International Journal of Cloud Computing (33 papers in press)
MSA: A Task Scheduling Algorithm for Cloud Computing
by Subhashree Mohapatra, Chhabi Rani Panigrahi, Bibudhendu Pati, Manohar Mishra
Abstract: Cloud computing is an effective technology to perform huge-scale and complex computing. Achieving a minimum makespan is the prime motto of any task scheduling algorithm for a cloud computing environment. This paper aims to propose a new task scheduling algorithm named as Min-sufferage scheduling. While selecting a task for execution, the Min-sufferage picks up the task with minimum sufferage value. The proposed method has been tested on numerous set of tasks and resources. The experimental results indicate that the proposed Min-sufferage algorithm results in less makespan as compared to Sufferage algorithm.
Keywords: Sufferage; Makespan; Cloud Computing.
Flow-based Dynamic Load balancing algorithm for the Cloud networks using Software Defined Networks
by Wilson Prakash, Deepa Lakshmi
Abstract: In recent days, cloud computing has become an outstanding technique in all areas that provide various computing resources. The specialty of cloud computing is that it provides excellent services in monitoring, communication, software platform,and infrastructure. Basically, cloud network limits its services in a specific location on the particular zone. When the targeted area is near, the specific server completes the request generates from the user. Few servers are busy while some servers are idle when it comes to the services. In this case, the provided resources are not used efficiently. To overcome this problem, this paper proposes a concept known as flow-based dynamic load balancing algorithm in cloud networking with Software Defined Networks (SDN). The fundamental principle of this proposed algorithm is distributing client request equally to all the available servers in a specific cloud network. The main aim of load balancing algorithm is optimizing the resource utilization by increasing the throughput, increase flow completion time, decrease the response time and eliminating the overloading of any individual resource. SDN is mainly used in applying various software functionalities into the hardware devices such as routers, switches etc.
Keywords: Load balancing; Data center; SDN; Traffic Engineering.
Medical Knowledge Extraction scheme for Cloudlet Based Healthcare System to avoid Malicious Attacks
by Anjali Chandavale, Anuja Gade
Abstract: Along with growth in technology, it has become the need of time that the response of doctor should be within few seconds along with his 24x7 availability. To perform this task, it is essential to share medical information that contains patients sensitive information. The medical information sharing involves information collection, information storage and sharing of this information. Security protection to this medical information is one of the concerns. As this medical information is finally stored at remote cloud, protection to the whole healthcare scheme against intrusions or malicious attacks is another important concern. Hence the current researchers are focusing on cloudlet based healthcare systems. Along with growth in technology, young generation always choose to examine health associated information and doctors recommendation for any health associated problem on web. In current question answering system, it remains challenging to extract medical knowledge from the clamorous question- answers pair and remove unwanted information. To overcome these challenges, in this paper, we propose Medical Knowledge Extraction scheme for Cloudlet based Healthcare System to avoid Malicious Attacks. In this proposed system, medical information sharing is done in energy efficient fashion using proposed modified Number Theory Research Unit (NTRU) algorithm. The proposed modified Number Theory Research Unit algorithm is used to perform the encryption of users physiological conditions i.e. body information. Collaborative Intrusion Detection System (CIDS) is used to detect and avoid malicious attacks. Medical Knowledge Extraction (MKE) method finds valid remedial triplicates from clamorous Question-Answer (Q-A) pairs and evaluate the reliability along with doctors proficiency using truth discovery method. rn Proposed modified NTRU algorithm gives 20% to 30% better delivery ratio as compared with the existing RSA algorithm. The response time of proposed system is 8 seconds which results in substantially reduction in time and cost for end user.rn
Keywords: Healthcare; Number Theory Research Unit (NTRU); Decision Making System; Proficiency score; reliable.
Dynamic Energy-saving Approaches in Mobile Cloud Computing: Issues, Challenges and Approaches
by RAJALAKSHMI KRISHNAMURTHI, Mukta Goyal
Abstract: The use of mobile devices is ubiquitous in the current information era. However, there is a critical need for the technical integration of mobile computing with cloud computing to enhance the performance of mobile devices. Mobile cloud computing promises several efficient ways of handling various constraints such as energy consumption, data transmission, bandwidth utilization, weak network connectivity, and user mobility. This paper addresses these issues and discusses the benefits of integrating mobile communications with cloud computing. and the main focus is on computation offloading into the cloud as an effective techniquefor overcoming the energy constraints of mobile devices. The characteristics and implications of various computation offloading techniques are explored.
Keywords: Mobile Cloud Computing; Dynamic Energy Saving; Content Offloading.
Secure Cloud Computing using Homomorphic Construction
by Swathi Velugoti, M.P. Vani
Abstract: When customers are transfering their private data to anyrnthird party, then there is much responsibility of both security andrncompliance. There exist hurdles in mainly two aspects - One is inrnthe users or customers point of view, where they want to ensurernthe privacy of its input parameters and results. Another is to cloudrnservers point of view, where cloud entity is worried about feasiblenessrnof encrypted/transformed inputs and operating on them. Thus, usersrnwould not like to submit their private problem data inputs to the cloud.rnencrypting the private data prior to submission to cloud is a usualrnsolution.rnHere in this paper, we have suggested a framework for cloud datarnsecurity and presented the detailed construction of our proposedrnhomomorphic encryption system. Along with that the security andrncorrectness analysis of our proposed construction is given. The schemernsatisfies both Additive and Multiplicative homomorphism properties
Keywords: Homomorphic encryption; Privacy; Security; Cloud computing; Confidential data.
Optimization of Automatic Web Services Composition Using Genetic Algorithm
by Mirsaeid Hosseini Shirvani
Abstract: In recent years, with the expansion of organizations, service-oriented architecture is known as an effective tool for creating applications. Hence, the need to use web services in organizations to reduce costs is felt more than ever. The purpose of web service composition is to determine a proper mix of user requests that cannot be met by a simple web service. In this paper, a genetic-based algorithm is proposed for combining cloud services that ensures multiple clouds work efficiently. The proposed method also provides an overview of the weaknesses of other available methods in terms of computational complexity in automated selection of web services and makes it possible to fulfill the demands of the composition of web services in a more optimal way. It is worth noting that the simulation results show the superiority of the proposed method compared to other methods analyzed in the paper.
Keywords: Web Service, Web Services Composition, Service-Oriented Architecture, Quality of Service.
Keywords: Web Service; Web Services Composition; Service-Oriented Architecture; Quality of Service.
A Secure and efficient multi cloud-based data storage and retrieval using hash-based verifiable secret sharing scheme
by Majid Farhadi, Hamideh Bypour, Seyyed Erfan Asadi
Abstract: As the availability of many smart devices rises, fast and easy access
to data as well as sharing more information is felt. Cloud computing is a
computational approach that shares configurable resources such as network,
servers, storage space, applications and services on the Internet, and allows
the user to access services without the expertise or control of the technology
infrastructure. The confidentiality, integrity, and availability of the data, reducing
computational cost and communication channel between the data owner (user)
and cloud service providers (CSPs) are essential parts of cloud computing. In the
paper, we propose a new scheme to construct a secure cloud data storage based
on the verifiable secret sharing scheme with public verifiability to protect data
integrity. In the new scheme, the validity of secret shares can be publicly verified
without leaking the privacy of secret shares in the verification phase. Moreover,
the verification phase does not depend on any computational assumptions.
Furthermore, the proposed scheme cannot only detect the cheating but also
identify who are the cheaters. It is worth noting that the proposed scheme is more
efficient compared with the other secret sharing-based cloud data storage since
heavy and complex computation is not required.
Keywords: Cloud computing; cloud data storage; verifiable secret sharing
scheme; public verifiability; hash function.
Stream of Traffic Balance in Active Cloud Infrastructure Service Virtual Machines Using Ant Colony
by Ankita Taneja, Hari Singh, Suresh Chand Gupta
Abstract: Cloud load balancing is the manner of distributing computing resources and workloads over a cloud computing infrastructure. It allows an enterprise to manage workloads through appropriate resource allocation in the cloud. Various load balancing techniques in cloud computing are reviewed and the work presented in this paper thoroughly analyzes and compares two well-known algorithms in MATLAB, the Ant Colony Optimization (ACO) Algorithm and Genetic Algorithm (GA). The objective is to produce an optimal solution for cost and execution time through balancing the workload. It is observed through experimental observations that ACO based load balancing possess incurs low cost and low execution time as compared to the GA for a constant workload over a fixed number of cloud machines. However, the execution time follows a different trend when workload increases and more machines are utilized to handle the increased workload; it rises sharply in ACO as compared to the GA.
Keywords: ACO; ant colony optimization; GA; genetic algorithm; load balancing; cloud computing; pheromone matrix; pheromone table; IAAS; infrastructure as a service.
Memory constraint Parallelized resource allocation and optimal scheduling using Oppositional GWO for handling big data in cloud environment
by Chetana Tukkoji, Seetharam Keshav Rao
Abstract: In cloud computing, task scheduling is one of the challenging troubles, especially when deadline and cost are conceived. On the other hand, the key issue of task scheduling is to reach optimal allocation of users tasks for to optimize the task scheduling performance and reduce non-reasonable task allocation in clouds. Besides, in terms of memory space and time complexities, the processing of huge number of tasks with sequential algorithm results in greater computational cost. Therefore, we have improved an efficient Memory constraint Parallelized resource allocation and optimal scheduling method applying Oppositional GWO for resolving the scheduling trouble of big data in cloud environment by this paper. In parallel over distributed systems, the suggested scheduling approach applies the MapReduce framework to perform scheduling. The Map Reduce framework is consisted of two main processes; particularly, the task prioritization stage (with Fuzzy C-means Clustering method based on memory constraint) in Map phase and optimal scheduling (using Oppositional Grey Wolf Optimization algorithm) in reduce phase. Here, the scheduling is maximized to reduce the makespan, cost and to raise the system utilization.
Keywords: Oppositional Grey Wolf Optimization algorithm; Fuzzy C-means
Clustering; MapReduce; Task Prioritization; Virtual Machine Allocation; Apache Spark Distributed file System (SDFS).
An efficient load balancing scheduling strategy for cloud computing based on hybrid approach
by Mohammad Oqail Ahmad, Rafiqul Zaman Khan
Abstract: Cloud computing is a promising paradigm that is widely used in both academia and industry. Dynamic demand for resources by users is one of the prime goals of scheduling process of task in cloud computing. Task scheduling is NP-hard problem which is responsible for allocating the task to VMs and maximizing their utilization while minimizing the total task execution time. In this paper, the authors propose a load balancing scheduling strategy, Hybridization of RALB method using the PSO technique inspired by the honeybee behaviour proposed named as (PSO-RALB). This strategy optimize the results and perform scheduling based on resource aware load balancing scheme. The foraging behaviour of the honey bee optimization algorithm is utilized to balance load across VM and resource aware is used to manage the resources. The computational results show that proposed scheme minimize the makespan time, total processing time, total processing cost and the degree of imbalance factor when compared with existing techniques PSO standard and PSO based Load Balancing (PSO-LB) algorithms.
Keywords: Cloud computing; Load balancing; Honey bee foraging; Particle Swarm Optimization; PSO-RALB Algorithm; Degree of imbalance;.
End-to-End SLA Management in Federated Clouds
by Asma Al Falasi, Mohamed Adel Serhani, Younes Hamdouch
Abstract: Cloud services have always promised to be available, flexible, and speedy. However, in some circumstances (e.g. drastic changes in application requirements) a Cloud provider might fail to deliver such promises to their distinctly demanding customers. Cloud providers have a constrained geographical presence and are willing to invest in infrastructure only when it is profitable to them. Cloud federation is a concept that collectively combines segregated Cloud services to create an extended pool of resources for Clouds to competently deliver their promised level of services. This paper is concerned with studying the governing aspects related to the federation of Clouds through collaborative networking. We propose a network of federated Clouds, CloudLend, that creates a platform for Cloud providers to collaborate, and for customers to expand their service selections. We also define and specify a service level agreement (SLA) management model in order to govern and administer the relationships established between different Cloud services in CloudLend. We define a multi-level SLA specification model to describe QoS terms, in addition to a game theory-based automated SLA negotiation model. We also define an adaptive agent-based SLA monitoring model. Formal verification proved that our proposed framework assures customers with maximum optimized guarantees to their QoS requirements, in addition to supporting Cloud providers to make informed resource utilization decisions. Additionally, simulation results demonstrate the effectiveness of our SLA management model.
Keywords: Cloud Computing; Federated Clouds; SLA Management; Game Theory; QoS Requirements.
A Cloud Data Collection Platform for Canine Behavioral Prediction using Objective Sensor Data
by Zachary Cleghern, Marc Foster, Sean Mealin, Evan Williams, Timothy Holder, Alper Bozkurt, David Roberts
Abstract: Training successful guide dogs is time and resource intensive, requiring copious professional and volunteer labor. Even among the best programs, dogs are released with attrition rates commonly at 50\%. Increasing success rates enables non-profits to meet growing demand for dogs and optimize resources. Selecting dogs for training is a crucial task; guide dog schools can benefit from both better selection accuracy and earlier prediction. We present a system aimed at improving analysis and selection of which dogs merit investment of resources using custom sensing hardware and a cloud-hosted data processing platform. To improve behavioral analysis at early stages, we present results using objective data acquired in puppy behavioral tests and the current status of an IoT-enabled ``Smart Collar'' system to gather data from puppies while being raised by volunteers prior to training. Our goal is to identify both puppies at risk and environmental influences on success as guide dogs.
Keywords: Cloud Computing; Canine Behavior; Behavioral Prediction; Sensor Data; Internet-of-Things; Machine Learning; Wearable Computing; Guide Dogs.
Special Issue on: ICACB18 Advanced Computing and Communication Systems
Multi-Objective Optimization Techniques for Virtual Machine Migration-based Load Balancing in Cloud Data Center
by Meenakshi Priya, R.Kanniga Devi
Abstract: This paper aims to balance the load in Cloud Data Center (CDC) by migrating Virtual Machines (VM) across hosts using Multi-Objective Optimization techniques. The unpredictable rate of demand for the cloud services leads to load fluctuation and subsequently load imbalance in Cloud Data Center. Hence, to balance the load in Cloud Data Center, this work presents Multi-Objective Optimization Technique-based Load Balancing (MOOT-LB) method. This work proposes two Multi-Objective Optimization techniques namely, Multi-Objective Particle Swarm Optimization (MOPSO) and Multi-Objective Differential Evolution (MODE) for load balancing the Cloud Data Center. These techniques identify an optimal set of hosts and set of VMs to be migrated from the source hosts and identify the target hosts for migration in an efficient way. The objectives are to minimize the frequency of VM migration and migration time. To evaluate the performance of the proposed techniques ClouSim 3.0.3 simulator is used. The performance of the proposed techniques is compared, and the results show that MOPSO based load balancing technique achieves better performance than MODE based load balancing technique.
Keywords: VM Migration; Load Balancing; Multi-Objective Optimization; MOPSO; MODE.
Optimized Handoff Mechanism using RFID tags for a Communication Based Train Control System
by Moahanasundaram Ranganathan, Kathirvel Brindhadevi, Arushi Sana, Ankit Malhotra
Abstract: In a Communication Based Train Control System, one of the most important aspects is ensuring that a Mobile Station is connected continuously to the Zone Controller to ensure continuous and precise transmission of the trains location, speed and other relevant information. Due to the continuous motion of the train handoffs between different APs is necessary to remain connected. This work aims to propose a location based handoff method which will minimize handoff latency and provide better connections. The location based handoff mechanism uses the concept of placing RFID tags at predetermined locations. Handoff is triggered at the location where the RFID tag is present and a connection is made to the new AP based on the channel information obtained. The architecture used in the scenario is a WiMAX architecture and connections are maintained with the help of routers contained within the mobile stations. The proposed handoff method provides better handoff latency and error free transmission in a system where precise information transmission is necessary.
Keywords: Communication Based Train Control System; WIMAX; Access Point; RFID; Mobile Station.
A survey: Comparative study on IoT and CoT (Internet of Things and Cloud of Things)
by Kathirvel Brindhadevi, Moahanasundaram Ranganathan, Navin Kumar, Rishikesh Y. Mule
Abstract: In todays era, an exponential rise in technology is changing our lives essay. Tens of millions of devices are getting connected to the internet every day. This extensive growth of the internet and increasing number of interconnected devices has given a rise to many new-age technologies. The Internet of Things (IoT) provides an innovative means of communication amongst these new-age technologies and the Web world. Cloud Computing enables a convenient, on demand and scalable network access to a shared pool of configurable computing resources. Integrating Cloud computing with Internet of Things brings the concept of Cloud of Things. The Cloud of Things framework is based on a combination of ubiquitous distributed sensing units, outcomes stored in the cloud for awareness. rnThis paper surveys a comparative study on internet of things and cloud of things demonstrating how cloud of things has prevalent benefits over internet of things. Also, this paper explains how introduction of fog computing concept in cloud and IoT improves efficiency of Cloud of Things.
Keywords: Cloud of Things; Internet of Things; Cloud computing; Fog Computing.
Enhanced Classification of LISS-III satellite image Using Rough Set theory and ANN
by Anand Upadhyay
Abstract: Land use and land cover classification are one of the major aspects to detect land coverage in particular area. Same goes for water, forest, and mangroves. So by keeping these parameters in mind, our objective is to identify water, land, forest, and mangroves from a LISS-III satellite image by using rough set theory and artificial neural network. LISS-III is multi-spectral camera operating in four different bands. There are many problems related to the classification of the satellite image i.e. universal classifiers, parameter setting of classifiers and features. The classification accuracy is one of the major issues related to classification of satellite image therefor in this paper rough set based artificial neural network is used for classification of the satellite image. The rough set theory is used to reduce the number of the feature vector for improved classification of satellite image using the artificial neural network.
Keywords: LISS-III (Linear Imaging and Self Scanning Sensor); classification; satellite image; accuracy etc.
Analysis of Methane (CH4) and Nitrous Oxide (N2O) emission from paddy rice using IoT and Fuzzy logic
by Shriya A. Jadhav, Anisha Lal
Abstract: Most climate scientists accepted main cause of the "greenhouse effect" is human expansion. The study of rice paddy feilds revels the fact that they are the substantial sources of greenhouse gases such as methane (CH4) and nitrous oxide (N2O). So rapid increase in rice production will result in speedy rise in the level of emission of these gases. Therefore, the purpose of this research is to forecast the level of emission of methane (CH4) and nitrous oxide (N2O) from the paddy farm. Internet of Things can provide an integrated solution for data acquisition, monitoring and measuring methane (CH4) and nitrous oxide (N2O) concentration in air. Here we propose to define a fuzzy rule set considering the various environmental factors and conditions which are causing Greenhouse Gas emission, the fuzzy rule based system can provide a solution using linguistic variables based on which a decision support system can be designed. The combination of IoT and fuzzy logic can be used in the development of intelligent system for pattern recognition, event prediction and decision making.
Keywords: methane (CH4); nitrous oxide (N2O); Internet of things; IoT; Fuzzy ; rice paddy.
A Case study for an Incremental Classifier model in big data
by Blessy Trencia Lincy S S, Suresh Kumar Nagarajan
Abstract: Big data is a term that implies enormous voluminous of data which cannot be handled by the existing traditional systems. With the evolving standards and technologies this volume has reached to a rate, such that even if provided with the huge amount of data it is a challenging task to obtain useful insights or knowledge out of it. Thus, this is a foremost and most important challenge for the researchers and scientists to transform the data or manipulate the data for analysis and processing them with the significant purpose of gaining insights out of it. Prediction plays a vital role in various applications involving business decision making processes or in healthcare industries, and many other domains. This helps in determining the future events, or to understand and analyse the events to predict the future outcome. This in turn increases the performance of the system in terms of accuracy, reduction in cost, speed and many other aspects. In this paper, a incremental classifier model is applied for performing the classification with the evolving new instances of data and analysed as a case study. The experiment is carried out with the healthcare datasets to understand and analyse the suggested model and the proposed model is said to provide better performance that deals with large data.
Keywords: Big data; Incremental model; Classification; Predictive analytics.
A HYBRID ENCRYPTION METHOD HANDLING BIG DATA VULNERABILITIES
by Priyanka G, Anisha Lal
Abstract: As Big Data hits the maximum number of companies in all domains, secured data transfer can be done by cryptography. With increasing threats to Big Data, the security must focus on to avoid the attackers from the formation of any pattern to gain access to the information. Big data deals with the linguistic data which consists of low secured data and high secured data as well. Hence, the system should focus on providing multi-fold security and should avoid high-security common to all data categories. This paper presents a hybrid model for Big Data that ensures Data Confidentiality, Data Integrity, Access Control and Sequential Freshness by combining three symmetric key algorithms AES, DES and Blowfish for the encryption and decryption process in any desired order. Based on the level of security the combination of the algorithms can vary. This method of encryption and decryption process ensures safe data transformation between source and destination.
Keywords: Hybrid; Encryption;Big Data.
Special Issue on: IEEE SERVICES 2018 Edges, Fogs and Clouds as Engines of IoT
SMIoT: A Software Architecture for Maintainable Internet-of-Things Applications
by Ilse Bohé, Michiel Willocx, Vincent Naessens
Abstract: Developing sustainable IoT applications is by far no sinecure. Many IoT apps that are currently on the market can only be coupled with a fixed set of edge devices or technologies. On the contrary, IoT sensors and actuators evolve at a fast pace degrading the attractiveness of many IoT applications over time. The vendor lock-in trap is often triggered by sensor-centric application development. This paper presents application-centric development as an alternative approach to tackle maintainability problems in IoT ecosystems. The paradigm shift is supported by a layered architecture called SMIoT, which guides the design of advanced IoT ecosystems. Applying the architectural principles result in IoT apps that can easily cope with new technologies that come to the market, hence, increasing their lifetime and offering various infrastructural alternatives to end-users. Furthermore, the architectural principles are adopted in an Android framework implementation and validated through the design of a health care ecosystem.
Keywords: Internet-of-Things; Software architecture; Application development; Maintainable;Sustainable; Application-centric; Framework; Android; Edge devices; Tackling Vendor lock-in;.
Edge-centric Resource Allocation for Heterogenous IoT Applications using a CoAP-based Broker
by Simone Bolettieri, Raffaele Bruno
Abstract: The Edge/Fog computing paradigm has been recently advocated for future IoT systems to cope with the capacity and latency constraints of conventional cloud-centric IoT architectures. Fog nodes are not only needed to offload computing tasks from the centralised cloud but also to provide IoT applications with management services that facilitate deployment and improve Quality of Service. Indeed, in large-scale IoT deployments, it is expected that a large number of applications access the same resources (e.g. a sensor or an actuator), most likely hosted on constrained devices. Moreover, applications can have highly heterogeneous QoS requirements, e.g., regarding real-time constraints or frequency with which they desire to receive notifications from the monitored resources. However, IoT applications may be unable to autonomously adapt their access patterns for IoT resources to network dynamics and bandwidth limitations. To address these issues, in this work we design a fog-based broker that regulates the access to IoT resources transparently and effectively. Specifically, we develop an optimisation framework to determine the notification periods that maximise the applications' QoS satisfaction under network-related constraints. Then, we also propose practical algorithms that leverage measurements of the degree of reliability of application transmissions to infer the congestion level of the IoT resources, and adapt the notification periods accordingly. We have developed a software prototype of our broker by exploiting the standard features of the CoAP protocol. Then, we have validated the proposed solution through simulations and real experiments in an IoT testbed. Results show that, as the application demands increase, the proposed approach guarantees better QoS satisfaction, higher throughput and improved energy efficiency than a conventional CoAP proxy. Moreover, the efficacy of the optimal solution heavily depends on accurate estimates of the network capacity, which may be difficult to obtain in real-world IoT deployments.
Keywords: Internet of Things; fog computing; resource brokering; CoAP; emulation; prototype.
The Edge Architecture for Semi-Autonomous Industrial Robotic Inspection Systems
by Ching-Ling Huang
Abstract: Robots have been increasingly used in industrial applications, being deployed along other robots and human supervisors in the automation of complex tasks such as the inspection, monitoring and maintenance of industrial assets. In this paper, we shared our experience and presented our implemented software framework for such Edge computing for semi-autonomous robotics inspection. These systems combine human-in-the-loop, semi-autonomous robots, Edge computing and Cloud services to achieve the automation of complex industrial tasks. This paper describes a robotic platform developed, discussing the key architectural aspects of a semi-autonomous robotics system employed in two industrial inspection case studies: remote methane detection in oilfields, and flare stack inspections in oil and gas production environment. We outline the requirements for the system, sharing the experience of our design and implementation trade-offs. In particular, the synergy among the semi-autonomous robots, human supervisors, model-based edge controls, and the cloud services is designed to achieve the responsive onsite monitoring and to cope with the limited connectivity, bandwidth and processing constraints in typical industrial setting.
Keywords: Semi-autonomous robotics; remote methane leak inspection; Unmanned Aerial Vehicle (UAV); HMI (Human Machine Interface).
In-network Processing for Edge Computing with InLocus
by Lucas Brasilino
Abstract: As sensors and smart device infrastructure grows, networks are\r\nincreasingly heterogeneous and diverse. We propose an efficient and low-latency\r\narchitecture called InLocus, which facilitates stream processing at the networks\r\nedge. InLocus balances hardware-accelerated performance with the flexibility of\r\nasynchronous software control.\r\nIn this paper we extented InLocus architecture by implementing compute\r\nnodes in a a more traditional cloud-based solution in the form of Apache Kafka\r\nand Twitter Heron framework, as well as by introducing a new runtime approach\r\nfor the previously handwritten C Server. We utilize a flexible platform (Xilinx\r\nZynq SoC) to compare microbenchmarks between the latter and High-Level\r\nSynthesis (HLS) version in programmable hardware.
Keywords: In-network Processing; Edge Computing; Internet of Things,\r\nProgrammable Logic; FPGA; Offloading; Hardware Acceleration.
Special Issue on: ICBDSDE'19 Cloud Computing for Smart Digital Environment
An Improved Pricing Algorithm for Infrastructure as a Service Clouds
by Seyyed-Mohammad Javadi-Moghaddam, Asieh Andarzgoo, Mohsen Saberi
Abstract: Marketing in cloud systems enables users to trade and share resources. For the sales of services, Client applications and service providers negotiate to make a Service Level Agreement. Offering prices in the negotiation of services become one that is challenging. A federal cloud is an efficient approach of recent interest to better balance risk sharing between services provider and customer. This work presents a new algorithm to increase service provider revenue and reducing user costs simultaneously. The auction of remaining time spent on resources and interactions between federal clouds increases the profits of clouds, the number of successful requests, and reduces users' costs. The simulation results confirm the expectations of the proposed approach.
Keywords: Federal cloud; Pricing model; Service quality; Service level agreement.
Versioning Schemas of JSON-based Conventional and Temporal Big Data through High-level Operations in the ?JSchema Framework
by Zouhaier Brahmia, Safa Brahmia, Fabio Grandi, Rafik Bouaziz
Abstract: ?JSchema is a framework for managing time-varying JSON-based Big Data, in temporal JSON NoSQL databases, through the use of a temporal JSON schema. This latter ties together a conventional JSON schema, which is a standard JSON Schema file, and its corresponding temporal logical and temporal physical characteristics, which are stored in a temporal characteristic document. Conventional JSON schema and temporal characteristics could evolve over time to satisfy new requirements of the NoSQL database administrator (NSDBA) or to comply with changes in the modelled reality. Accordingly, the corresponding temporal JSON schema is also evolving over time. In our previous work (Brahmia et al., 2017, 2018b, 2019a), we have proposed low-level operations for changing such schema components. However, these operations are not NSDBA-friendly as they are too primitive. In this paper, we deal with operations that help NSDBAs to maintain these schema components, in a more user-friendly and compact way. In fact, we propose three sets of high-level operations for changing the temporal JSON schema, the conventional JSON schema, and the temporal characteristics. These high-level operations are based on our previously proposed low-level operations. They are also consistency-preserving and more helpful than the low-level ones. To improve the readability of their definitions, we have divided these new operations into two classes: basic high-level operations, which cannot be defined through other basic high-level operations, and complex ones.
Keywords: Big Data; NoSQL; JSON; JSON Schema; ?JSchema; Conventional JSON schema; Temporal JSON schema; Temporal logical characteristic; Temporal physical characteristic; Schema change operation; Schema versioning; temporal databases.
Special Issue on: Big Data Computing and Sustainable Cloud Communication Systems
Green Factors of Referential Value based Software Component Repository
by Pradeep Kumar, Shailendra Narayan Singh, Sudhir Dawra
Abstract: Reference based green computing has different dimensions like time, cost, space etc. Time and space required by software system is dependent upon development process of the software and algorithm used during software development. Green parameters of the software are directly proportional to the space and time required to process the softwares. In the referential software development system, new dimensions are coming into picture such as length of networks used, technology adopted in the network systems and response time of the middleware. In this paper, all green parameters including new dimensions proposed in the referential software development process have been calculated. The proposed technique presents the mapping of addresses required to simulate referential software‟s in the cloud computing scenario
Keywords: Referential software; Cloud computing; Green computing; Computer networks.
A Novel Approach for Merging Ontologies using Formal Concept Analysis
by Priya. Munusamy, Ch. Aswani Kumar
Abstract: : Ontologies are mainly used for knowledge sharing and also as a knowledge structure. Due to the rising nature of ontologies, the method of merging information in the corporate realm turns to be critical. In the existing methods, formal concept analysis does not provide an efficient pseudo-intent calculation and does not handle large context. The proposed technique focused the issue of ontology heterogeneity that blocks the ontology interoperability and proposed a novel algorithm called Advanced Formal Concept Analysis merge. The AFCA-Merge algorithm performs four phases to merge the given two ontologies. In the first phase, it obtains the perfect attribute for the matching object using decision tree and pseudo intent technique. In the second phase, the obtained results are warehoused in the linked list as a formal context. In the third phase, the perfect relationship among formal contexts from the linked list has been identified using backtracking techniques. Finally, the merging phase performs the merging between the identified relations. The experimental outcome shows that the AFCA-Merge provides 97% of precision, 82 % of recall and 89 % of accuracy which is better than the existing technique.
Keywords: AFCA-Merge; Formal Concept Analysis; Formal Context; Ontology Merging.
Spectral and Spatial Features Based HSI Classification Using Multiple Neuron Based Learning Approach
by Venkatesan Rudhrakoti, Prabu Sevugan
Abstract: With the improvement of remote sensing application, hyper spectral images have been used in large number of applications. And lot of works have been done to extract the features from remote sensing and accurate learning for classify the classes. The spectral and spatial data of images have been allows to classify the results with improved accuracy. Fusion of spatial and spectral data is an actual way in improv-ing the accuracy of hyper-spectral image classification. In this work, we proposed spectral with spatial details based on hyper-spectral image classification method using neural network classifiers and using multi neurons based learning approach is used to classify the remote sensing images with specific class labels. The features may be supernatural and latitudinal data is extracted using boundary values using decision boundary feature extraction (DBFE). These extracted features are trained using convolutional neural networks (CNN) for improve the ac-curacy for labeling the classes. The methodology entails of training with adding regularizer towards the loss function recycled for train the neural networks. Guidance is done using various layers with additional balancing constraints to evade falling into local minima. In testing phase, classify each remote sensing image and avoid false truth map. Experimental results shows that improved accuracy in class specifica-tion rather than other state of art algorithms.
Keywords: Hyper spectral imaging; Classification; Features extraction; Neural networks; Class labels.
Correlative Study and Analysis for Hidden Patterns in Text Analytics Unstructured Data using Supervised and Unsupervised Learning techniques
by E.Laxmi Lydia, S. Kannan, S.Suman Rajest
Abstract: Two-third of the data generated by the internet is unstructured text in thernform of Emails, audio, video, pdf files, word documents, text documents. Extraction ofrnthese unstructured text patterns using mining techniques achieve quick access tornoutcomes. Textual data available atonline contains different patterns and when thosernhuge incoming unstructured data enters into the system creates a problem whilernorganizing those documents into meaningful groups. This paper discusses documentrnclassification using supervised learning by focusing on the concept based algorithm andrnalso deals with the hidden patterns in the documents using unsupervised clusteringrntechnique and Topic-based Modeling for the analysis and improvement of systematicrnarrangement of documents by applying k-means and LDA algorithm. Finally, thisrnpresents the comparative study and importance of clustering than classification forrnunstructured documents.
Keywords: Text Analytics; Concept Based method; Data Representation and Storage,rnLatent Dirichlet Allocation(LDA)Algorithm.
Feature vector extraction and optimization for multimodal biometrics employing face, ear and gait utilizing artificial neural networks
by Haider Mehraj, Ajaz Hussain Mir
Abstract: Cloud Computing is the rapidly growing model for providing resources to users over internet. Numerous business establishments have adopted cloud computing environment as it has low upfront costs, is scalable and provides rapid deployment. Many consumers store their sensitive information on the cloud and as such there needs to be strong authentication mechanism in place so that only authorised users are able to access the cloud. Multimodal biometrics is an upcoming research area to explore for improving the security of cloud. In this work, a novel multimodal biometric fusion system using three different biometric modalities including face, ear, and gait, based on Speed-Up-Robust-Feature (SURF) descriptor along with Genetic Algorithm (GA) for enhanced cloud security is anticipated. Artificial neural network (ANN) is utilized as a classifier for each biometric modality. Our novel fusion process has been effectively tested by means of dissimilar images analogous to all subjects from three databases including AMI Ear Database, Georgia Tech Face Database along with the CASIA Gait Database. Because of these biometric traits, the anticipated method requires no significant user assistance and also can work from a long distance. Before going for the fusion, the SURF features are optimized using genetic algorithm and cross validated using artificial neural network. The evaluations are done on a publicly available database demonstrating the accuracy of the proposed system compared with the existing systems. It is observed that, the amalgamation of face, ear and gait gives better performance in terms of accuracy, precision and recall and fmeasure.
Keywords: Cloud Computing; Biometric Fusion; Feature Vector; SURF; GA; ANN; precision; recall; kappa; accuracy; Fmeasure.
Special Issue on: Impact of Machine Learning in the Cloud Computing Revolution
Using Augmented Reality to Support Children With Dyslexia
by Majed Aborokbah
Abstract: This paper presents the use of interactive improved reality interface to assist and support children with dyslexia and it is one of the most common learning disabilities in the world. This is a literacy-based learning difficulty that mainly effect in reading, writing, speaking, short-term memory, spelling and etc. Many more people perhaps as many as 1520% of the population as a whole have some of the symptoms of dyslexia. This paper introduces case studies with different learning scenarios of Arabic language which have been designed based on Human Computer Interaction (HCI) principles so that meaningful virtual information is presented for dyslexic children in an interactive and compelling way. The smart phones are considered as being potentially valuable learning tools, this due to their portability, accessibility and pervasiveness. The blending of Technology and education is something that is growing rapidly and becomes most popular. Augmented Reality (AR) is recent example of a technology that has been combined into the educational field. This work aims to integrate mobile technology and AR method to improve the dyslexic children (DC) academic performance, concentration and short-term memory. The design process includes the following steps of identify the research problem and determines the requirements to overcome dyslexia problems, collect carefully the data from different sources and the collected data will be used to construct the target product based on the prototype methodology. As the output come, it will contribute in improving the learning and basic skills of children with dyslexia.
Keywords: learning disabilities; learning tools; augmented reality;.
MOBILITY OF SINK BASED DATA COLLECTION PROTOCOL (MSDCP) FOR ENERGY BALANCING IN WSN
by LALITHA THAMBIDURAI, SaravanaKumar R
Abstract: A sensor node is that the significant part of a wireless sensor network. Sensor nodes have various roles in a network includes identifying data storage, data processing and routing method. Cluster is an organizational element for wireless sensor networks. The powerful environment of this network is very essential for them to be broken down into clusters to make easier responsibilities such as communication. Cluster heads are the group head of a cluster have greater data rate match the alternative cluster member. rnThey frequently needed to associate activities within the cluster. These methods comprise but are not controlled to data aggregation and forming account of a cluster .Base station is at the upper level of organized wireless sensor network. It generates communication link among the sensor network and the end user.The data in a sensor network can be used for an enormous variety of applications. A detailed application is form use of network data over the internet retaining a personal digital assistant or desktop computer.rn This paper contributions mobility based reactive protocol named Mobility of Sink based Data Collection Protocol (MSDCP).This protocol sensor with great energy and maximum quantity information are picked as cluster heads that gather data from the common nodes between the clusters. This data is placed unless mobile sink comes within the transmission area of cluster heads and request for the gathered data. One time the request is received from cluster head and it forward data to the mobile sink.rn
Keywords: WSN; MSDCP; Transmission Area; Cluster Head;.
Fuzzy-C means Segmentation of Lymphocytes for the Identification of the Differential Counting of WBC
by Duraiswamy Umamaheswari
Abstract: In the domain of histology, discovering the population of White Blood Cells (WBC) in blood smears helps to recognize the destructive diseases. Standard tests performed in hematopathological laboratories by human experts on the blood samples of precarious cases such as leukemia are time-consuming processes, less accurate and it totally depends upon the expertise of the technicians. In order to get the advantage of faster analysis time and perfect partitioning at clumps, an algorithm is proposed in this paper that automatically identifies the counting of lymphocytes present in peripheral blood smear images containing Acute Lymphoblastic Leukemia (ALL). That performs lymphocytes segmentation by Fuzzy C-Means clustering (FCM). Afterward, neighboring and touching cells in cell clumps are individuated by the Watershed Transform (WT), and then morphological operators are applied to bring out the cells into an appropriate format in accordance with feature extraction. The extracted features are thresholded to eliminate the regions other than lymphocytes. The algorithm ensures 98.52% of accuracy in counting lymphocytes by examining 80 blood smear image samples of the ALL-IDB1 dataset.
Keywords: Fuzzy c-means; medical image processing; morphology; segmentation; watershed; WBC count; leukemia.