Embedded Devices and the Cloud

Devices in the Cloud: Driving Intelligence Where You Need It

The expansion of the Internet of Things is opening exciting new business opportunities with distributed connectivity and huge amounts of data. Knowing how and where to manage such intelligence is key to being able to realize its potential.


  • Page 1 of 1
    Bookmark and Share

Article Media

Analysts predict that 15 billion intelligent devices will be connected in the Internet of Things by 2015. Ultimately, connected devices will control everything from indoor temperatures to in-dash navigation, from the flow of energy through our cities to the flow of intravenous fluids. The prospects are at once exciting and daunting: exciting because of the potential to rejuvenate industries and create entirely new streams of revenue; daunting because of the complexities involved in making sure all those devices perform as promised.

Let’s look at the exciting part first the opportunity to adopt new business models and revenue streams. One Wind River customer is a manufacturer of forklifts (Figure 1). They traditionally made money selling or leasing equipment a product. Now, with the confluence of embedded technology, wireless connectivity and the Cloud, they are seeing the opportunity to evolve from selling products to selling subscription services to get paid for the usage of the product based on tonnage per month and not just for the product itself. Smart sensors on the forklifts can record and report how many tons they are picking up, how far they are traveling and other variables relevant for the new business model.

Figure 1
Using an expensive piece of equipment as a service available for subscription offers opportunities like opex and capex savings due to increased asset uptime, but it requires the ability to update firmware remotely as well as acquire use and maintenance data.

That is not much different from the telecom carriers who will gladly sell you a smartphone at a deep discount in order to sell you monthly, renewable voice and data plans. It’s a model that more and more hard-goods industries are interested in adopting. And it is only possible because of the ability to connect smart onboard devices with Cloud-based applications and analytics.

In our enthusiasm for the new possibilities, however, we cannot overlook the real challenges to achieve this vision. While the challenges are many, they tend to fall into three broad categories: connectivity, security and manageability.

Connectivity: With new products currently being developed, we can start building intelligence and connectivity into them from scratch. Today, though, much of the industry’s effort is focused on connecting legacy or “brownfield” devices that until now have stood alone. Most of these devices were not designed with connectivity in mind. On the contrary, many were designed to make connectivity difficult in order to protect them from network-borne threats. Now, builders and operators of large-scale systems want to take advantage of the efficiencies and economies that the IoT promises. To reap those benefits, they must figure out not only how to connect them but also how to protect them.

Another complication is that there is no single standard for connecting to networks. Many brownfield devices use proprietary protocols and will require gateways to connect with IP-based networks. And if they are already IP-based, they may be using a wide variety of protocols. Developers will need to be able to build gateways that support virtually any communication protocol.

The availability, accessibility and cost of bandwidth are another constraint. For equipment operating in remote locations, the transmission of data from onboard devices to Cloud-based applications via satellite can be an expensive proposition. We need to figure out ways to move data efficiently to where it needs to be. That starts with figuring out more precisely which data is needed at which level.

Security: As we become increasingly reliant on intelligent, interconnected devices in every aspect of our lives, how do we protect them from intrusions that could compromise personal privacy or threaten public safety? The number of network-based attacks on embedded devices that control critical infrastructure is increasing at an alarming rate, as is their sophistication. Security is arguably the overriding issue in the Internet of Things, inseparable from performance and reliability.

Security needs to be factored in at every level, from the devices to the gateways to the Cloud-based systems that control them (Figure 2). Virtually every known type of hardware and software security measure comes into play in the IoT. Secure booting at the device level, access control and authentication, application whitelisting, firewalls and intrusion prevention systems are just some of the tools at hand to respond to security threats.

Figure 2
Security for connected devices and services must be applied at various levels in a system of connected devices, and targeted for the specific characteristics and needs associated with those levels.

Manageability: Once you have addressed the connectivity and security issues, the next challenge is how to manage the device remotely over time. You need to be able to provision it securely with software updates as they become available. You need to send security patches as vulnerabilities are uncovered. You need to be alerted when the device is not performing properly, and you need the capability to diagnose problems remotely when they happen. And you need to be able to perform tasks like these without the risk of disruption or downtime.

It Starts with an End-to-End View

Players in the IoT market need to take an end-to-end view that encompasses the endpoint device, the connectivity layer, the gateway and the application running in the Cloud. We need to understand what the entire system is meant to do and the role each component plays, or could potentially play, in its overall operation. By looking along the whole continuum, we can identify opportunities to deliver intelligence where it’s needed to optimize performance.

Consider, for example, the issues of processing capacity and bandwidth. One of the big challenges in the IoT is how to deal with the enormous volumes of machine-generated data flowing through it, from devices through gateways to applications in the Cloud and back. It puts a huge strain on bandwidth, which can drag down performance and drive up costs. Conventional thinking is that all that data is necessary for Cloud-based applications to do what they’re supposed to do analyze the data from the devices, monitor performance, make decisions and send instructions back to the devices.

What if, instead, you could push a lot of the computing the intelligence down to the gateway, or even to the device itself? Imagine if the device could perform various services, such as smart data aggregation and filtering, and figure out which data needs to be sent to the Cloud. The Cloud applications would continue to perform the heavy-duty analytics and create statistical models for crunching the data. Ultimately, though, once the model has been fine-tuned and is running properly, a scaled-down version of it can be sent to the device, which can then take over much of the processing.

This vision is quickly becoming a reality. With a “write once, run everywhere” development platform and a scripted programming language optimized for resource constraints, developers will be able to create small-footprint applications that bring intelligence to the device level. Bandwidth constraints will be less of an issue as smart devices become more selective in the data they send to the Cloud, and systems distribute their processing more efficiently.

Translating Technology Advancement to Business Advantage

Of course, the end benefit of optimal IoT performance is the business advantage it delivers, whether that means saving money through increased productivity or making more money by creating sources of revenue that didn’t exist before.

Predictive maintenance is one of the more compelling advantages of the IoT. Take, for example, a wind turbine. Typically located in remote places, on mountainsides or even at sea, uptime is critical for turbines. To prevent failure, operators historically sent crews of technicians out to the turbines to perform routine inspections and preventive maintenance according to fixed schedules—a labor intensive process with no guarantee that failure won’t occur. Today, smart sensors can predict failure in real time with great accuracy based on any number of symptoms, such as changes in blade vibration patterns. Automatic software adjustments can often rectify the issue without the need for a crew onsite.

Moreover, control systems in the Cloud can collect data not only from the wind turbines themselves, but also from external sources, such as reports on airborne dust accumulations from the National Weather Service. The system can aggregate and analyze all these variables and determine precisely when a specific piece of equipment needs servicing or repair, avoiding unnecessary disruption or downtime and dramatically reducing maintenance costs.

Cost reduction is what initially attracts many companies to IoT solutions. However, once they experience the power of data analytics, they begin to identify new business and revenue opportunities. Our forklift business mentioned earlier is one such example. Another is a medical device manufacturer that evolved from marketing stand-alone to connected biofeedback devices. The devices can transmit data directly from patient to doctor, eliminating the need for an office visit. In the process of creating greater efficiency and reducing costs for both patient and provider, the company realized it was accumulating valuable data on multiple patients with similar conditions. Individual patient data, of course, must be kept confidential by law. Once aggregated and rendered anonymous, however, the data yields patterns that are very useful to anyone who wants to better understand a particular medical condition. The company could create a new source of revenue by providing data on the progression of a disease for medical researchers and insurers.

The full potential of the Internet of Things is only beginning to be recognized, let alone realized. Those of us on the technology side need to keep working closely with our customers to understand their business drivers. That enables us to develop the tools needed to overcome the challenges connectivity, security and manageability and turn exciting possibilities into realities.

Wind River
Alameda, CA
(510) 748-4100