Elastic models for cloud in e learning applications. V.P.Sampath 1, V.P.Sandhya 2

Loading....

Loading....

Loading....

Loading....

Loading....

Full text

(1)

Elastic models for cloud in e learning applications. V.P.Sampath1, V.P.Sandhya2

1

EEE dept,SMKFIT,OMR,Thaiyur village,Chennai,Tamilnadu,India.

2

EIE dept, Valliammai Engineering college, Chennai, Tamilnadu,India. . ramsampath78@rediffmail.com¹ sandh.vp@gmail.com²

Abstract

There are a number of cloud-based applications available in the e-learning sector as well. Smart cloud computing enables cloud servers to provide smart learning services to users through additional intelligent processes on existing cloud systems. As results, it can provide customized contents to each user. We propose a smart cloud robotic elastic model with rescheduling for elearning. The cloud robotics architecture leverages the combination of a virtual ad-hoc cloud

formed by machine-to-machine (M2M)

communications among participating

robots, and an infrastructure cloud

enabled by machine-to-cloud (M2C)

communications. Cloud robotics utilizes elastic computing models, in which resources are dynamically allocated from a shared resource pool in the cloud, to support task offloading and information sharing in robotic applications.

Keywords:M2M,M2C,SaaS,IaaS,PaaS,clo ud computing; context-awareness; smart learning service

1.Introduction

The cloud computing environment provides the necessary foundation for the integration of platform and technology.It integrates teaching and research resources distributed over various locations by utilizing existing conditions as much as possible to meet the demands of the teaching and research activities.Robotic systems have brought significant economic

and social impacts to human lives over the past few decades .Industrial robots have been widely deployed in factories to do tedious, repetitive, or dangerous tasks. These preprogrammed robots have been very successful in industrial applications due to their high endurance, speed, and precision in structured factory environments. To enlarge the functional range of these robots or to deploy them in unstructured environments, robotic technologies are integrated with network technologies to foster the emergence of networked robotics.

Networked robotics, similar to standalone robots, faces inherent physical constraints as all computations are accomplished in the robotic network, and information access is restricted to the collective storage of the network. With the rapid advancement of wireless communications and recent innovations in cloud computing technologies, some of these constraints can be overcome through the concept of cloud robotics, leading to more intelligent, efficient and yet cheaper robotic networks. This paper is organized as follows: the related works and Section 2.1 deals with the motivation and Section III is dedicated to the. Cloud Computing and Cloud-Based Applications.Section IV deals with the proposed idea ie SLA Enforcement and Rescheduling. We conclude and summarize this paper in Section V.

(2)

As far as the study [1],which allows the deployment of inexpensive robots with low computation power and memory requirements by leveraging on the communications network and the elastic computing resources offered by the cloud infrastructure.The studies[2] where the platform advancements enable context awareness in a smart cloud computing environment and smart services for innovative learning processes. Wireless technologies have changed the way learners access and share resources, acquire knowledge, and collaborate with each other. Such technologies may include various mobile devices such as hand-held computers and smart phones, embedded sensors in those devices,high-speed wireless networking technologies such as 4G networks that allow those heterogeneous. The platform advancements enable context awareness in a smart cloud computing environment and smart services for innovative learning processes. Recent studies[3] introduce cloud computing to learning, build an e-learning cloud, and make an active research and exploration.The literature[4] motivation is to present the benefits of applying SOA principles in the design of an infrastructure to support a robot undertaking more complex tasks as in which provides insight how semantic web and web services can be applied on robotics in order to facilitate cooperation between robots for joint tasks execution.

2.1Motivation

E-learning cloud is a migration of cloud computing technology in the field of e-learning,which is a future e-learning infrastructure, including all the necessary hardware and software computing resources engaging in elearning.After these computing resources are virtualized, they can be afforded in the form of services for educational institutions, students and businesses to rent computing resources. The key components involved in the process of scheduling an application

on a VM are admission control, VM manager, job scheduler and SLA manager. The admission control component decides whether the requested VM (for an application) can be allocated and the QoS requirements can be met with a given number of available resources. If an application is accepted for execution, SLA is signed and penalty rates are negotiated with the user. The VM manager will initiate a VM and allocate it to a server having the required capacity.

3. Cloud Computing and Cloud-Based Applications

A cloud computing project that uses elastic based model for smart learning and an overview of context-awareness in a learning environment. The cloud computing environment with respect to s-learning offers new ideas and solutions in achieving interoperability among heterogeneous resources and systems. The cloud services mean that the Internet can be used as huge workspace, repository, platform, and infrastructure. Learners can access to the Internet from anywhere at anytime, using widely spread mobile devices but the existing cloud computing technologies are only passively responsive to users’ needs. This situation necessitates proactive cloud services rather than passive services. Since learners typically carry mobile devices of some kind at their hands, the volume of information and services processed through the devices continues to increase. One important offering of cloud computing is to deliver computing Infrastructure-as-a-Service (IaaS). In this type of cloud, raw hardware infrastructure, such as CPU, memory and storage, is provided to users as an on-demand virtual server.Aside from client-side reduced total cost of ownership due to a usage-based payment scheme, a key benefit of IaaS for cloud providers is the increased resource utilization in data centers. Due to the high flexibility in adjusting virtual machine capacity, cloud providers can consolidate traditional web

(3)

applications into a fewer number of physical servers given the f

peak loads of

individual applications have few with each other .

3.1.Cloud computing elastic models

The most important responsibility of the servers is to perform elastic processes such as Elastic Computing for Infrastructure as a Service (IaaS), Elastic Management for Platform as a Service (PaaS) and Elastic Deployment for Software as a Service (SaaS) constantly. The elastic processing can be described as collecting user information that is pulled by the sensors in the users’ mobile devices and process the pulled information in real-time so that it can accommodate users’

changing situation dynamically.

Cloud Computing based on elastic computing for 4S model has the capability to provide a smart learning environment. SaaS is the largest contributor to the Cloud computing market with a contribution of 89% of total Cloud computing revenue. The SaaS market is primarily driven by the CCC and CRM segments that together account for 50% of SaaS revenues and will continue growing by more than 15% per year. Most adopters of SaaS are companies with relatively straightforward requirements without need for d customization.For this reason manufacturing, operational and financial solutions requiring specific functionalities and integration are slower to move to an on demand paradigm.However, we expect both Cloud service providers and Independent Software Ven

differentiate by offering an increasing number of solutions developed exclusively using PaaS.

It encourages learning system standardization and provides a means for managing it. A traditional e

system can display single content on a single device or multiple contents on one device. The SCC can deliver s

the users so they can use multiple devices applications into a fewer number of fact that the individual applications have few overlaps

Cloud computing elastic models The most important responsibility of the servers is to perform elastic processes such as Elastic Computing for Infrastructure as a Service (IaaS), Elastic Management for Platform as a Service (PaaS) and Elastic Deployment for Software as a Service onstantly. The elastic processing can be described as collecting user information that is pulled by the sensors in the users’ mobile devices and process the time so that it changing situation dynamically. The Smart Cloud Computing based on elastic computing for 4S model has the capability to provide a smart learning environment. SaaS is the largest contributor to the Cloud computing market with a contribution of 89% of total Cloud computing revenue. S market is primarily driven by the CCC and CRM segments that together account for 50% of SaaS revenues and will continue growing by more than 15% per year. Most adopters of SaaS are companies with relatively straightforward requirements without need for deep customization.For this reason manufacturing, operational and financial functionalities and integration are slower to move to an on demand paradigm.However, we expect both Cloud service providers and Independent Software Vendors to differentiate by offering an increasing number of solutions developed exclusively

It encourages learning system standardization and provides a means for managing it. A traditional e-learning system can display single content on a e device or multiple contents on one device. The SCC can deliver s-learning to the users so they can use multiple devices

to render multi learning contents. The multi learning contents can be played in different devices separately to form a virtual class.

Fig1:Elasticity

Fig2:Cloud computing layers Architecture

The architecture is organized into two complementary tiers: a machine machine (M2M) level and a machine cloud (M2C) level. On the M2M level, a group of robots communicate via wireless links to form a collaborative computing unit (i.e., a virtual ad

benefits of forming a collaborative computing unit are multi

computing capability fr

robots can be pooled together to form a virtual ad-hoc cloud infrastructure. Second, within the collaborative computing unit, information can be exchanged for collaborative

making in various robot applications. Finally, it al

within communication range of a cloud access point to access information stored in the cloud infrastructure or send computational requests to the cloud. the M2C level, the centralized cloud to render multi learning contents. The multi learning contents can be played in different devices separately to form a

:Cloud computing layers

The architecture is organized into two complementary tiers: a machine (M2M) level and a machine-to-(M2C) level. On the M2M level, a group of robots communicate via wireless links to form a collaborative computing unit (i.e., a virtual ad-hoc cloud). The benefits of forming a collaborative computing unit are multi-fold. First, the computing capability from individual robots can be pooled together to form a hoc cloud infrastructure. Second, within the collaborative computing unit, information can be exchanged for collaborative decision making in various robot-related applications. Finally, it allows robots not within communication range of a cloud access point to access information stored in the cloud infrastructure or send computational requests to the cloud. On the M2C level, the centralized cloud

(4)

infrastructure provides a pool of shared computation and storage resources

be allocated elastically for real demand.

The elastic computing model allows the group of networked robots to offload computation-intensive tasks for remote execution, resulting in “remote

robots. Moreover, the benefits of a large volume of storage provided by the centralized cloud are two-fold. First, it can unify a large volume of information about the environment, which can be organized in a format usable by robots. Second, it can provide an extensive library of skills or behaviors that are related to task requirements and situational complexities, making it feasible to learn from the history of all cloud-enabled robots.

Modern learning services typically deal with multi-media resources such as graphics, video, images, text

such resources provide an efficient learning environment that helps learners understand the topic of interest better. The awareness of user behavior in the learning process can be very helpful in providing the right contents at the right time. The learning services that include the concept of such awareness and the capability of handling multi-media resources efficiently can be termed smart learning systems. use of context-awareness for user behavior and a way to deliver the corresponding contents to the users.

Fig3:Learning cloud architecture

infrastructure provides a pool of shared ation and storage resources that can be allocated elastically for real-time

The elastic computing model allows the group of networked robots to offload intensive tasks for remote execution, resulting in “remote-brain” , the benefits of a large volume of storage provided by the fold. First, it can unify a large volume of information about the environment, which can be organized in a format usable by robots. Second, it brary of skills or behaviors that are related to task requirements and situational complexities, making it feasible to learn from the history

Modern learning services typically deal media resources such as video, images, text etc., since such resources provide an efficient learning environment that helps learners understand the topic of interest better. The awareness of user behavior in the learning process can be very helpful in providing s at the right time. The learning services that include the concept of such awareness and the capability of media resources efficiently termed smart learning systems.The awareness for user behavior the corresponding

:Learning cloud architecture

Hardware layers are resource layer as a dynamic and scalable physical host pool, software resource layer that offers a unified interface for elearning developers, resource management layer that achieves loose coupling of software and hardware resources, service layer, containing three levels of services (software as a service, platform as a service and infrastructure as a service), application layer that provides with content production, content delivery, virtual laboratory, collaborative learning, assessment and management features

Fig4:context model

The concept of the context model in context-awareness was introduced, which includes the static

descriptions of the user and physical situation. The context model deals with the context objects and the relations among them. The results of the context

allow learning efficiency and outcomes for smart learning, such as learners’ knowledge interests, needs, expertise, and experiences. Using the context mo

smart cloud model

necessary contents to users precisely. In order to collect user’s behavior, sensors in users’ devices were used. Based on the sensing information, the

environment can forecast and prepare the contents by analyzing the collected information. Such a process enables smart learning services to provide the contents to the users at an appropriate time.

It basically supports peer M2M

Capabilities at M2M Devices, M2M Gateways, and M2M Servers. It also supports three M2M reference points: dIa Hardware layers are resource layer as a dynamic and scalable physical host pool, software resource layer that offers a unified interface for elearning developers, management layer that achieves loose coupling of software and hardware resources, service layer, containing three levels of services (software as a service, platform as a service and infrastructure as a service), application layer that provides t production, content delivery, virtual laboratory, collaborative learning, assessment and management features.

The concept of the context model in awareness was introduced, which and dynamic descriptions of the user and physical situation. The context model deals with the context objects and the relations among them. The results of the context-awareness allow learning efficiency and outcomes for smart learning, such as learners’ rests, needs, expertise, and experiences. Using the context model, the can provide the necessary contents to users precisely. In order to collect user’s behavior, sensors in users’ devices were used. Based on the sensing information, the cloud computing environment can forecast and prepare the contents by analyzing the collected information. Such a process enables smart learning services to provide the contents to the users at an appropriate time. standards. It basically supports peer M2M Service Capabilities at M2M Devices, M2M Gateways, and M2M Servers. It also supports three M2M reference points: dIa

(5)

between an M2M Device/Gateway Application and an M2M Device/Gateway, mId between an M2M Device/Gateway and an M2M Server, and mIa between an M2M Network Applications and an M2M Server. dIa and mIa provides uniform interface for M2M Applications. In addition, constrained M2M devices are supported as a kind of M2M Device Application.Most importantly, the Scalable Service Platform is integrated with SC-to-SC interaction capabilities, which, as a unique feature, enables Device-to-Device (D2D), Gateway-to-Gateway (G2G), and Service-to-Server (S2S) direct communications and in turn significantly improves system reliability, scalability, and overall performance.

The availability of lower cost devices, sensors, and actuators with increased computing and lower power has created a huge opportunity for growth in M2M Service Applications. To rapidly realize this growth potential requires faster time to market, lower costs, and re-use of applications within vertical applications and possibly across other markets. The user situation contains the detailed information about users. The user preferences in the user situation specify user actions and required services. A user action indicates some preference in user’s requests for learning services. The required service should help users acquire knowledge in the area of interest, share experience, and collaborate with each other in learning. Each user’s personal information such as personal context is secured by some security setting such as user’s schedule and location. The physical situation includes each terminal’s MAC address, capability, software interface status, and types of software applications. The terminal capability describes the process speed, memory, screen size, resolution, and interface types. The terminal application type describes software applications installed in the terminal. The application type is based on quality of service (QoS) parameters, such

as response time, delay, jitters and bandwidth. It can be categorized into four types, namely (1) conversational service, such as VoIP; (2) real-time (RT) service such as Internet Protocol Television (IPTV) and mobile TV; (3) non-real-time (NRT) services such as email or ftp; (4) interactive services such as web browsing. The context model has user situation information such as user’s requests and the devices they are using. Using this information, the SCC can provide user-aware smart learning service based on E4S. The E4S handles the pulling of sensing information, the analysis of context from the pulled information, the generation of smart content, and the push of smart learning service to individual terminals with different contexts.

The Smart Content generates the fusion content for the user’s device using the harmony adaptation. The harmony adaptation has two steps—Fusion Content Adaptation and Device Synchronization process. The Fusion Content Adaptation presents the synchronization among the fusion contents in ActionNo.The Device Synchronization performs the process of synchronization between devices. For the synchronization of the fusion contents, the Fusion Content Adaptation (FCA) uses the contents that are indicated by Semantic description from Smart Prospect. The adapted contents include <All time {start, duration, delay, end}> and <time {start, duration, delay, end}> information. For the time synchronization, the FCA uses Interpreter Playout Schedule (IPS) to schedule the order of playout.

The basic approach of EC2 is that the user stores their data within the system, paying relatively low rates for data storage. When a user has a job to run, they can pay for as many computing nodes as needed, which are charged at an hourly rate. While nodes are being rented, the user has complete control of the system, having root access to the nodes. These nodes can thus be configured as the user desires with whatever packages and system software

(6)

needed. In particular, the nodes are networked, so can communicate with each other. This allows the user, for example, to run a version of MPI on the nodes and so run a job in parallel. The attraction of the technology is that if the user does not run any jobs, the only cost is for data storage. When a job or jobs run, as many CPUs as useful can be deployed. This changes the mind-set of the researcher: the cost of the job is determined by the total computation time. If an algorithm parallelizes effectively on n CPUs, a problem using that algorithm costs roughly as much to solve using n CPUs in one hour as using one CPU in n hours. m separate jobs cost as much to run sequentially as concurrently.

4.Proposed idea

The main idea is to monitor the resource demand during the current time window in order to make decisions about the server allocations and job admissions during the next time window.The resource allocation problem within a datacenter that runs different type of application workloads, particularly non-interactive and transactional applications. We propose admission control and scheduling mechanism which not only maximizes the resource utilization and profit, but also ensures the SLA requirements of users.Datacenter resource allocation is monitored and reconfigured in regular intervals. At a given point of time, it is assumed that if a host is idle for certain amount of time or not running any applications, then it will be switched-off.

4.1 SLA Enforcement and Rescheduling

1)Let the user request for a VM with capacity Ck. A request is accepted when the datacenter can schedule the VM with capacity Ck on any server assuming all hosted Web VMs are running at 100% utilization and without considering resources used by dynamic HPC VMs. The Web VM is scheduled based on the best-fit manner.

2)If new Web VM is deployed on a server hosting both a dynamic HPC VM and Web VMs, then the future resources available to the dynamic HPC VM will get affected. This scarcity of resources will delay the completion time of HPC job. Thus, the HPC VM will be paused and rescheduled (migrated) to other servers if the HPC job is missing its deadline after deployment of new Web VM.

3)The rescheduling of HPC job is done in such a way that the minimum penalty occurs due to SLA violation. In these cases, since, while scheduling of new Web application, the full utilization of resources by other VMs is considered. Therefore, there will not be any perceptible effect on the execution of other VMs. It can be noted that since static HPC Vm (denoted by red color) is hosted therefore, the available resources on the server for executing new Web application will be the amount of resources unused by HPC VM. Input: Current Utilization of VMs and Current Resource Demand.

Output: Decision on Capacity Planning and Auto-scaling

Notations: VMweb−i: VM running Transactional (Web) Applications;

CurResDemand(VMweb−i): Current

Resource Demand;

CurAllocResVMweb−i: Current

Allocated Capacity;

ReservedRes(VMweb−i): Reserved VMs Capacity Specified in SLA; VMhpc−i: VM running HPC Application

1: forEach VMweb−i do

2: Calculate the current resource demand CurResDemand(VMweb−i)

3: if CurResDemand(VMweb−i) < CurAllocResVMweb−i then

4: Reduce the resource capacity of VMweb−I to match the demand

5: else

6:if CurResDemand(VMweb−i) ≤ ReservedRes(VMweb−i) then

7: Increase the resource capacity of VMweb−i to match the demand

(7)

8: Reduce correspondingly the resource capacity allocated to HPC application (VMhpc−i

) on the same server 9: else

10: if SLA contains Auto-scaling Option

then

11: Initiate new VMs and offload the application demand to new VMs

12: end if 13: end if 14: end if 15: end for

16: forEach Batch Job VMhpc−i do

17: if slack resources available on the server where HPC VM is running then 18: Allocate the slack resources 19: end if

20: Recompute the estimated finish time of the job

21: Reschedule the Batch Job VM if missing the deadline.

22: end for

Smart Prospect is mainly responsible for describing the contents in ActionNo— time, memory, resolution and supported application types. The description is needed for fusion content delivery, because ActionNo specifies the fusion learning content in the Fusion learning DB. For the delivery of fusion content to a user’s device, the SCC is required for harmony adaptation. In the harmony adaptation, the most important part is synchronization. The synchronization part controls the time for synchronization among fusion contents in the same ActionNo. To access the information such as memory, resolution, or application type of the contents in ActionNo, the Smart Prospect uses a Semantic Description using of UVA (Universal Video Adaptation) model that has been developed .The UVA model uses the video content description in MPEG-7 standard and MPEG-21 multimedia framework. The Semantic Description based on the UVA model includes the effort to build a new architecture that supports content with formal semantics. The semantic

description provides the accurate and meaningful information for the fusion content. The semantic description uses XML, ontology and Resource Description Framework (RDF) that help define fusion content clearly and precisely. It also represents systematic information about the contents. The role of the ontology is to formally describe the shared meaning of vocabulary used. The ontology describes the basic fusion learning contents of some domain where learning takes place (e.g., history of science). It includes the relations between these concepts and some basic properties. Based on the ontology, all learning content in the Action No are associated each other. For example, the description of the video content used in semantic description

5.Conclusions

Cloud computing is a solution to many problem of computing. Even we are in IT ages complication of computing has created much disaster to computer world. Lots of crisis has happen in business world as well as in academic environment. Data security, storage, processing power is limited while using traditional computing. Data are also in risk and not available all time.In order to deliver such customized contents to the users at right time, the SCC followed Elastic 4S—Pull, Smart-Push, Smart-Prospect and Smart-Contents. All the services are based on the collected data through the sensors in user’s device. We have utilized the E4S model and analyzed the sensed information within the category of context. The context-aware model handles the fusion media adaptation, synchronization, and transmission for a smart learning service. We have considered various requirements that for the users, the networks, and the cloud. But by using of cloud computing the entire problem is solved.

6.References

[1]Robotics:Architecture, Challenges and Applications Guoqiang Hu, Member,

(8)

IEEE, Wee Peng Tay, Member, IEEE, and Yonggang Wen, Member, IEEE.

[2] Smart Learning Services Based on Smart Cloud Computing Svetlana Kim, Su-Mi Song and Yong-Ik Yoon Department of Multimedia Science, Sookmyung Women’s University, Chungpa-Dong 2-Ga, Yongsan-Gu 140-742.

[3] Md. Anwar Hossain Masud, Xiaodi Huang,“An E-learning System

Architecture based onCloud Computing” World Academy of Science, Engineering and Technology

[4]Cloud robotics: Towards context aware networks, João M. Quintas Pedro Nunes Institute Automations and Systems Laboratory

Coimbra/Portugal,Proceedings of the IASTED International Conference November 7 - 9, 2011 Pittsburgh, USA

Figure

Updating...

References

Updating...

Related subjects :