Simulation Modelling Practice and Theory 50 (2015) 96108Contents lists available at ScienceDirect

Simulation Modelling Practice and Theory

journal homepage: www.elsevier .com/locate /s impatMobile storage augmentation in mobile cloud computing:Taxonomy, approaches, and open issueshttp://dx.doi.org/10.1016/j.simpat.2014.05.0091569-190X/ 2014 Elsevier B.V. All rights reserved.

Corresponding author. Tel.: +60 1125273423.E-mail addresses: nazanin.aminzadeh@gmail.com (N. Aminzadeh), sanaei@ieee.org (Z. Sanaei), sitihafizah@um.edu.my (S.H. Ab Hamid).Nazanin Aminzadeh , Zohreh Sanaei, Siti Hafizah Ab HamidFaculty of Computer Science and IT, University of Malaya, Kuala Lumpur, Malaysiaa r t i c l e i n f o

Article history:Available online 15 June 2014

Keywords:Mobile cloud computingCloud computingBig dataMobile dataMobile data storageTaxonomya b s t r a c t

Worldwide employment of mobile devices in various critical domains, particularlyhealthcare, disaster recovery, and education has revolutionized data generation rate. How-ever, rapidly rising data volume intensifies data storage and battery limitations of mobiledevices. Mobile Cloud Computing (MCC) as the state-of-the-art mobile computing aims toaugment mobile storage by leveraging infinite cloud resources to provide unlimitedstorage capabilities with energy-dissipation prevention. Researchers have alreadysurveyed varied MCC aspects and its challenges, but successful futuristic Mobile StorageAugmentation (MSA) approaches demand deep insight into the current storage augmenta-tion solutions that highlights critical challenges, which are lacking. This paper thoroughlyinvestigates the main MSA issues in three domains of mobile computing, cloud computing,and MCC to present a taxonomy. Also, it examines several credible MSA approaches andmechanisms in MCC, classifies characteristics of cloud-based storage resources, andpresents open issues that direct future research.

2014 Elsevier B.V. All rights reserved.1. Introduction

Mobile data volume has been increased drastically in recent years. Analysys Masons report [1] forecasts 6.3 times growthof mobile data volume traffic amid 2013 and 2018. One reason, among the bevy of inducements is the exploitation of mobiledevices in various critical domains of healthcare [2], disaster recovery [3], and education [4]. The voluminous and rapidlygenerated data in various modalities, which is known as big data [5], demands high capacity, flexible, and reliable storageinfrastructure.

However, mobile devices are characterized by storage constraints, limited processing, and a short span battery. Storagelimitation and energy consumption are critical factors for resource constrained non-stationary computing devices, especiallysmartphones. High performance and energy-efficient data storage ensures the battery lifes augmentation. Despite advance-ments for augmenting a mobile devices storage including employment of flash and Secure Digital (SD) cards, current richmobile applications [6] demand higher storage capacity. Reducing the effects of mobile devices deficiencies and unreliablewireless connections in comparison with wired networks, are the ultimate goal in the plethora of efforts [7,8] and research torealize end-users demand.

The emergence of the cloud as a rich resource with unlimited computing and storage capacities can be considered as agood solution to mobile devices resource constraints. The use of a wireless medium, the offloaded data, the dependency

http://crossmark.crossref.org/dialog/?doi=10.1016/j.simpat.2014.05.009&domain=pdfhttp://dx.doi.org/10.1016/j.simpat.2014.05.009mailto:nazanin.aminzadeh@gmail.commailto:sanaei@ieee.orgmailto:sitihafizah@um.edu.myhttp://dx.doi.org/10.1016/j.simpat.2014.05.009http://www.sciencedirect.com/science/journal/1569190Xhttp://www.elsevier.com/locate/simpat

Table 1List of acronym definitions.

Acronym Description

ACID Atomicity, Consistency, Isolation, DurabilityDBMS Data Base Management SystemDSP Decryption Service ProviderESP Encryption Service ProviderGPS Global Positioning SystemsI/O Input/OutputIaaS Infrastructure as a ServiceICN Information Centric NetworkiSCSI Internet Small Computer System InterfaceLMH Large Mobile HostMANET Mobile Ad-hoc NetworkMCC Mobile Cloud ComputingMNO Mobile Network OperatorsMSA Mobile Storage AugmentationPC Personal ComputerPDA Personal Digital AssistantsPP-CP-ABE Privacy Preserving-Ciphtertext Policy-Attribute

Based EncryptionSaaS Storage as a ServiceSAL Storage Abstraction LayerSD Secure DigitalSIM Subscriber Identity ModuleSLA Service Level AgreementSMH Small Mobile HostSOA Service Oriented ArchitectureVM Virtual MachineWAN Wide Area NetworkWi-Fi Wireless FidelityWLAN Wireless Local Area Network

N. Aminzadeh et al. / Simulation Modelling Practice and Theory 50 (2015) 96108 97on a specific vendor, and data replication are among several challenges in this domain. Comprehensive studies [911] havereviewed and tried to address challenges in this area. However, the latest endeavor is to deploy cloud resources to augmentcomputing [12] and storage [13] capabilities for a multitude of mobile devices which leads to the state-of-the-art MCC.

The MCC paradigm combines cloud computing, mobile computing, and networking [14] to enhance the performance andcapacity of mobile devices. It is characterized by inherited mobility and rich services from mobile and cloud computingwhere a resource poverty (storage, computation, and battery) can impede the vision of time-, location-, and system type-freeubiquitous computing [15].

In the previous works [1419], the MCC domain have been comprehensively investigated from various perspectives. In[19], authors presented an extensive survey of heterogeneity in the MCC domain, presented an MCC definition, identifiedmajor MCC challenges, devised its taxonomy, and highlighted several crucial open issues that help to identify future researchdirections. Cloud-based augmentation [16] surveys the recent mobile augmentation efforts that employ cloud computinginfrastructures to enhance computing capabilities of resource-constraint mobile devices, especially smartphones. To theextent of our knowledge, investigation of storage augmentation issues in the MCC domain is a nascent literature and requirescomprehensive study and analysis.

In this paper, we comprehensively analyze MSA issues in the context of mobile computing, cloud computing, and MCCwhere each domains issues are investigated. Based on the investigated issues, we classify MSA issues into three classesof mobile device, cloud-based and converged issues. Based on a review of prominent MSA approaches in MCC, we classifycloud-based storage characteristics in a taxonomy that encompasses architecture, capacity, tiering, mobility, location, andback-end connectivity. The paper highlights several open issues in MCC for MSA to pave the way for future efforts. Mobiledevices and smartphones are used interchangeably in this paper. Table 1 provides a list of acronyms used throughout thepaper.

The remainder of this paper is presented as follows. Section 2 presents the motivation for MSA in MCC based on a devisedtaxonomy of issues. Section 3 reviews current approaches for MSA in MCC. Section 4 provides a taxonomy of cloud-basedstorage characteristics in MCC. Open issues are highlighted in Sections 5 and 6 concludes the paper.2. Motivation

Contemporary smartphones are dominant mobile devices that not only provide the basic telephony features of traditionalcellphones, but also incorporate the functionalities of several other digital devices, particularly Personal Digital Assistants(PDA), Global Positioning Systems (GPS), sound recorders, and digital cameras [20] and are contributing to rapid digital datageneration rate by producing data files, including emails, spreadsheets, bank statements, and multimedia files (i.e., video,

98 N. Aminzadeh et al. / Simulation Modelling Practice and Theory 50 (2015) 96108music, images, and any location related information) [21]. However, such rapid data generation by mobile devices namedmobile data demands extremely large elastic storage on mobile devices which is beyond their intrinsic capabilities. Mobiledevices are characterized by storage constraints, limited processing, and a short span battery. Hence, storage limitation andenergy consumption are critical factors for resource constrained non-stationary computing devices. Moreover, issues, partic-ularly mobility, limited wireless throughput, and a small user interface are among real challenges that affect data availabilityin mobile computing [16]. High performance energy-efficient storage ensures the battery life prolonging.

Despite the momentous role of mobile devices in various aspects of our daily life, resource poverty is a mobile devicesmajor deficiency that impedes an end-users rich experience and demands. The incorporation of SD cards to expand storagecapacity in addition to the available built-in and on-SIM (Subscriber Identity Module) storage on smartphones as an instanceof mobile devices, cannot compete with end-users demand. To address poor storage capacity, a plethora of research hasfocused on two varied hardware and software level approaches; shrinking big data volume and augmenting storagecapacities.

To shrink big data, researchers in [22,23] try to apply quantum theory to optimize the use of current storage by up to 80%.Quantum data represent bits of data that use quantum bits or qubits to reduce the amount of bits used by each data in thestorage. Currently, a bit uses 12 atoms; however, in 2012, IBM [24] demonstrated feasibility of using only eight atoms tostore one bit data. Quantum theory can realize the possibility of dedicating only one atom to each bit to ensure a tremendousreduction in storage utilization (up to 80%). However, the fully realization of this vision demands further extensive research.Alternatively, storage augmentation approaches aim to alleviate storage shortcomings by leveraging lightweight methods.

The review of storage augmentation approaches in mobile devices, cloud resources, and MCC has led us to devise ataxonomy of issues as depicted in Fig. 1. The taxonomy encompasses three main classes of mobile device, cloud-based,and converged issues, which are described in more details as follows.2.1. Mobile device issues

In the work [16], a taxonomy of augmentation motivation has been presented based on the intrinsic deficiencies of mobiledevices. Storage augmentation approaches mainly facing issues pertaining to intrinsic and non-intrinsic deficiencies ofmobile devices. Hence, the same taxonomy can be applied for the classification of storage augmentation issues. The process-ing power of mobile devices, energy resources, local storage, visualization capabilities, data safety, security, and privacy arethe vital issues that pose challenges and encumber efficient and effective mobile data storage.

Mobile devices are characterized by limited processing power, which implies restricted cache equipment. Restrictedcache imposes more I/O operation and increased energy consumption and processing time for data-intensive operations.Data transmission in intermittent wireless networks is an energy-consuming task. Intermittency due to the unreliabilityof a wireless link or a planned disconnection to save battery life are inevitable [25]. Therefore, data availability on mobiledevices becomes a significant research direction. However, being nomadic, mobile devices cannot benefit from the vast datastorage that is available to their immobile counterparts, which confines the amount of locally-stored data. Moreover, amobile DataBase Management Systems (DBMS) functionality is influenced by the limited processing power and memoryof the host mobile devices. The size-constraint screens and the gracile keyboards of mobile devices as visualization limita-tions are other impediments in the design of storage augmentation approaches.

Researchers endeavor to address aforementioned issues by major MSA approaches including caching [26], hoarding, andreplication mechanisms [27] as well as broadcasting [28] and summarization [25] for mobile environment data.Fig. 1. Taxonomy of MSA issues in MCC.

N. Aminzadeh et al. / Simulation Modelling Practice and Theory 50 (2015) 96108 992.2. Cloud-based issues

Storage augmentation issues are not merely peculiar to the mobile environment. Advancements in data storage technol-ogy from magnetic tapes and disks, optical/magneto-optical storage media, and flash memory to a storage area network,network attached storage, and even storage virtualization approaches [29] cannot compete with the drastic growth of digitaldata to terabytes and petabytes [9], which calls for additional data-intensive computing and management mechanisms.

The advent of the cloud as a rich computing resource [30] has provided virtually infinite data storage to the industry andindividuals. The cloud as proposed in [30] is a type of parallel and distributed system consisting of a collection of intercon-nected and virtualized computers dynamically provisioned and presented as one or more unified computing resources basedon service-level agreements established through negotiation between the service provider and consumers. Cloud comput-ing exploits cloud resources to provide ubiquitous, convenient, on-demand network access to a shared pool of configurablecomputing resources (e.g., networks, servers, storage, applications, and services) that can be rapidly provisioned and releasedwith minimal management effort or service provider interaction [31].

However, reliability, Service Level Agreement (SLA), performance, portability, security, and data confidentiality [9] can beconsidered major cloud-related storage augmentation issues. Reliability can be realized with redundancy, which may jeop-ardize guarantees for Atomicity, Consistency, Isolation, and Durability (ACID) transactions. Active standardization to mitigatea customers data extraction as an approach for the data lock-in and portability issue, the concurrent execution of an appli-cations treads, optimized online scheduling [32] and the rich encryption as mechanisms for performance and security areamong some of the effective approaches for addressing the aforementioned issues [33].

2.3. Converged issues

The power of MCC originates from the convergence of mobile and cloud computing, which merges mobility, richresources, and functionality as renowned advantages of each domain [15]. Resource poverty processing power, storagecapacity, and battery power are inborn characteristics of mobile devices, and the consequent mobile device storageaugmentation issues can be alleviated by leveraging MCC. MCC is a rich mobile computing technology that leverages theunified elastic resources of various clouds and network technologies toward an unrestricted functionality, storage, andmobility to serve a multitude of mobile devices anywhere and anytime through the channel of the Ethernet or Internetregardless of the heterogeneous environments and platforms based on the pay-as-you-use principle [19].

However, MCC is heterogeneous by nature because of inhomogeneous converged computing and networking technolo-gies [19]. MCC heterogeneity is a consequence of diverse hardwares, operating systems, development languages, datastructures and network technologies. Although mobile computing is augmented through rich resources of the cloud, theheterogeneity-based issues pose real challenges for storage amelioration in MCC. Data portability and interoperability,energy efficiency, long WAN latency, and data fragmentations are among the major issues of storage augmentation inMCC as identified in our proposed taxonomy. Addressing mobile data storage issues in MCC has leaded to several MSAapproaches. In the next section, we critically review the state-of-the-art approaches.

3. Mobile-cloud storage augmentation approaches

In this section, we review credible efforts that aimed to address storage amelioration and the I/O performance of mobiledevices by leveraging cloud resources and lightweight methods. We analyze and synthesize the approaches from severalaspects, including mobility, energy efficiency, latency, reliability, and architecture which are presented in Table 2.

Attribute-Based Data Storage (ABDS)The ABDS system [34] is part of the security framework presented in [34]. The framework encompasses two prominentcomponents: namely, a privacy preserving CP-ABE (PP-CP-ABE) and an ABDS scheme. By leveraging PP-CP-ABE, the cloudhandles encryption and decryption tasks that entail an excessive load on resource-poor mobile devices. Two core parts,the Encryption Service Provider (ESP) and the Decryption Service Provider (DSP), provide data confidentiality and securityservices without being aware of the encryption key and data contents. Similar to Phoenix [35], ABDS exploits the datapartitioning mechanism. However, ABDS differs from Phoenix by outsourcing the encryption and decryption tasks tothe public cloud. The ABDS scheme partitions data to a number of blocks and manages the encryption operations onthe required blocks independently. Consequently, when modifying a file on the cloud, the update and encryption tasksare merely applied on specific blocks, which balance the storage and communication overhead through data storageand retrieval.Fig. 2 depicts ABDS framework. The data owner refers to the mobile nodes in the framework. The data owner accesses thestorage service provider services in a secure, energy-efficient, and optimized manner by utilizing outsourced services oftwo prominent components (i.e., DSP and ESP).Encryption through a data partitioning approach minimizes the data management and the cloud vendors charged costs.However, encryption requires additional control information for data blocks, which imposes extra overhead for the data

Table 2Comparison of cloud-based storage augmentation approaches.

ABDS SmartBox WhereStore Wukong Phoenix E-DRM EECRS MOMCC UbiqStor SAMI

Mobility Immobile Immobile Immobile Immobile Mobile Mobile Mobile Mobile Immobile ImmobileStorage augmentation High High High High Low Low Low Low High HighEnergy Efficiency Low Low Low Low High High High High Medium MediumArchitecture ClientServer ClientServer ClientServer ClientServer P2P P2P P2P P2P Hybrid HybridTiering Single-tier Single-tier Single-tier Single-tier Single-tier Single-tier Single-tier Single-tier Multi-tier Multi-tierWAN Latency High High High High Low Low Low Low Hybrid HybridData Blocking Low Low Low Low High High High High Medium MediumReliability High High High High Low Low Low Low Medium MediumAvailability High High High High Low Low Low Low Medium MediumSecurity and Trust High High High High Low Low Low Low Medium MediumData Safety High High High High Low Low Low Low Medium MediumLocality Low Low Low Low High High High High Medium MediumI/O Performance Varied Varied Varied Varied Low Low Low Low Medium MediumBack-end Connectivity Wired Wired Wired Wired Wireless Wireless Wireless Wireless Wired WiredCommunication Platform Internet Internet Internet Internet Ethernet Ethernet Ethernet Ethernet Internet/EthernetContext Awareness Low Low Low Low High High High High Medium MediumClient Networking Cellular/WLAN

100N.A

minzadeh

etal./Sim

ulationModelling

Practiceand

Theory50

(2015)96

108

Fig. 2. ABDS framework.

N. Aminzadeh et al. / Simulation Modelling Practice and Theory 50 (2015) 96108 101management tasks. Moreover, the data block size should be identified optimally to satisfy the goal of minimizing datamanagement overhead tasks.SmartBoxSmartBox [21] is a solution for mobile storage capacity expansion through cloud resources. It employs cloud spacereferred to as shadow storage for storing and retrieving personal data by assigning a unique account for each mobiledevice. SmartBox is a rich personal repository and provides a public space as common storage for sharing data amongassociated mobile devices. Data access and navigation is provided via traditional hierarchical namespace and employsan attribute-based method for semantic query, which uses publisher-provider metadata.Nevertheless, the SmartBox design conforms to write once read many policy, which impedes its application as an enter-prise solution. In addition, it requires a permanent connection to access data on the cloud, which is not feasible in manycases.WhereStoreA location-aware data storage approach is considered by WhereStore [36] to enhance smartphones I/O performance byusing cloud storage resources. The prediction and caching of adjacent place information through the replication of prom-inent data retrieved from the cloud is proposed in this solution to optimize data transfer time and to conserve energy.Although several human mobility models and routing techniques [37] are proposed to figure out movement plan ofmobile users, predicting the users destination efficiently and specifying the appropriate state and time for caching arenon-trivial tasks, which require further investigation.WukongWukong [13] is an additional I/O performance optimization and storage enlargement effort and is a cloud-based file ser-vice that provides user-friendly, transparent data access to inhomogeneous remote cloud storage services for mobiledevices. By leveraging the Storage Abstraction Layer (SAL) and plugin procedures, cloud data and services, namely Drop-box and Amazon S3 can be accessed transparently and remotely via mobile applications as if they are available locally.Wukong realizes the provision of a uniform interface to access various cloud services simultaneously. The optimizationstack, cache management, and pre-fetch mechanisms are utilized to improve the throughput and minimize long WANlatency. Moreover, the application of encryption and compression mechanisms enhances data security and communica-tion, respectively, and completes the contribution of this approach.However, limited wireless bandwidth and I/O operation costs are impediments that hinder Wukong in minimizing longWAN latency. Moreover, the employment of efficient compression methods for multimedia files is required for thesuccess of this proposal. The proposed compression method is more beneficial for minimizing a text files size. These

102 N. Aminzadeh et al. / Simulation Modelling Practice and Theory 50 (2015) 96108objects encompass file types with a high ratio of compression. In addition, the prefetching method is more efficient insequential reading, which requires an improvement in random reading to enhance the performance of the work. How-ever, in current prototype, defining large intervals among the open and read processes would save time during readoperations.PhoenixPhoenix [35] is a distributed communication and storage protocol that considers the power-aware storage sharing ofeager nearby mobile devices in the cloud for one-hop networks. Unlike E-DRM [38], which utilizes a periodic broadcastof buffered data, Phoenix ensures data availability on participant mobile nodes by breaking each data content to a numberof blocks and copying them on at least two or more mobile devices. Phoenix provides edge nodes with a true number ofdata blocks copied to maintain data availability and longevity in the self-organized cloud of mobile devices. Authors focuson the extending the data survival time by minimizing disconnections, by administrating node movement, and by leavingthe support for the data update tasks to databases built on Phoenixs platform.In addition, autonomous management and maintenance is realized by utilizing an advertising model. When a nodeparticipates, it starts an asynchronous advertisement timer to decide whether to broadcast a block as a winner or tobecome a follower contributor as a loser. The Phoenix performance is evaluated through testbed and simulation exper-iments. Simulation is performed via TOSSIM in TinyOS. The number of data block copies as well as the effect of nodesfailure and mobility are among the main factors examined in 100 simulations. The results show Phoenix support for dataavailability because the number of available block copies is rarely less than the number of nodes in the network wherenodes are immobile. In the case of node mobility, it ensures data longevity by maintaining at least two copies of blocks inthe network. However, the current evaluation considers only one-hop networks. The simulation in multi-hop networks[39] can improve Phoenixs applicability.Phoenixs dependency on defining the number of participant nodes, the lack of authentication over a wireless network,and the overhead imposed on the network, in addition to the security, data safety, and privacy issues are among thechallenging tasks that can hinder Phoenixs application in a real scenario.Eager replication extended Database State Machine (E-DRM)An E-DRM [38] is an energy-efficient replication approach for a Mobile Ad-hoc NETwork (MANET) [40]. A MANET encom-passes a network of wireless mobile servers and clients. The nodes of a MANET are nomadic and have no permanentaccess to electricity. Hence, MANETs are restricted from an energy source. Eager replication maintains data consistencyvia the guarantee of an identical value for all copies on nodes. Unlike Phoenix [35] that focuses on preserving a true num-ber of copies in the network, E-DRMminimizes the number of message broadcasts to conserve energy. In E-DRM, authorspropose an eager replication scheme for application on a Database State Machine (DSM) approach introduced in [41]. TheDSM supports atomic broadcasts in which nodes consent to the order and the set of delivered transactions.E-DRM considers two mobile node types in the architecture of a MANET. Small Mobile Hosts (SMH) have restrictedresources, and Large Mobile Hosts (LMH) have a rich resource capacity. SMHs contain only a part of the database toaccomplish query tasks. Complete databases reside on LMHs and are responsible for broadcasting, certifying, and updat-ing tasks. E-DRM minimizes message broadcasting overhead by buffering and periodic broadcasting at the beginning ofthe broadcast cycle. The broadcast cycle commences when a client requests a SMH to update or read data. SMHs committhe transaction to a LMH to be processed and broadcasted to other LMHs. LMHs commit or abort the transaction and cer-tify it as a correct transaction. The cycle ends when a LMH broadcasts the result to a SMH to be transmitted to the client.Fig. 3. E-DRM broadcast cycle.

N. Aminzadeh et al. / Simulation Modelling Practice and Theory 50 (2015) 96108 103Fig. 3 shows the broadcast cycle in detail. The E-DRM approach considers the start of the broadcast cycle for dispatchingtransactions to conserve energy. However, in real-time, all LMHs validate the transaction locally and conserve the energythat would be consumed for message broadcasting and consistency management.EECRSEECRS [42] is an energy-aware content retrieval scheme for mobile cloud. It proposes a directive-selective forwardingscheme for broadcasting interest packets in a MANET to obtain the contents. The EECRS approach is considered in anInformation Centric Network (ICN). The contents of an ICN are characterized by a hierarchical-based naming strategy.Hence, names identify the requested contents. The EECRS improves energy efficiency and scalability by eliminatinginterest packets duplication to diminish traffic load. In addition, it searches multiple caches in parallel to fill the missinggap of the networks node where interest packets fail to reach.Nonetheless, the inability of the EECRS to guarantee a 100% hit rate through the proposed approach hinders its full appli-cation. Authors develop an intelligent broadcast approach for content inquiry in case of missing nodes. Although thismethod ensures a content node hit, the imposed traffic overhead decelerates its performance and energy-efficiency.Market-Oriented Mobile Cloud Computing (MOMCC)MOMCC [43] is a service-oriented architecture-based mobile application development framework. The prominent engag-ing entities of the MOMCC architecture are the central supervisory entity as governor, service programmer, mobile host,and requester mobile node. In MOMCC, service developers create services and upload them in a central database for pub-lic availability. Individual mobile device owners who are interested in sharing their resources with nearby resource-poormobile devices register their interest with the governor. The governor authenticates and authorizes the registered mobiledevices that enter a request for hosting specific services. Moreover, the governor finds a secure proximate host for serviceexecution in response to the service requester. The MOMCC overcomes the imposed overhead of code offloading in [44]by restricting hosting to services executed without offloading. The participating hosts as well as the governor, service pro-grammer, and application developer will earn money on publishing, governing, hosting, and sharing services. The lack ofdirect attachment between the service developer and the requester enhances privacy in the proposed framework.The promise of this approach is the execution of services on a multitude of nearby mobile devices under the administra-tion of a centrally trusted governing entity. The employment of adjacent mobile devices improves the availability ofresources, which can be leveraged to alleviate mobile devices storage limitations. Moreover, the communication ofmobile devices over a WLAN minimizes latency and conserves the energy, which would otherwise be dissipated oncellular networks. However, the MOMCC architecture shortcomings, such as the restricted computing of host devices,the necessity for fine-grained services, and the dependency on networking impede its employment for enterpriseapplications and in offline mode.UbiqStorUbiqStor [45] is an internet Small Computer System Interface (iSCSI) cache server that reduces the response delay timebetween mobile users and storage servers. The access to remote fixed storage servers via the nearest immobile UbiqStorserver as a hybrid, multi-tier solution provides mobile clients energy-efficient access to rich storage capacity by eliminat-ing long WAN latency in case of direct access to non-proximate storage servers. Initiator, target, and block managementmodules are components of UbiqStor, which handles iSCSI read and write mobile clients requests through a cachingmechanism.Despite the enhancement of transfer latency and response delays, wireless network intermittency and cache serverproximity factors can affect UbiqStors performance and can hinder its successful application.Service-based Arbitrated Multi-tier Infrastructure (SAMI)SAMI [15] is an Infrastructure as a Service (IaaS) hybrid model for mobile cloud computing that addresses storageaugmentation issues in addition to its main purpose of data-intensive computations in MCC. It promotes the I/O perfor-mance as well as the storage capacity of resource-poor mobile devices through a hybrid architecture that encompassesthree layers of an infrastructure, an arbitrator and a Service-Oriented Architecture (SOA). Mobile device users can benefitfrom SAMIs multi-tier infrastructure, which provides access to distant fixed clouds, proximate MNOs, and adjacent MNOFig. 4. Cloud-based storage taxonomy.

104 N. Aminzadeh et al. / Simulation Modelling Practice and Theory 50 (2015) 96108authorized dealers. SAMIs Multi-tier infrastructure provides decreased communication latency and energy efficiency as aconsequence, to latency-sensitive applications by MNO authorized dealers. In case of demand for security-sensitiveservices, MNOs are beneficial by providing access to their resources via more secure cellular network in comparison toInternet channel. When MNOs resources are incapable of meeting data-intensive services demand or not accessible afteroperating hours, cloud infrastructures are exploited.Reducing latency, establishment of trust, alleviating portability, and preventing energy dissipation are among promisingadvantages of SAMI. Similar to UbiqStor [45], SAMI exploits a multi-tier architecture in its design. However, SAMI pro-vides higher trust to mobile service consumers by leveraging resources of MNOs beside cloud and adjacent MNO dealers.

Furthermore, Table 2 summarizes the comparison results of the reviewed approaches based on the defined categories inthe taxonomy presented in Fig. 1. Table 2 advocates differentiations between mobile and immobile storage augmentationapproaches. We deduce from Table 2 that the mobility factor plays an important role and affects other specified factors.Employing mobile cloud resources enhances energy efficiency, WAN latency, locality, and context awareness, while utilizingan immobile cloud back-end improves storage augmentation, data blocking, reliability, availability, data safety, security, andtrust.

4. Cloud-based storage characteristics

According to existing MSA approaches including the reviewed solutions in the previous section, we have devised acloud-based storage taxonomy. This classification is based on the prominent common characteristic and major differencesbetween various cloud storage resources including the architecture, capacity, tiering, mobility, locality, security, back-endconnectivity, and utilization cost factors. Fig. 4 depicts the cloud-based storage taxonomy and will be explained in thissection.

Architecture: The variety in the cloud resources architecture arises from diverse roles of the cloud in augmentingresource-poor mobile devices. In a clientserver architecture, a multitude of rich stationary servers owned by vendorsand enterprises serve mobile devices as their clients to enhance their storage capacity. Public and private cloud resourcesfall into this category. Mobile Network Operators (MNO), stationary computers in public places as well as authorizedMNO dealers, computing kiosks, and mobile devices are among other resources that have a server role and offers storageservices to storage-constrained mobile devices [16]. Amazon S3,1 Amazon EBS,2 HP Cloud Object Storage,3 and HP CloudBlock Storage4 are well-known instances of storage services offered via the public cloud.In peer-to-peer architecture, mobile devices are both a provider and consumer of services in a decentralized autonomousnetwork of proximate non-stationary devices that form a mobile storage service. Conferences, coffee shops, and universitycampuses are scenarios in which ad hoc networks can be formed among participants that generally share common interests.In this architecture, cloud storage services are exploited locally from the network of available mobile devices in the vicinityfor a period of time [35,38]. The storage capacity provided in a peer-to-peer architecture is incapable of competing with infi-nite storage that is accessible via the public cloud in a clientserver architecture. Moreover, a peer-to-peer architectureleverages a decentralized management, on the contrary to a clientserver architecture with centralized management. How-ever, the exploitation of edge devices storage in a peer-to-peer network alleviates long WAN latency and device energy con-sumption. Moreover, caching information of same interest on participants devices is another advantage of this approach,which lessens the traffic over a cellular network. Nonetheless, a guarantee of the persistency and redundancy of data is achallenging task in this architecture due to node mobility and a participants disconnection, which may cause data loss whenthe participant leaves the network or powers off the device [46]. The researchers efforts are toward the application of power-aware, real-time-aware, or partition-aware [38,47] approaches in wireless and ad hoc networks to conserve mobile devicesenergy, to minimize communication volume, and to maintain the availability of data in a well-timed manner.The hybrid architecture includes approaches that exploit nearby mobile devices that have a reduced latency as well as richservers supplying services in public and private clouds and that have proximate computers in the vicinity. Nevertheless, theadvantages of unleashing hybrid architectures necessitate novel implementation, administration, and resource schedulingprocedures that can lead to new challenges in MCC realm.Capacity: Storage is one of the main services provided by the cloud as Storage as a Service (SaaS). The unlimited storagecapacity of the cloud can augment resource-constrained mobile devices. SaaS capacitates keeping data and applicationson the cloud resources and access them remotely. These unified, elastic, reliable, secure resources are usually located farfrom mobile nodes. Hence, mobile devices require crossing an Internet channel via a wireless network to access the1 http://aws.amazon.com/.2 http://aws.amazon.com/ebs/.3 http://docs.hpcloud.com/object-storage/.4 https://docs.hpcloud.com/block-storage/.

http://aws.amazon.com/http://aws.amazon.com/ebs/http://docs.hpcloud.com/object-storage/http://https://docs.hpcloud.com/block-storage/

N. Aminzadeh et al. / Simulation Modelling Practice and Theory 50 (2015) 96108 105unlimited storage of a distant public cloud. Dropbox,5 Google Docs,6 Amazon S3, iCloud,7 and SkyDrive8 are among therenowned cloud storage services that facilitate the storage, sharing, and synchronization of data for desktop and mobiledevice users. Several efforts such as UbiqStor [45] and MiSC [48] tried augmenting mobile devices using remote storage serv-ers. The utilization of non-proximate cloud storage services over an Internet channel imposes increased WAN latency anddecreased security. An alternative solution is the use of constrained transient storage capacity via nearby immobile cloudresources [49] and mobile devices [35]. Although leveraging this approach supplies limited storage to lightweight mobiledevices in comparison to infinite storage capacity of the cloud, the reduced latency over cellular networks and Wi-Fienhances performance and addresses mobile devices energy deficiency problem.Tiering: Tiering refers to the number of layers a mobile device passes over to exploit cloud services. Approaches such as[45,15] are referred to as multi-tier. For example, the three-tier infrastructure of SAMI [15] encompasses distant cloudresources, proximate MNOs, and MNO-authorized dealers in the vicinity. Conversely, the one-tier category approachesleverage one of the cloud resources for storage augmentation.Multi-tier infrastructure provides rich computing beside energy- and latency-aware communication. It is realizedthrough the provision of several options for execution of services. While proximate cloud resources tier covers therequirements of latency-sensitive services, the choice of exploiting reliable and well known organizations resources tieris beneficial to the services with security and trust requirements. Service providers in the telecommunication are trustfuloptions for the provision of resources over the secure cellular network as a cloud. The public clouds as the equipped tierwith elastic, infinite storage and computing resources are the choice for data-intensive services when resources of trust-worthy, proximate tiers are insufficient.Mobility: Cloud resources are either mobile or immobile with respect to the mobility characteristic. When immobile,cloud resources are delineated by a richer source of computation and storage capacity and no energy shortcomings com-pared with mobile resources. Public and private cloud servers, MNOs, and fine-grained resources such as available desk-top computers in coffee shops and airports and cloudlets are instances of immobile storage resources. A cloudlet is aproximate immobile cloud solution that concentrates on the computational augmentation of non-stationary neighboringdevices. Similarly, it can be applied for storage enhancement in which the VM-based offloading method of the cloudlet isemployed for more storage capacity to proximate mobile devices. On the contrary, mobile resources provide context-awareness and lower latency, which can adversely impact the energy consumption and transmission period.Location: Cloud storage resources are either located in the proximity or are distant. Approaches utilizing a multi-tierinfrastructure can benefit from both and are referred to as hybrid resources. Distant resources, particularly public cloudstorage, are accessible via an Internet channel over a wireless medium. Long WAN latency and low security are commonchallenges in wireless networks and are intensified through an Internet channel. However, proximate cloud resources areaccessible via both Ethernet and Internet channel. Utilizing nearby mobile or fixed resources reduces latency, cost, andcommunication overheads [50]. Stationary computers in public places and nearby mobile devices specifically smart-phones, laptops, and tablets are resources that can serve client mobile nodes in the vicinity.Back-end Connectivity: This category considers the connectivity of back-end storage resources to the Internet. They areeither connected to a wired Internet connection or utilize a wireless medium. Mobile devices that play the role of serversto other mobile nodes or participate in a peer-to-peer architecture are supported via a wireless connection to the Internet.On the contrary, stationary servers, whether public or private cloud resources, or immobile computers in the neighbor-hood are equipped with a wired Internet connection.

5. Open issues

In this section, we present crucial open issues on MSA in MCC as several thought-provoking future research directions.5.1. Energy-awareness

Energy conservation is one of the ultimate goals in mobile computing, cloud computing, and MCC domains. Although lim-itation of resources, processing, storage, and battery life of mobile devices can be alleviated through rich cloud-basedresources [16], high pre- and post-offloading overheads of in/less-efficiently designed MCC solutions can remarkably inten-sify the energy limitations of mobile devices. Therefore, energy-awareness using lightweight mobile augmentation frame-works [51], communication-aware approaches [52], and fidelity adaptation [53] solutions are essential to meet theenergy conservation in mobile devices. Lightweight solutions could mitigate the offloading overheads (processing andlatency) [54] and ensure that the communication cost would not exceed the data computation benefits of mobile devices.Similarly, optimal communication-aware and fidelity adaptation solutions alleviate the energy efficiency issue by leveragingheterogeneous wireless networks (i.e., cellular and WLAN) for an energy-efficient trade-off for lightweight MCC solutions.5 https://www.dropbox.com.6 http://docs.google.com.7 https://www.icloud.com.8 www.skydrive.com.

http://https://www.dropbox.comhttp://docs.google.comhttp://https://www.icloud.comhttp://www.skydrive.com

106 N. Aminzadeh et al. / Simulation Modelling Practice and Theory 50 (2015) 961085.2. Data integrity

Although the anywhere, anytime, any device principal of MCC is beneficial to mobile device users, the variation and inho-mogeneity of storage services provided by varied cloud vendors cause data integrity issue and necessitates additional inves-tigation of MCC storage augmentation approaches. Data integrity issue jeopardizes the reliability of cloud service providersand rises data retrieval issue as a consequence. To mitigate data integrity issue, efforts similar to [13] seek to provide a uni-form interface to access various cloud storage services by exploiting middleware, SOA, and domain specific languageapproaches. Moreover, management of data, which are distributed in various cloud storage services [30] is promising toovercome data integrity issues in MCC. However, further research is required for a unified, standard storage infrastructureand management of heterogeneity to alleviate data migration.

5.3. Trust

One of the most important concerns of the mobile end-users is to establish trust for the largely growing cloud vendors.Although several studies have been conducted for building trust in cloud resources [15,5557], the transmission of user dataover insecure wireless networks (where stern supervision is lacking) and the Internet for storage of data in the cloud (whereusers do not have any control) is inevitable. Heterogeneity of cloud infrastructures as well as mobile devices resource pov-erty and intrinsic traits intensify trust issue. Protection of user data by leveraging encryption and decryption techniques [34],authorization, and authentication are among the approaches try to overcome trust issue in MCC. However, a necessity fornovel trust establishment techniques similar to [58] exists which are required to be lightweight with respect to limitedresources of mobile devices and wireless communication in MCC domain.

5.4. Data portability

Harnessing cloud resources as a comparatively more reliable and safe storage for data has alleviated the data lock-inproblem of mobile device users. Hence, the migration of data for messages and contact information between non-uniformmobile devices is facilitated. However, data portability is still an issue for cloud consumers and due to this problem, usersare unable to transfer their data from Android-based to iOS-based mobile devices. Moreover, the heterogeneity of cloud ser-vices offered by diverse cloud vendors causes a vendor lock-in problem. Vendor lock-in is an attractive issue in business [59],but a real concern for customers. Vendor lock-in makes customers dependent on services provided by specific cloud vendors,as the incompatibility with other services or the consequence of cloud service providers policies.

Homogeneity between the existing cloud services architectures and programming languages is desired by end-users toovercome code and data portability concerns. Solutions including adapter and particularly middleware [60] as well as stan-dardization such as Open Cloud Computing Interface (OCCI)9 are among approaches addressing portability issues in cloudcomputing. MCC data portability issues grant for further research that considers variant clouds policies, architectures, and stor-age structures to alleviate imposed challenges on the migration of data between dissimilar cloud storages and mobile devices.

6. Conclusions

This paper surveys the crucial intrinsic restrictions of mobile devices and storage augmentation issues in three domains ofmobile computing, cloud computing andMCC to devise a taxonomy of issues as the motivation for the emergence of effectiveand efficient MSA approaches in MCC. A number of approaches leverage data partitioning whereas other approaches exploitdata replication, cache management, or SOA. Based on a review of the credible MSA approaches, the paper proposes a tax-onomy of cloud-based storage.

A mobile devices local resource conservation, computational augmentation, and storage extension have been realizedthrough the convergence of mobile and cloud computing that has produced state-of-the-art MCC. This association and allits inestimable privileges originate multi-dimensional heterogeneity in various domains, including platforms, operating sys-tems, networks, and data structures. Pivotal sources of variations produce the generation of several non-trivial challenges formobile data, including portability, interoperability, integrity, and security. Therefore, the need for lightweight, energy- andcommunication-aware MSA approaches is vital for the successful adoption of mobile computing. Mobility is one of the prom-inent factors affecting other factors in taxonomizing cloud-based storage resources in a positive or negative manner. Hence,additional efforts are required to address a number of crucial MCC storage augmentation open issues. Energy-awareness,data integrity, trust, and data portability are open issues, which specify future research directions in this area.

Acknowledgments

This work is funded by the University of Malaya Research Grant RP005D13ICT and by the Malaysian Ministry of HigherEducation under the University of Malaya High Impact Research Grant UM.C/625/1/HIR/MOE/FCSIT/03.9 http://occi-wg.org/.

http://occi-wg.org/

N. Aminzadeh et al. / Simulation Modelling Practice and Theory 50 (2015) 96108 107References

[1] W. Rupert, Mobile Data will Grow 6.3 Times between 2013 and 2018 and Growth will Be Strongest Outside Europe and North America. .

[2] J. Wan, C. Zou, C.-F. Lai, M. Zhou, X. Wang, Cloud-enabled wireless body area networks for pervasive healthcare, IEEE Network 27 (5) (2013) 5661.[3] H.-Y. Kung, C.-H. Chen, H.-H. Ku, Designing intelligent disaster prediction models and systems for debris-flow disasters in Taiwan, Expert Syst. Appl. 39

(5) (2012) 58385856.[4] R. Ferzli, I. Khalife, Mobile cloud computing educational tool for image/video processing algorithms, in: IEEE Digital Signal Processing and Signal

Processing Education Meeting (DSP/SPE), 2011, pp. 529533.[5] M. Chen, S. Mao, Y. Liu, Big data: a survey, Mobile Networks Appl. 19 (2) (2014) 171209.[6] S. Abolfazli, Z. Sanaei, A. Gani, F. Xia, L.T. Yang, Rich mobile applications: genesis, taxonomy, and open issues, J. Network Comput. Appl. 40 (2013) 345

362.[7] S.S. Ali Feizollah, Nor Badrul Anuar, Rosli Salleh, Fairuz Amalina, Rauf Ridzuan Maarof, A study of machine learning classifiers for anomaly-based

mobile botnet detection, Malaysian J. Comput. Sci. 26 (4) (2014) 251265.[8] N.B. Anuar, M. Papadaki, S. Furnell, N. Clarke, Incident prioritisation using analytic hierarchy process (AHP): Risk Index Model (RIM), Secur. Commun.

Networks 6 (9) (2013) 10871116.[9] S. Sakr, A. Liu, D.M. Batista, M. Alomari, A survey of large scale data management approaches in cloud environments, IEEE Commun. Surv. Tutorials 13

(3) (2011) 311336.[10] A. Ranabahu, A. Sheth, Semantics centric solutions for application and data portability in cloud computing, in: IEEE Second International Conference on

Cloud Computing Technology and Science(CloudCom), Dayton, OH, USA, 2010, pp. 234241.[11] A. Sheth, A. Ranabahu, Semantic modeling for cloud computing, part 1, IEEE Internet Comput. 14 (3) (2010) 8183.[12] M. Shiraz, S. Abolfazli, Z. Sanaei, A. Gani, A study on virtual machine deployment for application outsourcing in mobile cloud computing, J.

Supercomput. 63 (3) (2012) 946964.[13] H. Mao, N. Xiao, W. Shi, Y. Lu, Wukong: a cloud-oriented file service for mobile Internet devices, J. Parallel Distrib. Comput. 72 (2) (2012) 171184.[14] Z. Sanaei, S. Abolfazli, A. Gani, R.H. Khokhar, Tripod of requirements in horizontal heterogeneous mobile cloud computing, in: WSEAS International

Conference on Computing, Information Systems and Communications (CISCO12), Singapore, 2012.[15] Z. Sanaei, S. Abolfazli, A. Gani, M. Shiraz, SAMI: service-based arbitrated multi-tier infrastructure for mobile cloud computing, in: IEEE International

Conference on Communications in China Workshops (ICCC), Beijing, China, 2012, pp. 1419.[16] S. Abolfazli, Z. Sanaei, E. Ahmed, A. Gani, R. Buyya, Cloud-based augmentation for mobile devices: motivation, taxonomies, and open challenges, IEEE

Commun. Surv. Tutorials 16 (1) (2013) 337368.[17] S. Abolfazli, Z. Sanaei, A. Gani, Mobile cloud computing: a review on smartphone augmentation approaches, in: WSEAS International Conference on

Computing, Information Systems and Communications (CISCO12), Singapore, 2012.[18] M. Shiraz, A. Gani, R.H. Khokhar, R. Buyya, A review on distributed application processing frameworks in smart mobile devices for mobile cloud

computing, IEEE Commun. Surv. Tutorials 15 (3) (2013) 12941313.[19] Z. Sanaei, S. Abolfazli, A. Gani, R. Buyya, Heterogeneity in mobile cloud computing: taxonomy and open challenges, IEEE Commun. Surv. Tutorials 16

(1) (2013) 369392.[20] Gartner Reveals Top Predictions for IT Organizations and Users for 2012 and Beyond, 2011. .[21] W. Zheng, P. Xu, X. Huang, N. Wu, Design a cloud storage platform for pervasive computing environments, Cluster Comput. 13 (2) (2010) 141151.[22] P. Shor, Algorithms for quantum computation: discrete logarithms and factoring, in: Proceedings IEEE 35th Annual Symposium on Foundations of

Computer Science, 1994, pp. 124134.[23] P.W. Shor, Polynomial-time algorithms for prime factorization and discrete logarithms on a quantum computer, SIAM J. Comput. 26 (5) (1997) 1484

1509.[24] C. Metz, IBM Busts Record for Superconducting Quantum Computer, 2012. .[25] J.F.R. Darin Chan, Summarisation for mobile databases, J. Res. Pract. Inf. Technol. 37 (3) (2005) 267284.[26] Y. Jun, L. Shan, X. Lizhen, Research of cache mechanism in mobile data management, in: Proceedings IEEE Ninth Web Information Systems and

Applications Conference, 2012, pp. 159162.[27] H. Hpfner, E. Mansour, D. Nicklas, Review of data management mechanisms on mobile devices, Inf. Technol. 51 (2) (2009) 7984.[28] A. Vancea, B. Stiller, CoopSC: a cooperative database caching architecture, in: 19th IEEE International Workshops on Enabling Technologies:

Infrastructures for Collaborative Enterprises, Larissa, 2010, pp. 223228.[29] K. Goda, M. Kitsuregawa, The history of storage systems, Proc. IEEE 100 (Special Centennial Issue) (2012) 14331440.[30] R. Buyya, C.S. Yeo, S. Venugopal, J. Broberg, I. Brandic, Cloud computing and emerging IT platforms: vision, hype, and reality for delivering computing as

the 5th utility, Future Gener. Comput. Syst. 25 (6) (2009) 599616.[31] P. Mell, T. Grance, The NIST Definition of Cloud Computing, 2011. .[32] N. Cordeschi, M. Shojafar, E. Baccarelli, Energy-saving self-configuring networked data centers, Comput. Networks 57 (17) (2013) 34793491.[33] M. Armbrust, A.D. Joseph, R.H. Katz, D.A. Patterson, Above the Clouds: A Berkeley View of Cloud Computing, Tech. Rep. UCB/EECS-2009-28, EECS

Department, University of California, Berkeley, 2009.[34] Zhibin Zhou, Dijiang Huang, Efficient and secure data storage operations for mobile cloud computing, in: Network and Service Management (CNSM),

2012 8th International Conference and 2012 Workshop on Systems Virtualiztion Management (SVM), Las Vegas, NV, 2012, pp. 3745.[35] R.K. Panta, R. Jana, F. Cheng, Y.-F.R. Chen, V.A. Vaishampayan, Phoenix: storage using an autonomous mobile infrastructure, IEEE Trans. Parallel Distrib.

Syst. 24 (9) (2013) 18631873.[36] P. Stuedi, I. Mohomed, D. Terry, WhereStore: location-based data storage for mobile devices interacting with the cloud, in: MCS 10 Proceedings of the

1st ACM Workshop on Mobile Cloud Computing & Services: Social Networks and Beyond, New York, NY, USA, 2010.[37] B. Jedari, F. Xia, A survey on routing and data dissemination in opportunistic mobile social networks, IEEE Commun. Surv. Tutorials (99). doi: 10.1109/

SURV.2014.022714.00153.[38] A. Moon, H. Cho, Energy efficient replication extended database state machine in mobile ad hoc network, in: IADIS International Conference on Applied

Computing, LISBON, PORTUGAL, 2004.[39] E.R. Sanchez, L.M. Murillo, B. Montrucchio, M. Rebaudengo, An adaptive power-aware multi-hop routing algorithm for wireless sensor networks, in:

Eighth IEEE International Conference on Information Technology: New Generations (ITNG), 2011, pp. 112116.[40] Jun-Zhao Sun, Mobile ad hoc networking: an essential technology for pervasive computing, in: IEEE International Conferences on Info-Tech and Info-

Net. Proceedings, vol. 3, Beijing, China, 2001, pp. 316321.[41] F. Pedone, R. Guerraoui, The database state machine approach, Distrib. Parallel Databases 14 (1) (2003) 7198.[42] Y. Lu, B. Zhou, L.-C. Tung, M. Gerla, A. Ramesh, L. Nagaraja, Energy-efficient content retrieval in mobile cloud, in: Proceedings of the Second ACM

SIGCOMM Workshop on Mobile Cloud Computing MCC13, New York, USA, 2013.[43] S. Abolfazli, Z. Sanaei, M. Shiraz, A. Gani, MOMCC: market-oriented architecture for mobile cloud computing based on service oriented architecture, in:

1st IEEE International Conference on Communications in China Workshops (ICCC), 2012, pp. 813.[44] M. Asif, S. Majumdar, R. Dragnea, Hosting web services on resource constrained devices, in: IEEE International Conference on Web Services (ICWS

2007), 2007, pp. 583590.[45] M. Ok, D. Kim, M.-s. Park, UbiqStor: a remote storage service for mobile devices, Parallel Distrib. Comput.: Appl. Technol. 3320 (2005) 685688.

http://www.analysysmason.com/About-Us/News/Insight/Mobile-data-Oct2013/http://www.analysysmason.com/About-Us/News/Insight/Mobile-data-Oct2013/http://refhub.elsevier.com/S1569-190X(14)00079-3/h0130http://refhub.elsevier.com/S1569-190X(14)00079-3/h0135http://refhub.elsevier.com/S1569-190X(14)00079-3/h0135http://refhub.elsevier.com/S1569-190X(14)00079-3/h0140http://refhub.elsevier.com/S1569-190X(14)00079-3/h0145http://refhub.elsevier.com/S1569-190X(14)00079-3/h0145http://refhub.elsevier.com/S1569-190X(14)00079-3/h0150http://refhub.elsevier.com/S1569-190X(14)00079-3/h0150http://refhub.elsevier.com/S1569-190X(14)00079-3/h0155http://refhub.elsevier.com/S1569-190X(14)00079-3/h0155http://refhub.elsevier.com/S1569-190X(14)00079-3/h0160http://refhub.elsevier.com/S1569-190X(14)00079-3/h0160http://refhub.elsevier.com/S1569-190X(14)00079-3/h0165http://refhub.elsevier.com/S1569-190X(14)00079-3/h0170http://refhub.elsevier.com/S1569-190X(14)00079-3/h0170http://refhub.elsevier.com/S1569-190X(14)00079-3/h0175http://refhub.elsevier.com/S1569-190X(14)00079-3/h0180http://refhub.elsevier.com/S1569-190X(14)00079-3/h0180http://refhub.elsevier.com/S1569-190X(14)00079-3/h0185http://refhub.elsevier.com/S1569-190X(14)00079-3/h0185http://refhub.elsevier.com/S1569-190X(14)00079-3/h0190http://refhub.elsevier.com/S1569-190X(14)00079-3/h0190http://www.gartner.com/newsroom/id/1862714http://refhub.elsevier.com/S1569-190X(14)00079-3/h0195http://refhub.elsevier.com/S1569-190X(14)00079-3/h0200http://refhub.elsevier.com/S1569-190X(14)00079-3/h0200http://www.wired.com/wiredenterprise/2012/02/ibm-quantum-milestone/http://www.wired.com/wiredenterprise/2012/02/ibm-quantum-milestone/http://refhub.elsevier.com/S1569-190X(14)00079-3/h0205http://refhub.elsevier.com/S1569-190X(14)00079-3/h0210http://refhub.elsevier.com/S1569-190X(14)00079-3/h0215http://refhub.elsevier.com/S1569-190X(14)00079-3/h0220http://refhub.elsevier.com/S1569-190X(14)00079-3/h0220http://csrc.nist.gov/publications/nistpubs/800-145/SP800-145.pdfhttp://refhub.elsevier.com/S1569-190X(14)00079-3/h0225http://refhub.elsevier.com/S1569-190X(14)00079-3/h0230http://refhub.elsevier.com/S1569-190X(14)00079-3/h0230http://refhub.elsevier.com/S1569-190X(14)00079-3/h0235http://refhub.elsevier.com/S1569-190X(14)00079-3/h0240

108 N. Aminzadeh et al. / Simulation Modelling Practice and Theory 50 (2015) 96108[46] P. Padmanabhan, L. Gruenwald, A. Vallur, M. Atiquzzaman, A survey of data replication techniques for mobile ad hoc network databases, VLDB J. 17 (5)(2008) 11431164.

[47] K. Wang, B. Li, Efficient and guaranteed service coverage in partitionable mobile ad-hoc networks, in: Proceedings of IEEE INFOCOM, vol. 2, 2002, pp.10891098.

[48] J.-H. Kim, B.-S. Moon, M.-S. Park, MiSC: a new availability remote storage system for mobile appliance, Lecture Notes in Computer Science, vol. 3421,Springer, Berlin, Heidelberg, 2005, pp. 504520.

[49] M. Satyanarayanan, P. Bahl, R. Caceres, N. Davies, The case for VM-based cloudlets in mobile computing, IEEE Pervasive Comput. 8 (4) (2009) 1423.[50] S. Abolfazli, Z. Sanaei, M. Alizadeh, A. Gani, F. Xia, An experimental analysis on cloud-based mobile augmentation in mobile cloud computing, IEEE

Trans. Consumer Electron. 60 (1) (2014) 146154.[51] M. Shiraz, A. Gani, A lightweight active service migration framework for computational offloading in mobile cloud computing, J. Supercomput. 68 (2)

(2014) 978995.[52] K. Kumar, Cloud computing for mobile users: can offloading computation save energy?, Computer 43 (4) (2010) 5156[53] R. Balan, Powerful change part 2: reducing the power demands of mobile devices, IEEE Pervasive Comput. 3 (2) (2004) 7173.[54] M. Sharifi, S. Kafaie, O. Kashefi, A survey and taxonomy of cyber foraging of mobile devices, IEEE Commun. Surv. Tutorials 14 (4) (2012) 12321243.[55] M. Mowbray, S. Pearson, A client-based privacy manager for cloud computing, in: Proceedings ACM of the Fourth International ICST Conference on

COMmunication System softWAre and middlewaRE COMSWARE 09, New York, USA, 2009.[56] Q. Wang, C. Wang, J. Li, K. Ren, W. Lou, Enabling public verifiability and data dynamics for storage security in cloud computing, in: ESORICS09

Proceedings of the 14th European Conference on Research in Computer Security, Saint Malo, France, 2009, pp. 355370.[57] S. Shamshirband, N.B. Anuar, M.L.M. Kiah, A. Patel, An appraisal and design of a multi-agent system based cooperative wireless intrusion detection

computational intelligence technique, Eng. Appl. Artif. Intell. 26 (9) (2013) 21052127.[58] W. Wang, Z. Li, R. Owens, B. Bhargava, Secure and efficient access to outsourced data, in: Proceedings of the 2009 ACM Workshop on Cloud Computing

Security CCSW 09, ACM Press, New York, USA, 2009, p. 55.[59] S.S. Manvi, G. Krishna Shyam, Resource management for Infrastructure as a Service (IaaS) in cloud computing: a survey, J. Network Comput. Appl. 41

(2014) 424440.[60] G. Blair, P. Grace, Emergent middleware: tackling the interoperability problem, IEEE Internet Comput. 16 (1) (2012) 7882.

http://refhub.elsevier.com/S1569-190X(14)00079-3/h0245http://refhub.elsevier.com/S1569-190X(14)00079-3/h0245http://refhub.elsevier.com/S1569-190X(14)00079-3/h0250http://refhub.elsevier.com/S1569-190X(14)00079-3/h0250http://refhub.elsevier.com/S1569-190X(14)00079-3/h0250http://refhub.elsevier.com/S1569-190X(14)00079-3/h0255http://refhub.elsevier.com/S1569-190X(14)00079-3/h0260http://refhub.elsevier.com/S1569-190X(14)00079-3/h0260http://refhub.elsevier.com/S1569-190X(14)00079-3/h0265http://refhub.elsevier.com/S1569-190X(14)00079-3/h0265http://refhub.elsevier.com/S1569-190X(14)00079-3/h0270http://refhub.elsevier.com/S1569-190X(14)00079-3/h0275http://refhub.elsevier.com/S1569-190X(14)00079-3/h0280http://refhub.elsevier.com/S1569-190X(14)00079-3/h0285http://refhub.elsevier.com/S1569-190X(14)00079-3/h0285http://refhub.elsevier.com/S1569-190X(14)00079-3/h0290http://refhub.elsevier.com/S1569-190X(14)00079-3/h0290http://refhub.elsevier.com/S1569-190X(14)00079-3/h0290http://refhub.elsevier.com/S1569-190X(14)00079-3/h0295http://refhub.elsevier.com/S1569-190X(14)00079-3/h0295http://refhub.elsevier.com/S1569-190X(14)00079-3/h0300

Mobile storage augmentation in mobile cloud computing: Taxonomy, approaches, and open issues1 Introduction2 Motivation2.1 Mobile device issues2.2 Cloud-based issues2.3 Converged issues

3 Mobile-cloud storage augmentation approaches4 Cloud-based storage characteristics5 Open issues5.1 Energy-awareness5.2 Data integrity5.3 Trust5.4 Data portability

6 ConclusionsAcknowledgmentsReferences