International Journal of Computer Sciences and Engineering Open Access Research Paper Volume-2, Issue-8 E-ISSN: 2347-2693 Energy Saving For Mobile Users Using Cloud Computing Via S3 K.Venkateswarlu 1*, M.Sri Lakshmi 2, Dr.S.Prem Kumar 3 1 M.Tech, Dept. of CSE, G.Pullaiah College of Engineering and Technology, Kurnool, JNTUA, India 2 M.Tech, Asst. Prof, Dept. of CSE, G.Pullaiah College of Engineering and Technology, Kurnool, JNTUA, India 3 M.Tech, Ph.D, Professor & HOD,Dept. of CSE, G.Pullaiah College of Engineering and Technology, Kurnool, JNTUA, India. www.ijcaonline.org Received: 16/07/ 2014 Revised: 28/07/ 2014 Accepted: 24 /08/ 2014 Published: 31 /08/ 2014 ABSTRACT: With a rise in usage of mobile devices it's continuously expected that a mobile device perform the execution of all applications the approach a desktop device do. Mobile devices became associate integral a part of somebody's life. However, with restricted process power, memory & battery time period of mobile phones it becomes tough to execute computationally intensive applications like image process. Computation offloading provides a way to save lots of energy within which a number of the applying computer code parts are often offloaded from mobile device to run on a distant server. Offloading the computation from mobile devices into cloud can end in extended battery time period, improved knowledge storage capability and process power. Offloading the computation from mobile devices (low in resources) into cloud (resourceful machine) solves the matter. This analysis presents service primarily {based} offloading mechanism that permits execution of multiple computations and repair based access mechanism for higher purpose. Here to perform multiple computations, multiple services square measure run each at consumer aspect (mobile devices) and server aspect (Cloud). The cloud modules contains of Image clump, Image retrieval and image search service whereas robot module encompass Image storage and management, Login and registration,image search and image transfer service. The execution of multiple services can guarantee quicker execution and increase in battery time period. Keywords- Mobile applications, Energy saving, Cloud computing, computation offloading, Mobile Cloud Computing (MCC), Service based offloading mechanism I. INTRODUCTION Cloud computing could be a new paradigm within which computing resources like process, memory, and storage don't seem to be physically gift at the user s location. Instead, a service supplier owns and manages these resources, and users access them via the web. For instance, Amazon net Services lets users store personal information via its straightforward Storage Service (S3) and perform computations on hold on information exploitation the Elastic Compute Cloud (EC2). this sort of computing provides several blessings for businesses as well as low initial capital investment, shorter start-up time for brand spanking new services, lower maintenance and operation prices, higher utilization through virtualization, and easier disaster recovery that create cloud computing a lovely choice. Reports recommend that there square measure many edges in shifting computing from the table high to the cloud. What regarding cloud computing for mobile users? The first constraints for mobile computing square measure restricted energy and wireless information measure. Cloud computing will give energy savings as a service to mobile users, although it additionally poses some distinctive challenges. Mobile systems, like sensible phones, became the first computing platform for several users. Varied studies have known longer battery time period because the most desired feature of such Corresponding Author: K.Venkateswarlu systems. During this project, we tend to propose a mechanism to avoid wasting transportable energy by offloading to cloud Fig 1.Energy saving for mobile users using cloud computing via S3(simple storage service) Cloud computing could be a variety of computing wherever all the resources like hardware, package square measure provided to the user whenever they need. The infrastructure is provided to users on demand by service supplier. The user s square measure charged supported services they 2014, IJCSE All Rights Reserved 88
access. The services that cloud computing provides to the user may be classified as Software as a Service (SaaS), Platform as a service (PaaS), Infrastructure as a Service (IaaS). This ensures that the user s square measure charged with the kind of services they use from service supplier. However, there's no specific definition for cloud, NIST[6] defines Cloud computing as a model for sanctioning present, convenient, on-demand network access to a shared pool of configurable computing resources (e.g., networks, servers, storage, applications, and services) that may be chop-chop provisioned and discharged with bottom management effort or service supplier interaction. Hewitt [3] defines the key role of a cloud knowledge processing system in terms of storing data on the cloud servers, and exploitation cache memory to induce the information. Those purchasers may be PCs, laptops, Smartphone s so on. R. Buyya et al.[4] defines cloud computing as a combination of a gaggle of virtual machines with inner links. L. Youseff et al. [5] declare that cloud computing could be a combination of many new and existing ideas, like grid and distributed computing. Mobile Cloud Computing (MCC) is rising collectively of the foremost necessary twigs of cloud computing. Mobile users these days need that everything that they are doing on their laptop and laptops ought to even be done on their mobile phones. This has given an increase in Mobile Cloud computing. Mobile cloud computing could be a combination of Mobile computing and a cloud computing. The Mobile Cloud Computing Forum defines MCC as associate degree infrastructure wherever each the information storage and also the processing happen outside of the mobile device. Mobile devices have a downside of less energy, restricted memory and battery time period. so running a computationally intensive application like image process, video written material, optical character recognition, and Face detection on a mobile device needs a lot of battery time period, higher procedure power and a lot of disc space. A survey conducted in 2009 discovered that short battery time period is that the most unfavorable feature of mobile devices.(e.g., Apple s iphone 3GS [9],).According to survey conducted by Change wave, forty one p.c of respondents aforementioned the device s short battery life was an enormous concern. Thus, the limitation of energy has been the bottleneck of hand-held mobile devices. As per karthik et al.[2] There square measure four basic approaches to avoid wasting energy and increasing battery time period in mobile devices and that they square measure (i)adopt a brand new generation of semiconductor technology (ii)avoid wasting energy (iii)execute program slowly(iv)eliminate computation altogether through offloading. Amongst all the answer mentioned the most effective alternative was to dump the computation from mobile devices into the cloud within which computation isn't performed on mobile device rather it's done on cloud finish to scale back execution time, and increase in knowledge storage capability & process power & additionally extending battery time period. Offloading has started gaining interest within the mind of developers as a result of it overcomes the disadvantage related to mobile II. ASSOCIATED EXERTION Many connected researches are happen on offloading the information. the choice of wherever to position the execution (local or remote mode) ought to be anyway created supported the number of computation and communication that's needed by the appliance. a little quantity of communication combined with an outsized quantity of computation ought to be performed referable in remote mode, whereas an outsized quantity of communication combined with a little quantity of computation ought to be performed ideally in native mode. Face recognition a sample application [3] to gauge the tradeoff of offloading computation with the intuitive plan of the desired intensive calculus puts in commitment the hardware options of the mobile device. Whereas that, if an equivalent calculus square measure dead by alternative systems with higher hardware options, these processes square measure realised with less effort and in a lot of less time. Analyzing the intensive calculus dividing it in sub processes that square measure distributed between the mobile device and also the cloud infrastructure employing a cascade of classifiers supported the Ad boost algorithmic rule [4] to observe the presence of faces in a picture and also the Eigen faces algorithmic rule [5] to create the coaching and recognition of those faces. Finally, emulating the wireless channel between the mobile device and also the cloud server to look at however the end-to-end reaction time will have an effect on at application. And additionally emulating permits to search out limitations wherever we will get advantage with the utilization of this system. When involves the energy saving construct Power Booster, an automatic power model construction technique[6] that uses built-in battery volt- age sensors and information of battery discharge behavior to observe power consumption whereas expressly dominant the ability management and activity states of individual elements. It needs no external measure instrumentation. We additionally describe Power Tutor, a part power management and activity state musing primarily based tool that uses the model generated by Power Booster for on-line power estimation. Power Booster is meant to create it fast and simple for application developers and finish users to come up with power models for brand spanking new Smartphone variants. Combined, Power Booster and Power Tutor have the goal of gap power modeling and analysis for a lot of Smartphone variants and their users. III. PROBLEM STATEMENT With the fast development of mobile networking and device capability, energy potency becomes a crucial style thought thanks to the restricted battery lifetime of mobile terminals. Process energy value by processor is one in every of the foremost important power intense elements in mobile terminals. The emergence of mobile cloud computing (MCC) provides the chance to avoid wasting process energy through the method of offloading computation tasks to remote server(s). For offloading to induce most energy conservation, the way to separate the task & establish that task to run 2014, IJCSE All Rights Reserved 89
in cloud & that should be run in itinerant should be found. The advances in technology of the last decades have un- doubted turned yesterday s must-have devices into today s stock. Think of the phones with aerials of the late 80, or the Pentium four PCs of some years past. None of them is equivalent to the ability of these days Smartphone s, whose recent worldwide market boost is plain. We have a tendency to use Smartphone s to try and do several of the roles we have a tendency to wont to do on desktops, and lots of new ones. we have a tendency to browse the web, send emails, organize our lives, watch videos transfer knowledge on social networks, use on-line banking, realize our method by exploitation GPS and on-line maps, and communicate in revolutionary ways that. New apps square measure commencing at an improbable pace. Apple iphone commercial s decision to action There s associate degree app for everything says lots on this matter. nevertheless, the a lot of eager we have a tendency to get once exploitation our good phones by putting in new apps, the less happy wearer with the time period of the battery. the matter is that we have a tendency to rely on variety of crucial items of knowledge that square measure solely keep within the device (phone numbers, addresses, notes, appointments, etc.), or, in some cases, that may be got solely by exploitation the web on the fly as several people square measure wont to do. It s therefore necessary to stay our Smartphone operational that every day we have a tendency to concentrate to our battery and check out to avoid wasting it by reducing the quantity of phone calls, or by avoiding to observe too several videos, simply enough to be able to reach home and recharge it. However which means that we have a tendency to can t use our device to the fullest. Many researchers believe that cloud computing is associate degree excel- Lent candidate to assist cut back battery consumption of Smartphones,as well on backup user s knowledge. Indeed, several recent works have centered on building frame works that modify mobile computation offloading to package clones of Smartphones on the cloud (see [7], [8], [9] among others), still on backup systems for knowledge and applications keep in our devices [10], [11], [12]. Each mobile computation offloading and knowledge backup involves communication between the important device and also the cloud. IV. REVIEW OF LITERATURE: Several offloading strategies are planned to this point. Dejan et al. [1] enforced Mobile Augmentation Cloud Services framework for execution elastic mobile applications. The result was analyzed exploitation 2 completely different use case phone applications. The primary application enforced NQueens downside and second application enforced face detection and recognition of video files. The video file was processed with OpenCV and FFmpeg libraries. The result had well-tried that offloading the computation from mobile devices into clouds saves around ninety fifth of energy. X.Jhang et al. [12] planned a brand new elastic application model to boost the potential of resource-constrained mobile devices. One application is divided into multiple elements referred to as web lets.these web lets square measure then migrated to the cloud. This ensures increase in network information measure, storage and computation power. MAUI [7] provides fine-grained code offload to extend energy. Island has an extra advantage that in runtime it decides that strategies ought to be remotely dead and that methodology to dump, which ends up in finest energy savings. Island ensures that partitioning is completed with least resistance. The proxy implements the selections created by the island problem solver, handling each management and knowledge transfer supported the choice. The problem solver decides whether or not the tactic in program ought to be dead regionally or remotely supported the input from the profiler. The profiler gathers the identification data that is employed to higher predict whether or not future invocations ought to be. This work ensures application s performance and energy consumption by reducing the burden on applied scientist for program partitioning. V. MOBILE AUGMENTATION CLOUD SERVICES The goal of our MACS middleware is to modify the execution of elastic mobile applications. Zhang et al. [17] think about elastic applications to possess 2 distinct properties. First, Associate in nursing elastic application execution is split part on the device and part on the cloud. Second, this division isn't static, however is dynamically adjusted throughout application runtime. The advantages of getting such Associate in nursing application model are that the mobile applications will still run severally on mobile platforms; however can even reach cloud resources on demand and availableness. Thus, mobile applications don't seem to be restricted by the constraints of the prevailing device capacities. MACS design is delineate on Figure one. So as to use MACS middleware, the appliance ought to be structured mistreatment established golem services pattern. Golem is already established because the most outstanding mobile platform. In addition, its application design model permits decomposition of applications into service parts that As service-based implementation is adopted, for every service we will profile following metadata: 1. Type: whether or not may be offloaded or not 2. Memory cost: the memory consumption of the service on the mobile device 3. Code size: size of compiled code of the service 4. Dependency data on different services, for every connected module, we have a tendency to collect following: Transfer size: quantity of information to be transferred Send size: quantity of information to be sent Receive size: quantity of information to be received VI. SERVICE BASED OFFLOADING Mechanism: Here we have a tendency to propose a service based access mechanism that allows user to utilize multiple computation 2014, IJCSE All Rights Reserved 90
and allows service based design to be used for higher purpose as shown in figure I. we have a tendency to propose following services to be employed in server module and a mobile module. A. CLOUD MODULE: 1) Image clustering service This module can facilitate the appliance to reason or classify pictures into varied sections supported the tags related to these pictures. Image Clustering may be a suggests that for high-level description of image content. Clustering the pictures is a lot of advantage for reducing the looking time of images within the information. Thus the pictures requested by the user may be offered in less time. For Clustering 2 styles of algorithms area unit offered i.e. hierarchical formula and partitioned formula. Partitional algorithms construct varied partitions and so measure them whereas hierarchic algorithms produce a hierarchic decomposition of set of objects victimization some criteria. Hierarchical Clustering may be divided into bottom up (agglomerative) or prime down(divisive) bottom-up (agglomerative) Clustering starts with every item in its own cluster, finds the simplest combine to merge into a brand new cluster whereas prime down(divisive) formula starts with all the information during a single cluster, considering each attainable thanks to divide the cluster into 2. opt for the simplest division and recursively operate either side the set of objects victimization some criterion s suggests that [8] may be a acknowledge partitioned formula that partitions the information set into k sets such all points during a given subset area unit nearest to an equivalent centre. In detail, it every which way selects k of the instances to represent the clusters. Supported the chosen attributes, all remaining instances area unit appointed to their nearer centre. K-means then computes the new centers by taking the mean of all information points happiness to an equivalent cluster. The operation is iterated till there's no amendment within the gravity centers. If k can not be identified prior to time, varied values of k may be evaluated till the foremost appropriate one is found. For Clustering of pictures into the cloud we have a tendency to use Kmeans formula because of following advantages: Kmeans has a quickest time period wherever every iteration is linear. It takes K range of clusters as an input parameter and produces precisely k clusters. For Image Clustering Service we have a tendency to perform following steps: 1. Capture Image and connected details 2. Perform K-Means Clustering victimization image Keywords and class information 3. Generate clusters of pictures having similar keywords 4. Generate mapping model consisting of image clusters and associated pictures clustered and keeps in cloud, whenever the requirement happens to retrieve the pictures from a group of images it's needed to go looking a picture. For looking a picture from cloud the subsequent formula steps may be used. 1. Capture Search question Keyword 2. Establish clusters mapping to every keyword such 3. Filter clusters supported distance of keywords with the cluster suggests that 4. Establish pictures happiness to corresponding clusters 5. Forward list of filtered search pictures to shopper 3) Image retrieval service This module can facilitate storage and retrieval of image structure or image information supported specific necessities of a user. Whenever mobile user wish to transfer a specific image the image retrieval service is employed on the server aspect in conjunction with image transfer service on a shopper side to serve the request supported the mobile users demand. B. MOBILE MODULE 1) Image storage and management This module can give user to store pictures onto the search application which might be more facilitated to be retrieved victimization search mechanism. This module facilitate uploads the pictures from automaton mobile phones to cloud and even be wont to retrieve images. 2) Login and registration This module can give the system to demonstrate the user and thereby facilitate solely valid users to access the system. The automaton transportable can solely give access to licensed users thereby preventing unauthorized users to access the system. 3) Image search This is the computation offloaded mechanism provided by the appliance to go looking within the set of categorized pictures set on the server whereby the machine practicality is been dead at the server. Whenever a picture is to be searched from a group of pictures this service at automaton module can guarantee looking of a picture on the cloud aspect. 4) Image transfer The searched pictures area unit expedited here to be retrieved and downloaded onto the mobile device victimization the retrieval service set at the server. The retrieval service at the cloud aspect in conjunction with image transfer service at automaton module side ensures the transfer of the desired pictures on the user s transportable. 2) Image search service This module can facilitate the user to execute the search method that is offloaded to the server. Once the image is 2014, IJCSE All Rights Reserved 91
Fig 2- Service Based Offloading Mechanism 5) Performance analysis We currently assess the performance of the planned energy saving in humanoid devices mistreatment cloud computing schemes to point out that they're so light-weight. We are going to concentrate on the value of the potency of the battery life and our planned k-means technique. The experiment is conducted mistreatment java on a Jdk1.6 with Associate in Nursing Intel a pair of.1ghz processor,4gb memory, computer network interface, Android device and a cloud platform. Fig 3. Shows performance analysis VII. CONCLUSION Cloud computing will doubtless save energy for mobile users. Mobile cloud computing services would be miserably Different from cloud services for desktops as a result of they need to provide energy saving. Growth of complicated applications to mobile devices with support of cloud computing infrastructure demands higher understanding and battery life time. The computation in mobile devices need additional battery compare to cloud. For this reason during this paper, we have a tendency to confer an application designed to avoid wasting battery life time of a tool by offloading knowledge to cloud. From the obtained results, we have a tendency to contemplate that offloading computation from mobile devices to cloud computing infrastructure will be done simply. REFERENCES [1] Dejan Kovachev, Tian Yu and Ralf Klamma, Adaptive Computation Offloading from Mobile Devices into the Cloud, IEEE 2012. [2] Karthik Kumar and Y.-H. Lu, Cloud Computing for Mobile Users: Can Offloading Computation Save Energy? Computer, vol. 43, no. 4, pp. 51 56, April 2010. [3] C. Hewitt, Orgs for scalable, robust, privacy-friendly client cloud computing, Internet Computing, IEEE, vol. 12, no. 5,pp. 96 99, 2008. [4] R. Buyya, C. Yeo, and S. Venugopal, Market-oriented services as computing utilities, in High Performance Computing and Communications, 2008. HPCC 08. 10 th IEEE International Conference on. IEEE, 2008, pp. 5 13. [5] L. Youseff, M. Butrico, and D. Da Silva, Toward a unified ontology of cloud computing, in Grid Computing Environments Workshop, 2008. GCE 08. IEEE, 2008, pp. 1 10. [6] P. Mell and T. Grance, "The NIST Definition of CloudComputing,"2009.[Online]. Available:http://csrc.nist.gov/groups/SNS/ cloud- \computing/cloud-defv15.doc [7] S. Kosta, C. Perta, J. Stefa, P. Hui, and A. Mei, Clone2clone (c2c): Enable peer-to-peer networking of smartphones on the cloud, T-Labs, Deutsche Telekom, Tech. Rep. TR- SK032012AM, 2012. [8] S. Kosta, A. Aucinas, P. Hui, R. Mortier, and X. Zhang, Thinkair: Dynamic resource allocation and parallel execution in the cloud for mobile code offloading. in Proc. of IEEE INFOCOM 2012, 2012. [9] E. Cuervo, A. Balasubramanian, D. Cho, A. Wolman, S. Saroiu, R. Chandra, and P. Bahl, Maui: making smartphones last longer with code offload, in Proc. of MobiSys 10, 2010. [10] Joe and Y. Lee, Design of remote control system for data protection and backup in mobile devices, in Proc. of ICIS 2011,2011. [11] V. Ottaviani, A. Lentini, A. Grillo, S. D. Cesare, and G. Italiano, Shared backup & restore: Save, recover and share personal information into closed groups of smartphones, in Proc. of IFIP NTMS 2011, 2011. [12] C. Ai, J. Liu, C. Fan, X. Zhang, and J. Zou, Enhancing personal information security on android with a new synchronization scheme, in Proc. of WiCOM 2011, 2011. 2014, IJCSE All Rights Reserved 92