The next phase in our move to the rent-don't-buy economy shapes up with "device as a service".

James Kobielus, Tech Analyst, Consultant and Author

September 21, 2018

7 Min Read
Image: Pixabay

Ownership is becoming passé, or at least it’s no longer seen as the primary means for achieving the good life. As what’s been called the “sharing economy” “access economy,” or “collaborative economy” takes hold, consumers are more open to renting and other options that don’t require a transfer of ownership.

You can also think of this phenomenon as “property as a service.” People will pay for a property-grade consumption experience -- such as having full access to someone else’s car or home —when they only need it for a limited term and don’t wish to go into debt, assume title, or take on the other financial burdens associated with ownership. Some startups have branched into facilitating peer-to-peer renting of unused personal devices, such as cameras, musical instruments, and athletic gear.

As the Internet of Things spreads, we’re likely to see the advent of what some are calling “device as a service.” Contrary to what you may think, this doesn’t mean that most of us will be renting out our smartphones and gaming consoles to the highest bidder. Instead, device-as-a-service refers to the futuristic cloud-computing approach under which:

  • Users will be able to rent out the compute, storage, memory, and bandwidth resources of IoT endpoints individually or in larger virtualized clusters.

  • The infrastructure will harness the power of cloud-to-edge virtualization so that two or more IoT devices can be elastically and adaptively combined to handle dynamic workloads.

  • The end user will be presented with an AI-personalized holistic experience, as if all these scattered devices, yoked on demand, were simply nodes in an edge-oriented cloud service delivering a unitary application.

Another way of looking at this is “experience as a service,” with the cloud-to-edge infrastructure enabling us to rent a slice of IoT device bliss on-demand without our having to take possession of its physical instantiation. In a world where every piece of physical infrastructure is increasingly repurposable to support shifting workloads through edge virtualization, device-as-a-service has potential applications in such domains as:

Personal devices as a service: More consumers may choose to rent rather than own smartphones, tablets, appliances, and other personal edge devices, much the same way that time-sharing is done in the vacation rental industry. Within the IoT edge cloud, device-as-a-service might be used to dynamically reconfigure each acquired device to the next user’s specific requirements, predictively embedding profile information, environmental context, interest graphs, and other AI assets in physical devices to personalize the experience seamlessly.

Manufacturing and logistics devices as a service: More businesses might choose to rent rather than own their entire value chain, from the factory floor all the way out to the supply chain. Device-as-a-service could be used to flexibly reconfigure assembly lines, shipping containers, warehousing systems, long-haul trucks, and other physical assets to serve each user’s specific requirements, predictively managing equipment performance, prevent failures, and boost efficiency and throughput from end-to-end.

Network traffic management devices-as-a-service: More organizations might rent complex software-defined wide area networks that can be virtually instantaneously reconfigured and optimized for their specific workloads. Device-as-a-service could be used to ensure that all routers, controllers, and other networking devices behave as a seamlessly connected asset, using AI to ensure predictive bandwidth optimization, quality of service maintenance, and service level guarantees.

Bear in mind, though, that device-as-a-service is more of a vision in today’s IoT market than a concrete deployed reality. We’re seeing more discussions of it by consultants in the IoT arena than we’re seeing real-world implementations. But the approach appears to be gaining ground in industry efforts. In late 2017, Bosch contributed its Ditto codebase to the Eclipse Foundation under open source to catalyze the development of a cloud-native, infrastructure-agnostic device-as-a-service capability built around the emerging architectural concept of digital twins.

Before we examine the details of Eclipse Ditto, let’s remind ourselves what we mean when we refer to a digital twin, which is being implemented by a growing range of IoT and cloud vendors for industrial applications.

At heart, digital twin refers to a data construct that serves several functions within an IoT or other cloud that facilitates connections among edge devices:

  • Provides a virtual, cloud based, “digital master” representation of one specific real-world physical device, system, object, or other IoT-connected entity;

  • Mirrors that twinned entity and thereby helps to manage it, either through remote connection or through autonomous local operations;

  • Describes the current configuration, state, condition, behavior, location and other attributes of the twinned entity;

  • Facilitates aggregation, management, and analysis of the sensor data emitted by the twinned entity;

  • Supports data-driven functions such as closed-loop simulation, monitoring, maintenance, diagnostics, remediation, and other life-cycle management functions vis-à-vis that twinned entity; and

  • Drives the AI that guides the dynamic adjustments that a twinned entity makes to its environment.

As described on its Eclipse FAQ page, Ditto defines a hardware and software-agnostic API that enables potentially any IoT-connected entities to be “used as any other web service via its digital twin.” The API programmatic control of digitally twinned devices-as-a-service from user applications, application infrastructure, and cloud-native back-end services. It defines digital-twin interfaces that can be implemented throughout an edge cloud, such as embedded on an IoT edge device, installed in a mobile or web app, and configured into an IoT Internet gateway. It can also be implemented back-end edge-cloud middleware services such as those responsible for routing requests between hardware and applications, enforced access controls, managing device state, and managing changes and notifications across distributed IoT fabrics.

However, Eclipse Ditto does not define a lot of what would be necessary to stand up a working device-as-a-service implementation. The project is scoped so as to be agnostic to the following:

  • Frameworks, platforms, protocols, and data schemas for IoT networks that support device-as-a-service;

  • Reference software for running device-as-a-service in virtual machines, containers, or other runtimes on IoT gateways or other nodes;

  • Guidance for training digital twins’ corresponding AI models for device-as-a-service, for governing the master data store underlying digital twins, or for managing events and orchestrations among the digital twins corresponding to specific devices

Clearly, these out-of-scope matters are intended to facilitate the incorporation of device-as-a-service as an overlay in cloud, edge, IoT, and cloud-native computing environments. In addition, the specification doesn’t address how to align an end-to-end device-as-a-service with user intentions and interests. This will be fundamentally necessary if the proponents of this concept hope for it to catch on.

Though the digital twin concept got started in the industrial IoT, it isn’t intrinsically limited to that application domain. In a larger sense, it’s a modeling paradigm that can be applied to any IoT-connected endpoint, including human beings at edge nodes. Within the device-as-a-service paradigm, digital twins can serve as a framework for developing, training, and managing any intelligent virtual assistant that acts on behalf of a user who’s attempting to harness of vast array of connected nodes to their needs.

To serve this function, the digital twin would need to encode the intentions, knowledge, and interest graphs used by intelligent agents to calculate which actions are aligned with specific user intentions at any time. In this regard, I recommend an excellent Huawei blog on the vision of “intent-driven networking” that uses digital twins to optimizes user experience in device-as-a-service implementations for connected autonomous vehicles.

Just as desktop virtualization took virtualization to the edge years ago by enabling experience portability across distributed computing fabrics, we’re likely to see devices-as-a-service take this revolution to the next step. Within the coming 5 to 10 years, we’re likely to see this technology mature to the point where it will dilute people’s interest in owning any piece of IoT-connected property that is capable of being provisioned, rented, and personalized for them for any period of time, no matter how long or brief.

About the Author(s)

James Kobielus

Tech Analyst, Consultant and Author

James Kobielus is an independent tech industry analyst, consultant, and author. He lives in Alexandria, Virginia.

Never Miss a Beat: Get a snapshot of the issues affecting the IT industry straight to your inbox.

You May Also Like


More Insights