on the environment. Many of these applications (tasks), like those belonging to the healthcare system, scientific research, the Internet of Things (IoT), and others, are deadline-sensitive. Hence efficient scheduling of tasks is essential to prevent deadline violation, decrease makespan, and at the same time reduce energy consumption. To address this issue, we have considered the bi-objective optimization problem of minimization of energy and makespan and have proposed two scheduling approaches for independent, deadline-sensitive tasks in a heterogeneous Cloud environment. Our first approach is a greedy heuristic based on the Linear Weighted Sum technique. The second one is based on Ant Colony Optimization and uses a combination of heuristic search and positive feedback of information to improve the solution. Both approaches use a three-tier model where tasks are scheduled by taking into account the properties of three entities- tasks, VMs, and hosts. Moreover, we have proposed a suitable strategy for scaling of Cloud resources to improve energy-efficiency and task schedulability. Extensive simulations using Google Cloud trace-logs and comparison with some state-of-art approaches validate the effectiveness of our proposed scheduling techniques in achieving a proper trade-off between the energy consumption of the virtualized Cloud infrastructure and the average makespan of the tasks.

Unsupervised learning of morphology

Abstract: This article surveys work on Unsupervised Learning of Morphology. We define Unsupervised Learning of Morphology as the problem of inducing a description (of some kind, even if only morpheme-segmentation) of how orthographic words are built up given only raw text data of a language. We briefly go through the history and motivation of the this problem. Next, over 200 items of work are listed with a brief characterization, and the most important ideas in the field are critically discussed. We summarize the achievements so far and give pointers for future developments.

The role of artificial intelligence in photo-voltaic systems design and control: A review

Abstract: This paper is a review on the up to date scientific achievements in applying Artificial Intelligence (AI) techniques in Photovoltaic (PV) systems. It surveys the role of AI algorithms in modeling, sizing, control, fault diagnosis and output estimation of PV systems. It also summaries more than 100 research articles in the applications of AI techniques in PV research. A complete comparison between conventional and AI methods is carried out to prove the important role of the AI algorithms play PV systems. The paper compares between the reviewed works and outlines their contributions.

Task Scheduling in Cloud Computing based on Meta-heuristics: Review, Taxonomy, Open Challenges, and Future Trends

Abstract: Cloud computing is a recently looming-evoked paradigm, the aim of which is to provide on-demand, pay-as-you-go, internet-based access to shared computing resources (hardware and software) in a metered, self-service, dynamically scalable fashion. A related hot topic at the moment is task scheduling, which is well known for delivering critical cloud service performance. However, the dilemmas of resources being underutilized (underloaded) and overutilized (overloaded) may arise as a result of improper scheduling, which in turn leads to either wastage of cloud resources or degradation in service performance, respectively. Thus, the idea of incorporating meta-heuristic algorithms into task scheduling emerged in order to efficiently distribute complex and diverse incoming tasks (cloudlets) across available limited resources, within a reasonable time. Meta-heuristic techniques have proven very capable of solving scheduling problems, which is fulfilled herein from a cloud perspective by first providing a brief on traditional and heuristic scheduling methods before diving deeply into the most popular meta-heuristics for cloud task scheduling followed by a detailed systematic review featuring a novel taxonomy of those techniques, along with their advantages and limitations. More specifically, in this study, the basic concepts of cloud task scheduling are addressed smoothly, as well as diverse swarm, evolutionary, physical, emerging, and hybrid meta-heuristic scheduling techniques are categorized as per the nature of the scheduling problem (i.e., single- or multi-objective), the primary objective of scheduling, task-resource mapping scheme, and scheduling constraint. Armed with these methods, some of the most recent relevant literature are surveyed, and insights into the identification of existing challenges are presented, along with a trail to potential solutions. Furthermore, guidelines to future research directions drawn from recently emerging trends are outlined, which should definitely contribute to assisting current researchers and practitioners as well as pave the way for newbies excited about cloud task scheduling to pursue their own glory in the field.

NUMFabric: Fast and Flexible Bandwidth Allocation in Datacenters

Abstract: We present xFabric, a novel datacenter transport design that provides flexible and fast bandwidth allocation control. xFabric is flexible: it enables operators to specify how bandwidth is allocated amongst contending flows to optimize for different service-level objectives such as minimizing flow completion times, weighted allocations, different notions of fairness, etc. xFabric is also very fast, it converges to the specified allocation one-to-two order of magnitudes faster than prior schemes. Underlying xFabric, is a novel distributed algorithm that uses in-network packet scheduling to rapidly solve general network utility maximization problems for bandwidth allocation. We evaluate xFabric using realistic datacenter topologies and highly dynamic workloads and show that it is able to provide flexibility and fast convergence in such stressful environments.