PaperlayerView exact revision on arXiv

Source-linked AI summary

Internet of Things: An Overview

Farzad Khodadadi, Amir Vahid Dastjerdi, Rajkumar Buyya

arXiv:1703.06409v1cs.DCcs.NI

TL;DR

The chapter addresses how IoT can support ubiquitous, context-aware connectivity across heterogeneous devices and environments. It surveys architectures, security and privacy, communication means and protocols, and related integration approaches. It concludes with future directions and open challenges, including standardization, authorization, and communication protocols.

  • Problem

    IoT needs ubiquitous, context-aware platforms for interconnected, heterogeneous, and distributed devices, requiring integration from identification and communication to resource discovery and service integration.

  • Method

    The chapter surveys IoT architectures, security and privacy, network communication means and protocols, and approaches using service-oriented architectures and API definition languages.

  • Results

    The chapter highlights IoT research and advances while discussing cloud-, fog-, and mobile-computing solutions and API-oriented service use.

  • Takeaways & Limitations

    IoT development spans interconnected devices, people, and virtual environments, with practical attention to service integration and processing across constrained devices.

  • Takeaways & Limitations

    Standardization and regulatory policies constrain IoT development, while authorization and communication protocols remain open challenges.

Abstract

from arXiv · show

As technology proceeds and the number of smart devices continues to grow substantially, need for ubiquitous context-aware platforms that support interconnected, heterogeneous, and distributed network of devices has given rise to what is referred today as Internet-of-Things. However, paving the path for achieving aforementioned objectives and making the IoT paradigm more tangible requires integration and convergence of different knowledge and research domains, covering aspects from identification and communication to resource discovery and service integration. Through this chapter, we aim to highlight researches in topics including proposed architectures, security and privacy, network communication means and protocols, and eventually conclude by providing future directions and open challenges facing the IoT development.

1.1 Introduction

IoT seeks ubiquitous connectivity across heterogeneous devices, protocols, people, and computing resources. The chapter surveys its challenges and approaches, including cloud, fog, mobile computing, SDN, and containers.

  • 1.1 Introduction: IoT connects physical entities and virtual environments through seamless integration of people and devices.The entities include smart devices, sensors, humans, and context-aware communicating objects.
  • 1.1 Introduction: IoT requires ubiquitous connectivity across diverse devices and communication protocols, from tiny sensors to back-end servers.The ecosystem also integrates mobile devices, edge devices, and humans as controllers.
  • 1.1 Introduction: The chapter identifies scalability, heterogeneity support, total integration, and real-time query processing as underemphasized IoT requirements.It presents challenges and promising approaches based on recent IoT research and advances.
  • 1.1 Introduction: It discusses emerging IoT solutions based on cloud, fog, and mobile computing facilities.These solutions are presented alongside challenges and recent advances in the IoT ecosystem.
  • 1.1 Introduction: The chapter considers applying Software Defined Networking and containers to embedded and constrained IoT devices.These approaches are described as cutting-edge technologies being integrated with IoT.

1.2 Internet-of-Things Definition Evolution

IoT definitions evolve from connected objects toward ubiquitous, autonomous, smart, and service-exposing networks involving people, machines, and environments. The chapter also describes expanding markets, applications, investments, and remaining innovation needs.

  • Definition evolution: IoT definitions increasingly emphasize ubiquitous autonomous networks whose objects support identification and service integration.The discussion broadens IoT beyond basic connectivity and sensory requirements.
  • Definition evolution: IoT encompasses people, things, and places that can expose services to other entities through the Internet of Everything concept.Cisco uses IoE for this broader set of service-exposing entities.
  • Industrial IoT: Industrial IoT combines machine-to-machine communication with data analysis and machine learning for industrial tasks.Applications include supply-chain management, quality control, and reducing total energy consumption.
  • Smartness in IoT: IoT smartness is distinguished from sensor networks and divided into object smartness and network smartness.Smart networks are characterized by open standardized communication, addressable multifunctional objects, and reuse across applications.
  • Market share: $44.0 billion was the IoT market value in 2011, while cited forecasts estimated $498.92 billion for IoT and M2M by 2019 and $1423.09 billion for IoT by 2020.The same forecasts estimated Internet of Nano Things at approximately $9.69 billion by 2020.
  • Adoption and investment: IoT growth is accompanied by government investment, research projects, industrial initiatives, and cooperation among companies and smaller businesses.The chapter states that realizing the forecast market requires innovation and progress across different areas.

1.3 IoT Architectures

IoT architectures organize heterogeneous devices, services, communication layers, and user interfaces into interoperable systems that can adapt to mobility, failures, and changing conditions. The chapter presents an extended reference architecture emphasizing service management, discovery, APIs, security, privacy, and independent device management.

  • Architectural requirements: IoT connects heterogeneous objects, services, and people across physical and virtual realms through universally accessible communication.Interconnectivity and reliable operation are treated as critical architectural requirements.
  • Reference architectures: Reference architectures provide an abstract, bird’s-eye view that hides specific system constraints while organizing IoT components and interactions.Several research groups, including IoT-A and IoT-i, proposed architectures or strategic visions for integrating fragmented IoT sectors.
  • Extended reference architecture: The extended architecture places event processing, analytics, resource management, service discovery, message aggregation, ESB services, and API management above communication and physical layers.Web dashboards or smartphone applications provide access to and management of shared APIs.
  • Cross-layer concerns: Security, privacy, device management, object identification, and access control are treated as independent cross-layer components.The architecture separates these concerns because they are important across different layers and protocols.
  • SOA-based architecture: SOA supports IoT interoperability and scalability by using loosely coupled, reusable, composable services across heterogeneous devices and protocols.Its generic organization includes sensing, network, service, and interfaces layers, while component separation can preserve operation after failures.
  • Web-based services: Web APIs and REST-based methods are presented as alternatives to conventional SOAP approaches, while lightweight JSON exchanges can reduce communication and processing overhead on constrained devices.API-based service exposure also supports monitoring, pricing, multi-tenancy, and collaborative data sharing.

1.4 Resource Management

IoT resource management must discover, partition, provision, and schedule heterogeneous, dynamic resources while accounting for QoS, robustness, fault tolerance, scalability, energy, and service constraints. The section surveys containers, virtualization, code offloading, adaptive resource sharing, and semantic discovery as approaches to these challenges.

  • Resource partitioning: Resource management models IoT as a graph whose resources are partitioned to optimize utility criteria such as cost, energy, or performance before task scheduling.Figure 1.4 summarizes the resulting taxonomy of resource-management activities.
  • Resource-management requirements: Resource selection and provisioning directly affect IoT application QoS, while heterogeneity and dynamic resource conditions make management difficult.Large-scale deployments require robustness, fault tolerance, scalability, energy efficiency, QoS, and SLA considerations.
  • Containers and virtualization: Containers provide portable, platform-independent application environments and lightweight virtualization that can improve hardware utilization without expensive platform-specific requirements.Compared with virtual machines, containers require considerably less spin-up time and suit rapidly scaling distributed IoT applications.
  • Virtualization approaches: Virtualization techniques span Xen-, KVM-, microkernel-, and container-based approaches, with containers offering performance and security advantages through application sandboxing on a shared operating-system layer.Examples include Linux VServer, Linux Containers, and OpenVZ, while virtualized smartphones can run multiple Android systems on one device.
  • Resource-partitioning boundary: Task-grain scheduling across containers and virtualized environments challenges algorithms that treat these layers as black boxes.This issue arises because heterogeneous IoT devices may leverage virtualization to increase resource utilization.
  • Computation offloading: Code offloading addresses limited mobile-device resources by moving computation to other devices or clouds, improving power management, storage requirements, and application performance.Static approaches commonly require developers to manually annotate functions for remote execution, while dynamic parsing can adapt to network fluctuations and latency.
  • Semantic discovery: Semantic Web of Things approaches annotate resources and metadata with RDF and OWL, then support discovery and manipulation through SPARQL and related protocols.The section identifies scalable service discovery, composition, and integration as unresolved needs and describes discovery as locating devices and then finding target services.

1.5 IoT Data Management and Analytics

IoT data management must adapt acquisition, filtering, transmission, and analysis to distributed, heterogeneous, high-volume data. The section surveys cloud, edge, fog, and stream-processing approaches for scalable and low-latency analytics.

  • IoT data-processing procedures must be updated for distributed, heterogeneous environments characterized by velocity, volume, and variety.The required updates include data acquisition, filtering, transmission, and analysis.
  • Lambda Architecture combines batch, serving, and speed layers to support extensibility, scale-out, low-latency queries, and recent-data processing.The speed layer processes recent data for delay-sensitive queries, while batch and serving layers manage precomputed and dynamic queries.
  • Centralized data-mining algorithms are poorly suited to geographically distributed and heterogeneous IoT environments.Their centralized nature affects performance in these settings.
  • Cloud and edge resources support scalable, compute-intensive IoT analytics, while edge placement can reduce networking delays, costs, and sensitive-data exposure.Edge Cloud places resources near application consumers for aggregation and local processing.
  • Fog and edge computing extend cloud capabilities toward devices to address real-time processing, low latency, QoS, and SLA requirements.Edge devices such as smartphones, smart TVs, routers, and access points can contribute processing and storage.
  • Efficient real-time stream-processing engines require data fluidity, disorder handling, deterministic outcomes, integration of streaming and stored data, availability, auto-scaling, and partitioning.These requirements include failover and hot backup mechanisms.

1.6 Communication Protocols

IoT communication aggregates heterogeneous networks and protocols, making seamless connectivity possible but protocol interoperability and selection challenging. The section reviews protocol families, communication models, and deployment considerations.

  • IoT communication combines mobile networks, WLANs, wireless sensor networks, and mobile ad hoc networks.Seamless connectivity depends on communication speed, reliability, and connection durability.
  • Protocol interoperability remains challenging because IoT environments connect diverse communication systems and device-specific protocols.Segmentation and poor coherency can result from vendor-specific protocols and APIs.
  • Protocol selection must consider device specifications, future support, implementation ease, accessibility, security, performance, standardization, and documentation.Poor selection can create costly strategic mistakes, while insufficient documentation limits protocol usage.
  • Common IoT technologies include RFID, IEEE 802.11, IEEE 802.15.4, NFC, Bluetooth, 6LoWPAN, MQTT, and CoAP.These protocols span identification, local networking, low-power networking, and machine-to-machine communication.
  • IoT protocols include general-purpose protocols such as IP and SNMP and lightweight protocols such as CoAP for constrained devices.CoAP targets devices with tiny hardware and limited resources.
  • Publish/subscribe communication, used by MQTT, supports dynamic nodes through push notifications and queues for delayed message delivery.HTTP/REST and CoAP instead use request/response pulling to fetch messages.

1.7 Internet of Things Applications

IoT applications span household, industrial, civic, healthcare, and social domains. The section organizes enterprise applications by monitoring and actuation, business and data analysis, and information gathering and collaborative consumption.

  • IoT applications range from home automation to smart cities, e-government, logistics, transportation, supply chains, healthcare, and enterprise automation.The application domain is broad and requires seamless connectivity and addressability between components.
  • Enterprise IoT applications are categorized as monitoring and actuating, business process and data analysis, and information gathering and collaborative consumption.These categories organize applications by usage domain.
  • Smart homes and grids use connected devices to monitor conditions, control parameters, change configurations, and identify performance defects.Anomaly detection on collected data can support productivity improvements.
  • IoT adoption is discussed across society, industry, organizations, and individuals, with benefits including lower costs, transparency, daily-life improvements, and productivity growth.The cited classification distinguishes adoption levels by affected user or institution.
  • Healthcare IoT captures patient data through wearables and sensors, then applies data mining and machine learning to extract patterns and forecast risks.Remote systems can analyze household activity and health parameters to produce alerts that prevent incidents.
  • Privacy and security challenges remain major barriers to healthcare IoT adoption.These barriers affect both people and industries considering healthcare applications.

1.8 Security

IoT security is increasingly important as adoption expands, but constrained, distributed, heterogeneous, and physically accessible devices remain exposed to attacks. Resource limits also complicate cryptographic protection and system maintenance.

  • Increasing IoT adoption is accompanied by shifting attacks and malicious users targeting embedded-device security mechanisms.Wireless sensor-network communication is vulnerable to eavesdropping and man-in-the-middle attacks.
  • IoT devices are less protected than servers because physical access can simplify penetration and the number of potentially compromised devices is larger.Device proximity to users also increases the consequences of information leakage.
  • Resource constraints are a key barrier to implementing security in IoT environments.This constraint limits the direct use of resource-intensive protection mechanisms.
  • Cryptographic protection for confidentiality and authenticity consumes considerable bandwidth and energy on constrained devices.Lightweight and symmetric cryptographic techniques are therefore considered in RFID and wireless-sensor contexts.
  • IoT heterogeneity and distributed deployment make patching more time-consuming and can leave systems exposed.Existing RFID and wireless-sensor systems used in logistics, fleet management, farming, and smart cities remain vulnerable to attacks across layers.
  • Existing survey work has examined IoT attacks, but gives less attention to solutions and counter-attack practices.This identifies a security-research coverage gap rather than a performance result.

1.9 Identity Management and Authentication

IoT identity management must uniquely identify devices and support access control, authentication, and trust in dynamic, heterogeneous environments. These mechanisms affect how securely and efficiently diverse entities connect.

  • Authentication and trust: Management systems can enforce access-control policies, monitor them, and establish trust negotiations with external partners.
  • Authentication and trust: IoT security mechanisms must account for dynamism and heterogeneity, including devices that regularly enter and leave networks.Vehicular networks illustrate this setting through moving cars interacting with access points and roadside sensors.
  • Identity management: IoT devices require unique identification, supported by mechanisms including ucode, RFID, EPC, and URI-based identifiers.The passages describe ucode as generating 128-bit codes and EPC as creating unique identifiers.
  • Identity management: Global identification and location can reduce the complexity of expanding local environments and linking them with global markets.
  • Identity management: Identity management systems can delegate management locally when devices share geographical coordinates or belong to the same group.
  • Authentication and trust: Context-aware pairing, automatic authentication, and zero-interaction approaches aim to simplify secure network formation and influence IoT adoption.

1.10 Privacy

IoT privacy concerns are intensified by rapidly growing data generation, storage, and sensing of personal information. Proposed responses include centralized control, privacy-by-design, distributed algorithms, and privacy-enhancing technologies.

  • Privacy concerns: The growth of generated data and declining storage costs make it easier and cheaper to retain user data for long periods.The passages connect these trends with harvesting as much data as possible for later use.
  • Privacy concerns: IoT sensors can collect location, heartbeat, and motion data, making user control over shared data and access an ongoing privacy concern.
  • Privacy approaches: Privacy in distributed IoT environments can follow either a centralized approach or privacy-by-design, where each entity manages its own data flows.
  • Privacy approaches: Distributed privacy-preserving algorithms address data scattering and associated privacy tags when entities access only data chunks.
  • Privacy approaches: Privacy-enhancing technologies are presented as candidates for protecting collaborative protocols and sensitive data.The section also mentions rapidly deployable enterprise solutions using containers on virtual machines.

1.11 Standardization and Regulatory Limitations

IoT standardization is constrained by regulatory policies and the diversity of technologies, creating interoperability challenges across devices, protocols, providers, and users. Broad standards are presented as a way to ease participation and improve interoperability, although growth can make standardization more difficult.

  • Challenges: Regulatory policies and standardization limitations challenge interoperability among devices, authentication, identification, authorization, and communication protocols.
  • Challenges: These challenges can become barriers to IoT growth and adoption as organizations and industries increasingly incorporate IoT.
  • Potential benefits: Defining and broadcasting standards can ease entry for new users and providers and positively influence interoperability across IoT components and services.
  • Regulatory limitations: Increasing IoT growth can complicate standardization, while regulations governing radio-frequency access impose additional constraints.
  • Standardization efforts: The chapter lists organizations and technologies contributing to IoT standardization, including CoRE, 6LoWPAN, ROLL, IPv6, and Web of Things.

1.12 Conclusions

The chapter presents IoT as a distributed and heterogeneous paradigm whose development still faces open challenges from communication requirements through middleware. It highlights integration, communication, data analysis, and service-oriented approaches as directions for addressing this complexity.

  • Conclusions: IoT emerged as a paradigm for M2M communication, while the role of humans in future IoT scenarios remains an investigated topic.
  • Conclusions: Connectivity, interoperability, and integration are described as inevitable parts of IoT communication systems.
  • Conclusions: IoT’s distributed and heterogeneous composition makes integrating its many components and hiding complexity from users necessary.
  • Future directions: SOA architectures and API definition languages are proposed for service exposition, discovery, and composition.
  • Conclusions: Sensors and smart devices are identified as building blocks of IoT.
  • Future directions: The chapter highlights open challenges ranging from communication requirements to middleware development and provides typical solutions and research guidelines.
Loading 1703.06409v1…