PaperlayerView exact revision on arXiv

Source-linked AI summary

Cloud-Based Augmentation for Mobile Devices: Motivation, Taxonomies, and Open Challenges

Saeid Abolfazli, Zohreh Sanaei, Ejaz Ahmed, Abdullah Gani, Rajkumar Buyya

arXiv:1306.4956v1cs.DC

TL;DR

Mobile devices have limited processing, memory, energy, storage, and communication capabilities, while existing studies insufficiently explain how remote-resource characteristics affect augmentation. The paper surveys mobile augmentation, defines and taxonomizes CMA resources and approaches, and analyzes decision and performance factors. It concludes that CMA can enhance mobile processing, energy, storage, security, ubiquity, accessibility, and user-interface capabilities, while heterogeneity and communication constraints remain important limitations.

  • Problem

    Existing studies marginally discuss how remote-resource performance, distance, elasticity, availability, reliability, and multi-tenancy affect mobile augmentation.

  • Method

    The paper surveys mobile augmentation and classifies cloud-based resources and CMA approaches into distant fixed, proximate fixed, proximate mobile, and hybrid groups.

  • Results

    CMA can enhance mobile processing, energy, storage, data safety and security, data ubiquity, accessibility, and user-interface capabilities.

  • Takeaways & Limitations

    CMA provides a cloud-based back-end model for extending mobile-device capabilities beyond intrinsic processing and storage limits.

  • Takeaways & Limitations

    CMA performance remains constrained by heterogeneous mobile, cloud, and networking environments and by unreliable or interrupted communication links.

Abstract

from arXiv · show

Recently, Cloud-based Mobile Augmentation (CMA) approaches have gained remarkable ground from academia and industry. CMA is the state-of-the-art mobile augmentation model that employs resource-rich clouds to increase, enhance, and optimize computing capabilities of mobile devices aiming at execution of resource-intensive mobile applications. Augmented mobile devices envision to perform extensive computations and to store big data beyond their intrinsic capabilities with least footprint and vulnerability. Researchers utilize varied cloud-based computing resources (e.g., distant clouds and nearby mobile nodes) to meet various computing requirements of mobile users. However, employing cloud-based computing resources is not a straightforward panacea. Comprehending critical factors that impact on augmentation process and optimum selection of cloud-based resource types are some challenges that hinder CMA adaptability. This paper comprehensively surveys the mobile augmentation domain and presents taxonomy of CMA approaches. The objectives of this study is to highlight the effects of remote resources on the quality and reliability of augmentation processes and discuss the challenges and opportunities of employing varied cloud-based resources in augmenting mobile devices. We present augmentation definition, motivation, and taxonomy of augmentation types, including traditional and cloud-based. We critically analyze the state-of-the-art CMA approaches and classify them into four groups of distant fixed, proximate fixed, proximate mobile, and hybrid to present a taxonomy. Vital decision making and performance limitation factors that influence on the adoption of CMA approaches are introduced and an exemplary decision making flowchart for future CMA approaches are presented. Impacts of CMA approaches on mobile computing is discussed and open challenges are presented as the future research directions.

I. INTRODUCTION

Cloud-based Mobile Augmentation (CMA) uses cloud-based resources to enhance mobile-device computing, but effective adoption depends on understanding resource impacts and selecting suitable resources. The paper surveys CMA efforts, classifies resource types and approaches, and identifies decision and performance challenges.

  • Motivation and context: Remote execution with nearby surrogates faces reliability, security, and elasticity challenges because services can terminate unexpectedly or expose data.Cloud resources emerged partly to mitigate these shortcomings.
  • Motivation and context: CMA executes resource-intensive mobile application components in resource-rich cloud-based resources to enhance mobile-device capabilities.Cloud-based resources include distant clouds and proximate mobile nodes.
  • Decision challenges: Heterogeneous resources require schedulers and analyzers to assess influential factors and allocate appropriate resources to tasks.Examples include user preferences and available native mobile resources.
  • Decision challenges: CMA performance is constrained by application complexity, unauthorized remote-data access, wireless-network limitations, and cloud-computing challenges.The paper also highlights the difficulty of accurately estimating communication and computation time and energy.
  • Research gap: Prior studies marginally discuss remote-resource performance, distance, elasticity, availability, reliability, and multi-tenancy, leaving their augmentation impacts insufficiently explained.The paper frames this gap as requiring further research.
  • Paper scope: The paper classifies CMA resources and approaches into distant fixed, proximate fixed, proximate mobile, and hybrid groups.The classification follows resource mobility and physical-location traits.

A. Definition

Mobile augmentation encompasses hardware and software approaches that increase, enhance, or optimize mobile-device computing capabilities. The paper distinguishes augmentation from earlier approaches and motivates it by persistent processing, memory, energy, and application-demand constraints.

  • Earlier approaches: Load sharing migrates an entire computation job for remote execution, while remote execution conserves mobile resources and battery through remote computation or storage.Cyber foraging extends remote execution by dynamically exploiting wired infrastructure.
  • Earlier approaches: Earlier approaches require intensive code and data to be identified and partitioned, imposing substantial overhead on resource-poor mobile devices.The offloading and decision process must be repeated for each mobile user running the intensive application.
  • Augmentation model: Recent augmentation stores intensive code and data outside the device while retaining the rest locally, mitigating partitioning and migration overhead and alleviating storage constraints.Publicly accessible storage can also increase component reusability and support multiple users.
  • Definition: Mobile computation augmentation is a broad process of increasing, enhancing, and optimizing mobile-device capabilities through feasible hardware and software approaches.The term subsumes load sharing, remote execution, cyber foraging, and other augmentation approaches.
  • Motivation: Augmentation approaches support computing-intensive, data-intensive, and communication-intensive applications while increasing computing capabilities and conserving energy.Examples include speech recognition, enterprise applications, and online video streaming.
  • Motivation: Persistent mobile limitations include slow processing, limited RAM, short battery life, and constrained storage for growing applications and digital content.Memory leakage further intensifies mobile-memory restrictions.

2) Energy Resources:

Mobile augmentation offers alternatives to enlarging constrained smartphone hardware, including software techniques and cloud resources. These alternatives can extend processing, energy, storage, and presentation capabilities but remain subject to communication, security, portability, and design constraints.

  • Hardware and software approaches: Software techniques can improve visualization on small screens, but computation-intensive techniques may quickly drain batteries.Cloud-based computation is presented as an alternative for these presentation approaches.
  • Constraints: CMA approaches face security and privacy concerns, unreliable wireless connectivity, and heterogeneous platforms that hinder portability and continuous service.Mobility or disconnection can force local execution or re-initiation of augmentation.
  • Hardware and software approaches: Hardware augmentation is limited because larger processors, storage, screens, and batteries increase device heat, size, weight, and battery drainage.Smartphone hardware is not upgradable, making hardware-based augmentation slow and expensive.
  • Cloud resources: Cloud resources offer high computing power, elasticity, mobility support, low utilization overhead, and security compared with traditional surrogates.These differences support cloud infrastructure as a back-end technology for mobile augmentation.
  • Cloud resources: Remote storage expands mobile capacity and supports access to applications and data outside the device, but online connectivity is essential.Remote storage can also enable code updates without consuming mobile I/O resources.

III. IMPACTS OF CMA ON MOBILE COMPUTING

This section analyzes how CMA affects mobile computing and summarizes its advantages and disadvantages. It focuses on how cloud resources can mitigate mobile-computing deficiencies while introducing implementation and security concerns.

  • III. IMPACTS OF CMA ON MOBILE COMPUTING: The paper summarizes CMA’s advantages and disadvantages as mechanisms for addressing deficiencies in mobile computing.The discussion treats cloud resources broadly across the resource types examined in the paper.

A. Advantages

CMA augments mobile processing, storage, reliability, application capability, and battery life by shifting demanding workloads and data to cloud resources. However, device diversity and offloading constraints make flexible, low-overhead augmentation difficult.

  • 1) Empowered Processing:: CMA offloads CPU- and memory-intensive application components to cloud resources, allowing smartphones to deliver results beyond their native capabilities.This increases virtual transaction capacity and usable memory for resource-intensive applications.
  • 1) Empowered Processing:: Cloud resources provide higher availability and reliability than traditional surrogates, whose services may be arbitrarily interrupted or denied.
  • 1) Empowered Processing:: CMA enables computing- and memory-intensive applications, including content-based image retrieval, to run on smartphones with reduced concern for native resources.
  • 3) Expanded Storage:: A generic CMA approach remains needed to support heterogeneous devices with minimal configuration, processing overhead, and latency.The diversity arises from variation in smartphone and tablet hardware, platforms, APIs, features, and networks.
  • 2) Prolonged Battery:: Migrating heavy computation to the cloud can save mobile energy and extend battery life, although energy efficiency remains a major CMA challenge.
  • 3) Expanded Storage:: Cloud storage lets smartphones access extensive applications and data without repeatedly installing or removing content because of local space limits.Online connectivity is required, and cloud-stored applications can be updated without consuming mobile I/O transactions.

3) Expanded Storage:

Cloud-based augmentation expands mobile data access, sharing, protection, visualization, and cross-device application development. These benefits depend on connectivity, access control, and solutions to the broader complexity and security drawbacks of cloud services.

  • 5) Ubiquitous Data Access and Content Sharing:: Cloud storage improves data access by making digital content available anytime, anywhere, and from any device.This reduces the effects of temporal, geographical, and physical differences on access.
  • 5) Ubiquitous Data Access and Content Sharing:: Cloud storage supports authorized sharing and synchronizes data across heterogeneous devices with lower risk of corruption or integrity loss than direct device-to-device transfer.Reliable access control is required to manage permissions.
  • 6) Protected Offloaded Content:: Cloud storage and virtualization aim to protect remote code and data while preserving user privacy through guest isolation and vendor security policies.
  • 6) Protected Offloaded Content:: Cloud resources can perform intensive 2D or 3D rendering and stream screen images adapted to a smartphone’s size.Cloud-side processing can alter presentation techniques to match the device display.
  • 7) Application Development:: Cloud frameworks allow components to support distributed applications across dissimilar mobile devices and operating systems, reducing redevelopment time and cost.An Android application can be transitioned to iOS or Symbian by changing native components.

1) Dependency to High Performance Networking Infrastructure:

CMA systems depend on robust, high-performance networking, yet wireless limitations, congestion, security risks, development complexity, costs, and volatile cloud policies constrain deployment and service reliability.

  • Networking Requirements: CMA requires high-bandwidth, robust, reliable wireless communication and fast wired links to support intersystem communication and live migration.The paper identifies these networking requirements as necessary for computing anywhere and migrating heavy data and computations to nearby cloud resources.
  • Networking Requirements: 18) Mobile data operations such as storage, offloading, and VM migration can increase traffic, congestion, and packet loss over wireless networks.These operations distribute mobile data among helping nodes, making traffic management challenging.
  • Security and Privacy: Unauthorized access to remote cloud data threatens privacy, especially when sensitive code, financial data, or enterprise information is migrated.Enterprise users may face increased exposure of business data to competitors.
  • Development and Cost: Heterogeneous cloud environments increase application-development complexity and impose temporal and financial costs on developers.Developers must understand cloud operating systems, programming languages, and data structures across many mobile platforms.
  • Development and Cost: Cloud usage introduces SLA-based charges, including possible double charging for transferring data between cloud vendors.Users may choose local execution or application termination when infrastructure costs are unfavorable.
  • Provider Dependence: Nonnegotiable agreements, changing vendor policies, and absent external control make CMA service quality volatile.Cloud users cannot fully control provider performance or commitments over time.

IV. TAXONOMY OF CLOUD-BASED COMPUTING RESOURCES

The paper classifies cloud-based computing resources by distance and mobility into four groups, then compares their capabilities and limitations for CMA resource selection.

  • Resource Taxonomy: The taxonomy comprises distant immobile clouds, proximate immobile computing entities, proximate mobile computing entities, and hybrid resources.The classification is based on the distance and mobility traits of cloud-based resources and is intended to guide future infrastructure selection.
  • Distant Immobile Clouds: Distant stationary clouds provide highly available, scalable, and elastic resources but suffer performance effects from long WAN latency.Live VM migration can shorten the distance, but large VM sizes, unpredictable mobility, and intermittent wireless bandwidth make it non-trivial.
  • Distant Immobile Clouds: Virtualized clouds improve resource utilization by running several isolated VMs on one host, while VM hopping and VM escape threaten code and data security.VM hopping attacks co-resident VMs, whereas VM escape compromises the hypervisor and can control other VMs.
  • Proximate Immobile Computing Entities: Proximate immobile resources include underused public-place computers connected to power and wired Internet, but security and privacy weaknesses hinder their use.Owners and malicious users may access private data, falsify offloading results, or use these resources as attack points.
  • Proximate Mobile Computing Entities: Proximate mobile resources reduce client-server heterogeneity and support context-aware services, but limited device resources leave architecture, scheduling, and mobility unresolved.Mobile devices acting as servers are also vulnerable to loss, robbery, and security threats.

D. Hybrid (Converged Proximate and Distant Computing Entities)

Hybrid CMA combines proximate and distant resources to balance latency and computation power, but dynamic deployment, management, scheduling, and security trade-offs remain central concerns.

  • Hybrid Resource Allocation: Hybrid infrastructures combine proximate and distant mobile or immobile nodes to balance user requirements, especially latency and computation power.The approach assigns latency-sensitive code to nearby devices and intensive, less latency-sensitive tasks to farther resources.
  • Hybrid Management: Hybrid deployment, management, and resource scheduling are non-trivial in dynamic mobile environments.The paper proposes autonomic systems that automatically manage, optimize, and adapt hybrid infrastructures as a way to improve CMA quality.
  • Security and QoS: Hybrid clouds can improve CMA security, privacy, and QoS by trading among resources with different security, privacy, and trust characteristics.Security-sensitive computations may be executed in a secure distant cloud as part of a security-latency trade-off.
  • CMA Taxonomy: CMA is defined as executing resource-intensive mobile-application components in cloud-based resources to increase, enhance, and optimize device capabilities.The paper classifies state-of-the-art CMA approaches into distant fixed, proximate fixed, proximate mobile, and hybrid models.
  • Application Scenarios: Cloud resources are used in scenarios including live cloud streaming, enterprise applications, and social networking.The paper identifies CRM and enterprise resource planning as examples of enterprise use.
  • State-of-the-Art Approaches: CloneCloud can reduce execution time through thread-level migration, but synchronization and migration overhead can reduce cloud benefits for tightly coupled, data-intensive applications.Frequent encapsulation, migration, and mobile-cloud data synchronization increase communication demands.
  • State-of-the-Art Approaches: Representative approaches address screen rendering, portability, development complexity, storage, data locality, file access, and fine-grained application partitioning.Virtualized Screen moves rendering to the cloud; µCloud supports portability and offline usability; SmartBox expands storage; WhereStore caches location-relevant data; Wukong abstracts cloud services; CloneCloud migrates intensive threads without code changes.

B. Proximate Fixed

Proximate fixed CMA uses nearby stationary computers or cloudlets to augment mobile devices, while proximate mobile approaches harness nearby smartphones for distributed resource-intensive tasks.

  • Proximate Fixed: Cloudlet uses nearby resource-rich computers to offload intensive computation from mobile devices over Wi-Fi.The mobile device acts as a thin client, while computation is migrated to the nearby Cloudlet.
  • Proximate Fixed: Cloudlet execution restores the original host state after results are integrated, but deployment depends on available platforms and acceptable latency.The approach requires many powerful Cloudlets with popular mobile-platform base VMs, and few minutes of offloading latency is unacceptable to users.
  • Proximate Mobile: MOMCC harnesses nearby mobile devices through service-oriented building blocks, with providers registering resources and users locating secure hosts through a UDDI.Distant stationary clouds supplement nearby resources when local availability is insufficient.
  • Proximate Mobile: MOMCC creates a payment-based resource-sharing model in which participating smartphone owners can earn revenue for hosting services.The framework also faces challenges involving resource limitations, security, mobility, and business-plan design.
  • Proximate Mobile: Hyrax uses fault tolerance to mitigate frequent disconnections among nearby immobile smartphones, but its local-storage design limits interactive and data-intensive applications.Hyrax lacks cloud-storage connectivity and relies on constant historical multimedia data to avoid file sharing.
  • Proximate Mobile: Virtual Mobile Cloud Computing uses an ad-hoc cluster of nearby smartphones to reduce latency and network traffic while addressing hardware and platform heterogeneity.Required proxy-creation and RPC components are added to application code during the first execution for later offloading.

D. Hybrid

Hybrid CMA combines multiple resource types to improve augmentation flexibility and performance, while introducing coordination and decision-making challenges across heterogeneous infrastructures.

  • Hybrid: Hybrid CMA integrates different resource types to provide a smoother computing experience for mobile users.mCloud combines proximate immobile and distant stationary computing resources.
  • Hybrid: Hybrid offloading can forward a task from a nearby mobile resource to a distant cloud when local processing lacks sufficient resources.The foreign-tourist example sends an image to a nearby device first and forwards it to the cloud if processing fails.
  • Hybrid: SAMI proposes a three-level IaaS combining distant clouds, nearby MNOs, and nearby authorized dealers for computation and heavy-data storage.Its tiers use cellular, broadband, and Wi-Fi connections according to resource location and service needs.
  • Hybrid: SAMI aims to increase outsourcing flexibility, augmentation performance, and energy efficiency while supporting latency-sensitive services through nearby machines.Wi-Fi is used to conserve mobile energy relative to cellular networks.
  • Hybrid: MOCHA uses a mobile-cloudlet-cloud architecture for face recognition, combining nearby Cloudlets with distant immobile clouds.Its greedy task partitioning has significantly better response time than fixed partitioning when Cloudlets and heterogeneous-response-time clouds are available.
  • Hybrid: MOCHA requires prior knowledge of communication and computation latency for all available computing entities, making resource analysis time-consuming.This requirement applies to both Cloudlets and distant fixed clouds.

A. CMA Decision Making Factors

CMA decisions depend on mobile-device, content, environment, user, and cloud factors; these factors determine whether augmentation is feasible and which resources should be selected.

  • Decision factors: Decision systems categorize CMA factors into mobile devices, contents, augmentation environment, user preferences and requirements, and cloud servers.These factors support decisions about whether to perform augmentation and which resources to allocate.
  • Mobile devices: Native CPU, memory, storage, energy, mobility, and communication capabilities influence whether mobile devices should offload computation.The trade-off between energy consumed by augmentation and energy consumed by communication is especially important.
  • Contents: Code granularity, code size, data type, and data volume determine whether application content is suitable for local processing or augmentation.Latency-sensitive small data are efficient locally, while coarse-grained large code is infeasible to offload through low-bandwidth networks.
  • Augmentation environment: Wireless communication quality, bandwidth, congestion, utilization cost, and latency strongly affect CMA quality and constrain remote-server selection.Wireless communication is described as intermittent, unreliable, risky, and highly influential on augmentation performance.
  • Augmentation environment: Runtime dynamism increases decision complexity because movement, changing bandwidth, access-point distance, and resource availability alter augmentation conditions.Approaches should adapt rapidly to changes in the mobile environment.
  • User preferences and requirements: User preferences and requirements determine whether users permit remote execution and how CMA specifications should be customized.Users may differ in willingness to use Internet channels and in desired technical or non-technical settings.
  • Cloud servers: Cloud-resource performance, availability, elasticity, security, reliability, cost, and distance influence the credibility and performance of augmentation.Clouds offer elastic, pay-as-you-use resources that can reduce device ownership costs.

1) Heterogeneity:

Heterogeneity across mobile devices, networks, clouds, and data complicates CMA decisions and can degrade latency, service quality, and adoption when not managed effectively.

  • 1) Heterogeneity:: Network heterogeneity across Wi-Fi, 3G, and WiMAX makes mobility management and handover decisions central to CMA quality.Selecting an unsuitable network or transferring links at the wrong time increases WAN latency and jitter.
  • 1) Heterogeneity:: Growing digital-content volumes require partitioning and distributed storage, which complicate data integrity, consistency, management, and access control.Applying one access-control mechanism across diverse storage environments is also challenging.
  • 1) Heterogeneity:: Communication and computation latency are key performance metrics, particularly when mobile devices exploit distant cloud resources.Wireless distance, bandwidth, and speed variations affect offloading performance and application responsiveness.
  • 1) Heterogeneity:: Follow-Me Cloud supports endpoint mobility across IP subnets but introduces per-packet locator and identifier manipulation overhead.The overhead results from updating communication values as the node migrates.
  • 1) Heterogeneity:: Cloud computation latency affects application responsiveness and user-perceived performance in CMA systems.The cited research reports a 12X reduction in performance time violation.
  • 1) Heterogeneity:: Insufficient runtime knowledge degrades CMA performance because context information is large, dynamic, and difficult to store, manage, and process.Relevant context includes remote resources, network bandwidth, weather, and user voice or gestures.
  • 1) Heterogeneity:: CMA feasibility depends on user permission, resource availability, latency sensitivity, network conditions, and task characteristics.If users prohibit augmentation or delay-sensitive execution cannot meet requirements, processing may remain local or terminate.

VII. OPEN CHALLENGES

CMA deployment faces challenges spanning architectural fragmentation, resource-management complexity, mobility, and communication continuity. Future approaches require unified, adaptive mechanisms that coordinate heterogeneous resources while maintaining reliable service.

  • Open Challenges: CMA development is fragmented across cloud structures, techniques, and application types, while the absence of a unified architecture complicates deployment across mobile operating systems.A generic reference architecture is needed to support varied CMA families and multidimensional user needs.
  • Open Challenges: Heterogeneous communication technologies and infrastructures intensify management complexity, motivating autonomic mechanisms to coordinate distributed mobile and cloud resources.External management systems may further amplify complexity, so autonomy is presented as a feasible alternative.
  • Open Challenges: Continuous mobility remains difficult because providers and consumers can move during augmentation, causing redundant costs and reducing reliability.Web advancements and lightweight cognitive mobility management are suggested to reduce signaling traffic and latency.
  • Open Challenges: Task migration and result transfer incur substantial computation and communication costs, which mobility and wireless constraints further intensify.Resource discovery and allocation become especially challenging when using nearby computing devices.
  • Open Challenges: Intermittent wireless connectivity can force local execution or augmentation re-initiation when mobility breaks communication between consumers and providers.User mobility and wireless disconnection also reduce resource utilization.

F. Multipoint Data Bridging

CMA must bridge heterogeneous mobile and cloud systems while managing distributed data, security, virtualization, and mobility. These constraints make interoperability, content management, and low-overhead migration central challenges for dependable augmentation.

  • F. Multipoint Data Bridging: Multipoint data bridging requires offloading tasks, interoperating among varied clouds, and returning results across heterogeneous systems with different APIs and data structures.The goal is to access, interpret, process, share, and synchronize distributed contents in dynamic environments.
  • F. Multipoint Data Bridging: Distributed code and data facilitate heavy applications and large datasets, but managing the resulting heterogeneous content volume is difficult for smartphone users.Content-management demands grow as mobile users increasingly depend on cloud infrastructures.
  • F. Multipoint Data Bridging: Bandwidth limits, virtualization startup and migration, and secure VM management increase execution time and augmentation latency.Lightweight compression and paravirtualization are identified as possible ways to reduce data volume and virtualization overhead.
  • F. Multipoint Data Bridging: Offloading plain code or data over insecure wireless links and storing plain cloud data create major security and privacy concerns that impede CMA deployment.The paper calls for lightweight mobile-side security and robust cloud mechanisms.
  • F. Multipoint Data Bridging: Live VM migration must cope with bandwidth, intermittency, and mobility while reducing computation, energy, data-volume, and communication costs.Moving the VM with the user is needed to avoid increased access latency and degraded application performance.
  • F. Multipoint Data Bridging: Seamless access after VM migration requires preserving accessibility through the initial IP address when the VM changes physical machines.This supports uninterrupted access as the running VM moves between distributed servers.
  • F. Multipoint Data Bridging: Communication challenges including VM migration, mobility, handover, context awareness, location management, and unpredictable network conditions slow CMA adoption.Variable bandwidth, jitter, delay, throughput, and wireless-spectrum elasticity hinder upfront planning.
Loading 1306.4956v1…