Source-linked AI summary
Internet of Things (IoT): A Vision, Architectural Elements, and Future Directions
Jayavardhana Gubbi, Rajkumar Buyya, Slaven Marusic, Marimuthu Palaniswami
TL;DR
IoT must manage massive sensing data and converge heterogeneous technologies into an integrated architecture. The paper proposes a user-centric, cloud-based framework using public-private cloud interaction, concluding that scalable separation of networking, computation, storage, and visualization can support diverse IoT needs.
Problem
IoT generates enormous data requiring seamless storage, processing, presentation, and intelligent use, while extracting information across complex spatial and temporal sensing environments remains challenging.
Method
The paper develops a user-centric cloud-centric IoT vision and presents Aneka-based interaction between private and public clouds for shared services and resources.
Results
The proposed framework separates networking, computation, storage, and visualization to allow independent growth within a shared, scalable cloud-enabled environment.
Takeaways & Limitations
The paper concludes that IoT research should pursue convergence among wireless sensor networks, the Internet, and distributed computing through flexible cloud-supported infrastructure.
Takeaways & Limitations
The proposed cloud-centered architecture may not be the best option for every application.
Abstract
from arXiv · showhide
Ubiquitous sensing enabled by Wireless Sensor Network (WSN) technologies cuts across many areas of modern day living. This offers the ability to measure, infer and understand environmental indicators, from delicate ecologies and natural resources to urban environments. The proliferation of these devices in a communicating-actuating network creates the Internet of Things (IoT), wherein, sensors and actuators blend seamlessly with the environment around us, and the information is shared across platforms in order to develop a common operating picture (COP). Fuelled by the recent adaptation of a variety of enabling device technologies such as RFID tags and readers, near field communication (NFC) devices and embedded sensor and actuator nodes, the IoT has stepped out of its infancy and is the the next revolutionary technology in transforming the Internet into a fully integrated Future Internet. As we move from www (static pages web) to web2 (social networking web) to web3 (ubiquitous computing web), the need for data-on-demand using sophisticated intuitive queries increases significantly. This paper presents a cloud centric vision for worldwide implementation of Internet of Things. The key enabling technologies and application domains that are likely to drive IoT research in the near future are discussed. A cloud implementation using Aneka, which is based on interaction of private and public clouds is presented. We conclude our IoT vision by expanding on the need for convergence of WSN, the Internet and distributed computing directed at technological research community.
1. Introduction
The introduction presents IoT as a network of interconnected objects that sense, actuate, communicate, and provide services through pervasive networks and cloud computing. It frames the paper around IoT technologies, definitions, applications, cloud realization, analytics, and future challenges.
- IoT vision: IoT connects surrounding objects to networks through RFID and sensor-network technologies, extending computing beyond traditional desktop environments.The paradigm places networked objects throughout everyday environments and uses sensing technologies to support this shift.
- Cloud computing: Cloud computing supplies virtual infrastructure integrating monitoring, storage, analytics, visualization, and client delivery for on-demand access.Its cost-based model supports end-to-end service provisioning for businesses and users from anywhere.
- Enabling foundations: Successful IoT requires smart connectivity, context-aware computation, pervasive communication networks, and analytics capable of autonomous and smart behavior.The introduction identifies these capabilities as necessary to process and convey contextual information where it is relevant.
- IoT vision: IoT combines sensing, actuation, Internet standards, analytics, applications, and communications to interact with the physical world.The described network both harvests environmental information and provides services for information transfer and computation.
- Scale and opportunity: 2011 marked the point when interconnected devices outnumbered people; the paper reports 9 billion devices currently and 24 billion expected by 2020.It also identifies $1.3 trillion in revenue opportunities for mobile network operators across health, automotive, utilities, and consumer electronics.
- Paper organization: The paper covers IoT vision and technologies, definitions and taxonomy, applications, cloud-centric realization, Aneka/Azure analytics, challenges, and future trends.These topics are organized across Sections 2–7 as the paper’s stated contributions.
2. Ubiquitous computing in the next decade
Ubiquitous computing embeds technology into everyday life through interconnected devices, ubiquitous sensing, and distributed computing resources. The IoT vision extends this approach by combining sensor-actuator networks with cloud computing to analyze shared information and support smart environments.
- 2. Ubiquitous computing in the next decade: Ubiquitous computing aims to embed technology into the background of everyday life, with smartphones and handheld devices making environments more interactive and informative.Mark Weiser defined the ubiquitous computing discipline and smart environments.
- 2. Ubiquitous computing in the next decade: The Internet enables devices to communicate worldwide while exposing distributed computing resources and storage owned by multiple parties.Inter-networking is presented as a major step toward the ubiquitous computing vision.
- 2. Ubiquitous computing in the next decade: Miniature wireless sensor nodes sense, compute, and communicate over short distances, forming WSNs for environmental, infrastructure, traffic, and retail monitoring.WSNs result from converging MEMS, wireless communications, and digital electronics.
- 2. Ubiquitous computing in the next decade: Cloud computing provides efficient, secure, scalable, and market-oriented computing and storage for receiving, analyzing, interpreting, and presenting data from ubiquitous sensors.The cloud is described as a receiver of sensor data and a platform for computation and user access.
- 2. Ubiquitous computing in the next decade: An integrated Sensor-Actuator-Internet framework shares information across platforms and applications to create a common operating picture and enable control of unrestricted things.The framework is positioned as the core technology for shaping smart environments and supporting data-on-demand in the transition from web to web2 and web3.
3. Definitions, Trends and Elements
The section defines IoT as an intersection of internet-, things-, and semantic-oriented paradigms, centered on interconnected sensing and actuating devices that share information through unified frameworks. It also outlines IoT’s growing prominence, core hardware–middleware–presentation components, enabling technologies, and emerging data-management challenges.
- Definitions: IoT’s usefulness emerges where internet-oriented middleware, things-oriented sensors, and semantic-oriented knowledge intersect.The RFID group further characterizes IoT as a worldwide network of uniquely addressable interconnected objects using standard communication protocols.
- Definitions: IoT for smart environments interconnects sensing and actuating devices to share information across platforms and develop a common operating picture.This relies on large-scale sensing, data analytics, information representation, and cloud computing.
- Trends: IoT search volume has consistently increased since its emergence, while Wireless Sensor Networks search volume has declined.The passage projects this trend to continue for the next decade as enabling technologies converge.
- Elements: The IoT taxonomy comprises hardware, middleware, and presentation components for seamless ubiquitous computing.Hardware includes sensors, actuators, and embedded communication hardware; middleware provides on-demand storage and computing for analytics; presentation provides visualization and interpretation tools.
- Elements: RFID, low-power wireless sensor devices, resource naming, and metadata support IoT identification, sensing, communication, and access.RFID enables automatic identification, while URN, metadata, and IPv6 support unique and remote resource access.
- Elements: IoT’s unprecedented data generation makes storage, ownership, expiry, and energy-efficient data-center operation critical issues.The passage notes that the internet already consumes up to 5% of total generated energy and that demand is expected to rise.
4. Applications
The paper organizes IoT applications into domains based on network and user characteristics, spanning personal and home systems, enterprises, smart environments, utilities, video, water, agriculture, transportation, and logistics. These applications use shared sensing and data infrastructures to improve healthcare, energy and resource management, urban services, surveillance, and mobility.
- Application domains: IoT applications are classified by network availability, coverage, scale, heterogeneity, repeatability, user involvement, and impact.The paper groups applications into four broad domains and gives typical applications for each.
- Utility IoT: Utility IoT uses extensive networks and smart meters for resource management, cost optimization, efficient electricity consumption, and grid load balancing.Personal and home electricity-use data can be shared with utility companies to help optimize supply and demand.
- Personal and Home IoT: Personal and home IoT supports healthcare monitoring, aged-care at home, home equipment control, energy management, and social networking among interconnected objects.Body-area sensors can upload physiological data through smartphones, while home monitoring may enable early intervention and reduce hospitalization costs.
- Enterprise and Smart Environment IoT: Enterprise and smart-environment IoT networks support building utilities, environmental monitoring, factory maintenance, smart cities, retail, transportation, water, and agriculture.Applications include tracking occupants, managing HVAC and lighting, monitoring environmental conditions, and sharing urban data across impact areas such as health, mobility, pollution, and government services.
- Video-based IoT: Video-based IoT combines image processing, computer vision, and networking for surveillance, target tracking, suspicious-activity detection, and unauthorized-access monitoring.Automatic behavior analysis and event detection remain in their infancy, with expected breakthroughs in the next decade.
- Smart Transportation and Logistics: Smart transportation and logistics address congestion, pollution, freight delays, delivery failures, dynamic traffic information, and large-scale wireless sensor networks.Dynamic traffic data can improve freight movement, planning, and scheduling, while congestion affects economic and social activities.
5. Cloud centric Internet of Things
The paper develops an Internet-centric IoT architecture that integrates sensed information, analytics, and visualization across public and private clouds using Aneka. It presents cloud-based service sharing and autonomic resource management as foundations for delivering IoT applications through the Future Internet.
- Architecture: The proposed IoT architecture is Internet-centric, with Internet services as the main focus and data contributed by connected objects.The framework emphasizes interaction between cloud environments to support the architecture.
- Architecture: Aneka integrates public and private clouds so developers can combine sensed information, analytics algorithms, and visualization in one framework.This interaction is described as critical for application developers building IoT applications.
- Aneka Platform: Aneka is a .NET-based PaaS providing APIs, runtime support, and Task, Thread, and MapReduce programming models across public and private cloud resources.Its services support resource control, auto-scaling, reservation, monitoring, and billing.
- IoT Services: Automatic cloud management is proposed to host and deliver IoT services as SaaS applications while enabling reusable data and service sharing across application scenarios.Anomaly detection in sensed data is given as an example of a shareable Application-layer service.
- Autonomic Management: Aneka’s autonomic features primarily support application scheduling and dynamic resource provisioning through manager effectors controlling scheduling and provisioning.These components coordinate resource management for IoT application execution.
- Resource Management: The scheduler assigns resources to application tasks using QoS and provider-cost considerations, while dynamic provisioning instantiates or terminates computing, storage, and network resources.Provisioning negotiates with public and private cloud IaaS providers based on requirements, execution history, timing, cost, and budget availability.
6. IoT Sensor Data Analytics SaaS using Aneka and Microsoft Azure
The section presents an IoT sensor-data analytics SaaS architecture combining Microsoft Azure with Aneka for provisioning, cloud interoperability, execution management, and dynamic analytics-tool updates. Aneka supports hybrid private–public cloud operation and task-based execution, while MEF enables DLL-based analytics assemblies to become available to applications through folder updates.
- Azure–Aneka Integration: Aneka uses Azure as provisioning infrastructure and can launch any number of Azure instances to run applications.This integration combines Azure’s cloud platform with Aneka’s provisioning capabilities.
- Azure–Aneka Integration: Aneka provides PaaS features including Task, Thread, and MapReduce programming models, runtime execution, workload management, dynamic provisioning, QoS scheduling, and flexible billing.
- Cloud Interoperability: Aneka’s InterCloud model addresses cloud interaction by combining private- and public-cloud resources into a hybrid computing environment.
- Task-Based Analytics: The Aneka task programming model expresses analytics and artificial-intelligence applications as independent tasks that can run in any order on different data or operations.Independent tasks allow computationally demanding analytics workloads to use large-scale resources.
- Dynamic SaaS Updates: MEF enables developers to update shared DLL-based analytics tools dynamically, making an analytics assembly available to applications when placed in a designated folder.MEF is described as a .NET composition layer that improves flexibility and maintainability despite Azure administrative constraints.
7. Open Challenges and Future Directions · 7.1. Architecture
The paper presents a flexible, cloud-centric IoT architecture designed around user requirements while identifying privacy, interoperability, sensing, analytics, visualization, cloud, and WSN challenges. It emphasizes that early architectural choices will strongly shape IoT’s development and highlights a roadmap of future technological advances and applications.
- 7. Open Challenges and Future Directions: The cloud-centric vision proposes a flexible, open, user-centric architecture that lets different IoT participants interact according to their own requirements.It includes provisions for data ownership, security, privacy, and information sharing.
- 7. Open Challenges and Future Directions: IoT research must address privacy, participatory sensing, data analytics, GIS-based visualization, cloud computing, and traditional WSN challenges.The listed WSN challenges include architecture, energy efficiency, security, protocols, and Quality of Service.
- 7. Open Challenges and Future Directions: The stated goal is Plug n‘ Play smart objects that deploy in any environment and blend through an interoperable backbone.This goal follows the paper’s discussion of IoT-specific and WSN-related challenges.
- 7. Open Challenges and Future Directions: A research roadmap identifies key IoT technology developments and expected application outcomes for pervasive applications over the next decade.The roadmap is presented in Figure 8, alongside discussion of international initiatives supporting IoT’s success.
- 7.1. Architecture: Initial IoT architectural choices require investigation because they will have a severe bearing on the field’s subsequent development.The paper frames architecture as a major open challenge in IoT research.
- 7.1. Architecture: Early IoT architecture research has largely followed the wireless sensor networks perspective, with SENSEI and IoT-A addressing architecture challenges for different applications.The paper describes both European Union projects as successful in defining application-oriented architectures.
7.2. Energy efficient sensing · 7.3. Secure reprogrammable networks and Privacy
Energy-efficient IoT sensing must coordinate heterogeneous, fixed or mobile modalities and exploit spatial-temporal structure, while compressive sensing can reduce measurements and transmission power. IoT security requires cryptographic, non-cryptographic, cloud, reprogramming, and privacy protections against diverse threats.
- 7.2. Energy efficient sensing: Heterogeneous urban sensing must balance multiple modalities, network traffic, data storage, and energy use across fixed or mobile infrastructure.The framework must also accommodate continuous and random sampling while exploiting spatial and temporal data characteristics.
- 7.2. Energy efficient sensing: Compressive sensing reconstructs signals accurately from a small number of projections when sparsity and basis incoherence conditions hold.The recovery problem seeks the smallest l1-norm coefficient vector consistent with measurements.
- 7.2. Energy efficient sensing: Compressive wireless sensing uses synchronous communication to reduce each sensor’s transmission power by sending noisy projections to a central aggregator.This approach also affects data compression, network traffic, and sensor distribution.
- 7.3. Secure reprogrammable networks and Privacy: Large-scale IoT deployments face attacks that can disable availability, inject erroneous data, or expose personal information across RFID, WSN, and cloud components.Cryptography is described as the first defense against data corruption.
- 7.3. Secure reprogrammable networks and Privacy: RFID, particularly passive RFID, is highly vulnerable because it enables person tracking and lacks high-level intelligence on the devices.The paper identifies cryptographic methods as a potential solution requiring further research.
- 7.3. Secure reprogrammable networks and Privacy: Encryption protects confidentiality against outsiders, while message authentication codes protect data integrity and authenticity.Encryption does not stop insider attacks, which require non-cryptographic measures, particularly in WSNs.
- 7.3. Secure reprogrammable networks and Privacy: Remote wireless reprogramming is needed to install new sensor applications and update existing ones across network nodes.The passage introduces this requirement alongside security limitations in traditional network reprogramming.
- 7.3. Secure reprogrammable networks and Privacy: Hybrid clouds intensify security and identity-protection concerns because IoT data, tools, and economics are exposed across private and public clouds.IoT’s persistent data collection also motivates digital forgetting to protect personal data from positive or negative uses.
7.4. Quality of Service
IoT quality of service must support heterogeneous, multi-service networks carrying diverse traffic without compromising guarantees. Wireless resource constraints and growing cloud-based, high-capacity applications make QoS management a continuing research challenge.
- 7.4. Quality of Service: Heterogeneous networks must support multiple applications and traffic types while maintaining QoS across services.Traffic includes throughput- and delay-tolerant elastic applications, such as low-rate weather monitoring.
- 7.4. Quality of Service: Wireless networks make QoS guarantees difficult because shared-media resource allocation and management can create gaps in resource guarantees.These constraints arise across network segments and limit consistent resource provisioning.
- 7.4. Quality of Service: Cloud computing QoS requires increasing research attention as more data and tools become available on clouds.Dynamic scheduling and resource allocation algorithms based on particle swarm optimization are being developed.
- 7.4. Quality of Service: High-capacity applications may become a QoS bottleneck as IoT grows.The concern is linked to the increasing scale of IoT and demand for cloud-based resources.
7.5. New protocols · 7.6. Participatory Sensing
New IoT protocols must provide energy-efficient communication, resilient routing, and an effective data tunnel between sensors and the outer world. Participatory sensing offers localized, timely environmental data but requires fixed infrastructure to address missing and inconsistent samples.
- 7.5. New protocols: Energy-efficient MAC and appropriate routing protocols are critical to completing IoT sensing systems.Protocols form the backbone of the data tunnel between sensors and the outer world.
- 7.5. New protocols: Available MAC schemes include TDMA and FDMA for collision-free operation, while CSMA offers low traffic efficiency.FDMA requires additional circuitry in sensor nodes.
- 7.5. New protocols: Because individual sensors can drop out, IoT networks must self-adapt and support multi-path routing.Existing multi-hop routing protocols are categorized as data centric, location based, or hierarchical, with energy as the main consideration.
- 7.6. Participatory Sensing: Participatory sensing projects aim to provide low-cost environmental sensing localized to users.User-collected environmental data can function as social currency and enable more timely data generation.
- 7.6. Participatory Sensing: Participatory sensing can closely indicate environmental parameters experienced by the user because measurements are localized to that user.The approach is people centric and relies on data collected by users.
- 7.6. Participatory Sensing: Fixed-infrastructure IoT must provide reference data because participatory sensing suffers from missing samples and inconsistent collection.User participation, timing, location, and travel paths limit the ability to produce meaningful data.
7.7. Data mining · 7.8. GIS based visualization · 7.9. Cloud Computing
The section identifies challenges in extracting higher-level knowledge from complex IoT sensing data, visualizing heterogeneous spatio-temporal information, and building scalable, reliable cloud applications. It emphasizes temporal activity inference, Internet GIS frameworks, multi-stakeholder service integration, and adaptive resource scheduling with failure management.
- 7.7. Data mining: IoT data mining must progress from shallow supervised and unsupervised extraction of predefined events and anomalies toward inferring local activities from their temporal information.The paper frames useful information extraction across different spatial and temporal resolutions as a challenging artificial-intelligence problem.
- 7.8. GIS based visualization: Creative visualization is enabled by increasingly capable displays, including touch-based Plasma, LCD, LED, AMOLED, and emerging 3D technologies.These technologies improve data representation and navigation while creating further research and development opportunities.
- 7.8. GIS based visualization: IoT visualization requires further processing because ubiquitous-computing data is not always ready for direct consumption, especially for heterogeneous spatio-temporal data.The challenge concerns both data preparation and the complexity of representing heterogeneous information across space and time.
- 7.8. GIS based visualization: Visualizing geo-related, sparsely distributed IoT data requires new temporally varying 3D sensor representations and a framework based on Internet GIS.The proposed direction addresses heterogeneous sensors distributed across geographic space.
- 7.9. Cloud Computing: Integrated IoT and Cloud applications for Smart Cities must combine multiple stakeholders’ services, scale reliably and decentrally, support wired and wireless networks, and tolerate limited power and unreliable connectivity.These requirements apply across constrained access devices and data sources.
- 7.9. Cloud Computing: Cloud resource management must dynamically prioritize requests and provision resources so critical requests are served in real time.The scheduling system is also expected to use task duplication algorithms to manage failures and deliver results reliably.
- 7.9. Cloud Computing: Cloud application scheduling algorithms should support multi-objective optimization while coordinating resource provisioning for diverse application requirements.The passage introduces multi-objective optimization as a required scheduling capability.
7.10. International Activities
International IoT activities are gaining momentum through coordinated efforts across industry, academia, and government. Europe is consolidating research into a unified framework, while initiatives in Asia, the USA, and Australia advance related capabilities and standards.
- Global initiatives: IoT initiatives are gathering momentum worldwide across industry, academia, and multiple levels of government.Stakeholders are seeking a coordinated path for realizing IoT’s technological evolution.
- European coordination: Europe is consolidating M2M, WSN, and RFID research activities into a unified IoT framework through the EU-FP7-supported IERC.IERC aims to establish a European cooperation platform and research vision and serve as a global IoT research contact point.
- International programs: Japan, Korea, the USA, and Australia are advancing IoT capabilities through collaborations involving smart cities, smart grids, smart metering, broadband, and RFID technologies.These large-scale initiatives involve industry, associated organizations, and government departments.
- Chinese initiatives: Shanghai established an Internet of Things center with a total investment over US$ 100million to study technologies and industrial standards.Wuxi also founded an IoT industry fund and the “Sensing China” IoT Union, initiated by more than 60 telecom operators, institutes, and companies.
8. Summary and Conclusions
The paper presents IoT as an emerging technology that integrates communicating sensing and actuation devices to enable new capabilities from rich information sources. It proposes a user-centric, scalable cloud model using private–public cloud interaction to flexibly support diverse end-user needs.
- Communicating-actuating devices bring the Internet of Things closer by blending sensing and actuation into the background and enabling capabilities from rich information sources.
- The proposed user-centric model uses private and public clouds to place end-user needs at the forefront of IoT deployment.
- A scalable cloud framework provides flexibility for diverse and competing sectors while supporting IoT networking, computation, storage, and visualization.