In Part 1 of this BLOG series, SAP Internet of Things (Industrial Flavor) – Part 1, the focus was on setting the stage for how the consumer and industrial components of the Internet of Things play a complementary role. It also focused on how the existing infrastructures and connections will be reused and how new ones, like near-field communication (NFC), will augment these existing lines and provide lower latency and high-value data alongside of the traditional data flows. This next part will start to dive into more and more details on how this can be achieved today and what it means for an industrial business to deploy and manage something like this.
Consumerization of IoT
The real value of the IoT story is around the management and contextualization of the data that each connected device can bring to the network (M2M) and overall how it fits into a broader business process (IoT). The trick is assigning and categorizing this data and then managing this, often intermittent, data in a way that will benefit both a business process and end-users. Businesses need to make money and consumers like saving money, so simplifying the life of a consumer without sacrificing any personal security or privacy is clearly a key expectation of an inter-connected world. As an example, if a user’s phone knew they were probably waiting for a flight based on being at a gate for a period of time in a terminal, the airlines could immediately know travel intentions for a user. This can be done without identifying the user directly, but via the concept of near-field communication. Then by coupling this with a process for optimization of the consumer’s experience by asking basic questions around flight time or number from a reduced set of options the application could feed information about flight delays and alternate routes if the user chose to answer them. Once signed-in the airline could sell upgrades and other flight amenities to the person, including the ability to order food or specialty beverages instead of waiting for the meal cart or having the flag down a flight attendant mid-flight. As a consumer whom states they do not want big brother to know where they are, for at least this particular scenario, one could argue that it’s already too late once you’re in an airport. At this point airport security knows exactly what your travel intentions are since they were required to acquire a boarding pass in the first place. This end user tracking is not always the case and can be considered a violation of consumer and privacy and this is a touchy subject, but the service provided to the consumer may make them think twice about opting out.
M2M Processes Are Reused by IoT Orchestration
As we all known, machines don’t interact with each other unless there’s a well-defined business process or application in place that enables them to. Machines are not yet capable to make decisions and have intelligent conversations on their own without standards in place that enable this type of integration. It is no different than having a common language and measures for people in a region so that they can communicate and interact with each other. It’s more a matter of convincing things (people and machines) to adhere to common standards in order to minimize disruption to the flow of information by reducing complexity of the overall system.
In the above scenario there was an orchestration process and a well-defined set of data required that a user had to add in order to provide additional context around it in order for the system to properly identify and supply the proper information to them. Also, in order to provide full and personalized value to the end user additional information was required. Where else do these end-user scenarios apply, and further extending the topic away from the consumer, what does this mean for an industrial process? Which industries have the largest hurdles and which have the biggest potential for benefit from an interconnected world of devices and people? All of these are excellent questions and ones that we will continue to explore
What is IoT (Including M2M) to SAP
As previously stated, the value of IoT is found in the reuse of data across a wide range of business process and linking one or many machines or people to one or many other machines or people in a coordinated and traceable fashion. Thus providing for greater visibility and analytical reporting across personal, public, and corporate ecosystems. This is a very broad and general definition but it has implications for supporting multiple scenarios and different technical landscapes as well that are optimized to handle different use cases depending on what the goal of the integration or intelligence reporting, or analytics actually is. The IoT topic can be easily broken into multiple categories but there are two fundamental divisions and they are “Industrial IoT” and “Consumer IoT”. These two divisions encompass all of the potential IoT architectures which are 1 or many hub and spoke hierarchies and networks of devices. There are more segments as you get deeper into specific scenarios that cover multiple different topics and require different architectures to support them depending on if you are a customer to or vendor of various services and equipment. The same goes for managing access to the information provided by the Things that drive your or someone else’s business processes.
The above attempts to provides a clear and easy way to categorize the use cases by using a single central or multiple distributed hierarchies that are required to properly realize the scenario and are based on architecture modeling requirements rather than by business function. The Consumer IoT architecture model involves a single central cloud installation and applications that talk directly to the cloud and are “contextualized” by the cloud application as well. Devices are registered and classified and properties are shared across similar devices in order to simplify management of the consumer device cluster. The Industrial IoT scenario consists of one or many hub and spoke models in order to accommodate existing infrastructure and machine/equipment monitoring scenarios that have already been employed at the industrial automation layer and are required for local autonomy and reporting. This encourages and benefits corporations whom already have a robust set of M2M processes defined as they can reuse and leverage this work to help feed the broader IoT Orchestration framework. One could even argue that in various scenarios the use of near-field communication (NFC) is a mini form of M2M that can be re-used by an IoT process; something like Micro-M2M or Personal M2M. However, because the broader IoT layer knows the device, perhaps the device itself is even smart enough to know about the IoT layer around it and how it fits, additional linkages about how it’s operation impacts the world around it can also be quickly derived and presented to the user or users.
IoT for Consumers
The consumer IoT story plays a high level role around mobile device management and tracking of usage of services as well as location of the user are key pieces of information that provide context as to what is and what is not relevant for them. It also can make consumers nervous about the personal privacy even if no personal information is collected by the monitoring. This type of data collecting of how frequent consumers walk past various locations in a store and at what level on the shelves are looking at that time can be used for precision marketing purposes and targeting the right coupons to motivate a buyer to purchase an item they are close to and also upsell them on something that is a related product.
Here are a few consumer IoT examples, if a consumer pauses by ice cream in the freezer section, a coupon for chocolate syrup could be pushed to their phone (http://www.myfoxchicago.com/story/22847651/stores-tracking-shoppers-behavior-through-cell-phones-surveillance). This type of precision marketing and tracking of what a user is buying seems like it could be a very useful tool as very often store coupons go unused or are often lost in the day to day shuffle. If they were instantly available and accessible this would certainly help to reduce this providing value to a consumer without compromising their identity any more than a store rewards or credit card already does and would help retailers more reliably move stagnant inventory off of their shelves.
The other, and fancier for now, option will be around consumers and their ability to control other things around them with things that they are wearing. Such as the Audi Smart Watch (Audi unveils smart watch to start its driverless car) that enables an end user to remotely start and interact with their car. Again this type of technology is very cool and one can easily see the benefit of ease of access rather than fumbling for your keys on a cold or stormy day. However, the security question still remains that if I lose my watch do I also lose my car, or access to it?
Finally, having a device that can roam around your home as a hub and interact with you, and your family, along with various devices around your home seems to have a lot of appeal and also much lower security concerns. As proof of this at the introduction of this thing “JIBO” there was lots of venture capitalist money and many pre-orders as well (JIBO, Worlds First Family Robot). In fact one could argue that the lower the security concerns the higher the consumer adoption will be for these types of solutions. One could argue that having a portable camera moving around your home is an ideal security system making your home even safer.
In these Consumer IoT scenarios, there is a common theme of a direct to the enterprise and near-field communication enabled approaches so the local user and larger processes can be triggered and powered. This is of course leveraging local store tracking systems, surveillance equipment, cellular and Bluetooth networks. However, the bulk of theses processes typically resides centrally in a data center as its typically not critical enough for a stores’ operation to host locally.
IoT for Industrials
IoT scenarios that involve industrial assets typically leverage fixed network architectures and are more reliable for the local operations and also have been employed for a long time. However the maturity of the network and the architecture involved to communicate with inter- and intra- network assets are industry specific (Previously I posted this on industrial architectures: SAP MII, HANA/ESP and Lumira (Technical)). Industrial IoT scenarios typically leverage existing network infrastructure and require a different level of security when working in these systems as it is very critical to control access to these systems and does not operate like a normal corporate network does. These networks also tend to use a hub-and-spoke model an accommodate legacy and modern equipment which drives the automation control and monitoring applications. Manufacturers have taken great strides in achieving connectivity to their equipment layer but many still struggle today to manage these systems in a way that a broader IoT Orchestration process could leverage. Simply put, unless there is a concerted effort across the business to expose and manage automation and execution data there will be only very small capabilities for re-use of this data in a generic way by many orchestration processes. Even if there is a point to point integration targeted for process orchestration, this will likely not be sustainable nor is it scalable across an organization and it certainly does not enable a business end-user to discover and leverage this data on-demand and in a personalized way without additional technical support/development overhead.
So what is the right approach then for an industrial corporation and what possible value could they get out of such a scenario given the wide range of varying and heterogeneous data and processes that exist across an organization. For starters the orchestration needs to be centrally managed so that it can be shared across various functional and geographical regions. However, the execution of the various processes does not have to be limited to this single central environment provided there is central synchronization. This distribution of responsibilities with central coordination and monitoring is a well-known concept that is commonly referred to as Ubiquitous Computing but is in principle the same as IoT. This concept of distributed orchestration lends itself well to the industrial space because of the robustness and reliability that is inherent to this design where local systems are less disruptive and mitigate risk when they do fail or become temporarily unavailable (for whatever reason). In this layered model of course the layers can be combined and will often be depending on the size of the organization and processes that are most critical to the business.
Goals for IoT
Since every industrial corporation has some form of equipment in their business the interaction with this equipment needs to be quick, intuitive, and repeatable. As a result the easiest one to point out is the ability to both proactively and reactively manage equipment performance, reliability, and health. This is clearly something that can of huge value to a corporate but how could a business even start to quantify the value of such a program? One easy way is to understand the impact and criticality of each equipment to the operation. Most operations folks can very quickly point out which equipment is crucial to the business but could they tell you the cost in lost production, corporations that could be affected, priority of the fix or replacement? What about quickly identifying systemic production (incl. quality) issues that arise from aging equipment? All of these may seem like a bit of a dream but they are certainly achievable. The real question now is how to go about this?
One key thing that is core to the operations tracking and management is having the context from the central planning and coordination system (ERP) mapped down to the operations level (automation/execution systems) and those systems mapped back upwards to the enterprise. So what does this technical integration actually mean for a business process, it means that any event coming from the operationslandscape can immediately be tied back to a broader process and correlated across a variety of characteristics (ie: Geography, Equipment Vendor, Quality, Process Efficiency, OEM scoring, Energy & Emissions) for patterns. Of course it’s fine to focus on only one of these topics at a time but the ultimate goal should be to combine and reuse the data the same data across each of these verticals.
The big and obvious question here is how can you coordinate and combine this data in a way that makes sense for background analytics and on-demand. Contrary to popular belief the notion of an industrial corporation to copy all data into a single location in order to perform basic reporting should not be a key driver for an IoT solution. The ability to both stream data for analytics but also to retrieve data on-demand from a wide array of systems (sometime validated environments) also has to factor into the deployment strategy. Not only will this reduce the cost of the central warehouse but it will also enable better use of existing systems and prevent “low-touch” data from cluttering the warehouse. The approach of only replicate critical data (Stream or batching) and get the rest of it as needed is a viable and cost effective approach for any business. It also enables a much broader reach for a business user to dig and track down data that may have once been obscure but relevant for their analysis. This can be done because of the common context (ie: definitions) shared across the landscape.
I hope that you enjoyed the BLOG so far and stay tuned for Part #3 that explains various SAP software components and how they fit together to form a cohesive strategy for both Consumer and Industrial IoT scenarios.