Home Internet of Things Making Sense of IoT
IoT

Making Sense of IoT

0
1457
Thumb1

The more open a system is, the more it enables innovation and engenders trust. Yet, when it comes to open systems, the question that should be asked is ‘how open can we make it?’

Being open doesn’t mean that everything needs to be open. Much like the web, the more you are open near the bottom, the better, and more those layers become embraced by others:

  • Trust: Open at the bottom systems generate trust. Companies are not worried that you can block them out so they see little risk in joining.
  • Reach: By building an open base layer, you create an open playing field that everyone joins, allowing others to build products for corners that you don’t have the energy for. This avoids the potentially dozens of silos that choke network effects.
  • Momentum: By having great reach you build market momentum, everyone joins because ‘everyone else is joining”
  • Longevity: Open at the bottom platforms encourage more experimentation, creating new extensions and experimentations, much like the web grew by adding Javascript and CSS.
  • Product Control: Yet, nothing should be stopping companies from building their own proprietary agents. Companies can also build their own proprietary products on top of these open base layers.

The purpose of building an “open at the bottom” system is that it creates the trust, reach and momentum that allows a vibrant platform to be created. These layers all have a few common open approaches holding them together:

Open Standards & Protocols

These bottom layers, much like HTTP and the web, are fully open and based on standards. This is the ‘level playing field’ that gives everyone the confidence that lots of things can be tried.

Shared Conventions
This is the more experimental layer where competing conventions, much like schema.org, can evolve shared conventions

Competing Products
At this layer, self-contained products, that aren’t open at all, can be build. Not that alternatives can exit. While they are proprietary, they don’t lock out competition. This is clearly a risk but it’s critical to create a space that encourages others to join.

No single firm can provide all of the services necessary to make the IoT a reality. Apple built a closed, proprietary system around the iPhone and the manufacturers, developers, and carriers who were not part of that system and who needed an alternative platform. Today, the situation is somewhat different: there are many small players and a few large ones. Some of the challenges that need to be overcome are:

  1. Managing an open source core evolution while avoiding fragmentation/forking: This is a perennial problem with open source systems in a commercial environment. The basic problem is a prisoners’ dilemma. Total value is higher if everyone adheres to the standard, but individual parties may want to defect from the standard in order to differentiate their product and gain a competitive advantage.
  2. Cross-platform development tools: Just as Apple was first to market with the iPhone and quickly built a developer network, Amazon is first to market with the Echo. The goal should be to make it as easy as possible for services to port services available. This may result in some redundancy in the APIs, but that is a small sacrifice to make for to get the developers onboard.
  3. The role of software and hardware developers: Companies should make it easy for partners to join the ecosystem by providing toolkits, documentation, examples, and training. This involves a lot of effort and example code. Initially, it is probably necessary to work closely with a limited number of high-value partners, but they should broaden the developer universe as quickly as possible. Hardware development is, of course, different from software development. From a realistic perspective, most hardware devices that comprise the IoT will be developed in Asia following the Apple “factoryless goods producer” model, since that’s where the supply chain is focused.

Connecting literally everything is a bigger task than any one company can ever achieve. Companies must deeply rethink what it means to be connected. Not just technically, but also politically. They must break out of this ‘winner take all’ mentality that has hobbled the IoT industry to date. Actually “thinking differently” and creating an open system will earn companies the right to lead the world in connecting everything together.

Related Article:   All the best AI conference sessions at SXSW

A company should start with the inherit user/customer focus and sharing ML (machine learning) as the network-effect driver, but then focus enterprise IoT where the core value proposition becomes the ability to manage, process, and to analyze large data sets to make better business decisions.

In order to create value for businesses, the following points shall be taken into account by companies for conducting a business analysis:

  • Creating Value for Users (or improving the human condition): The IoT ecosystem optimization function should be aimed at producing maximum value for users. While understanding all stakeholder interests would be beneficial for an enterprise, an organization’s excellent relationships with its users is the only lasting source of value. While the belief that dramatically increased connectivity between individuals is enabling humanity to become a powerful ‘collective intelligence’ in certain respects, this connectivity is consistent with increased, not decreased, individual freedom and empowerment. The approach offered in this book allows to offer novel, value-added services to augment the user’s ability to make sense of his/her context and translate it into enhanced agency.
  • Appealing to all Stakeholders – Win: In multi-sided markets such as IoT, all stakeholders should benefit from solutions. Users, platform providers, hardware producers, service and maintenance companies and many more should each benefit.
  • Balancing Stakeholder Interests (Fair and effective Governance): Who can participate under which rules will be a major factor influencing competition between IoT ecosystems. The rules of the ecosystem should be transparent, balanced among stakeholder interests, and continuously updated based on new data and insights. That said, the IoT ecosystem must always remain a platform for permissionless innovation. It is the platform providers’ responsibility and opportunity to expand the certification programs and partner policies to shape a fair and effective ecosystem.
  • Ecosystem Openness: Nearly every IoT system to date has been a closed silo. If the IoT is going to encompass nearly every sector, it must be appropriately open at the device, control, and data layers. This clearly has some business risks, but the ecosystem will only take off if a system is built which encourages user trust, reduces 3rd party product risk, encourages cooperation among participants, and provides incentives for strategic partners to participate enthusiastically.
  • MI (Machine Intelligence) as an IoT Network Effect generator: Widespread IoT adoption will only happen when network effects drive the utility of connected objects. If an organization has strength in MI that would be an excellent starting point as the focus can be on the design to maximize generation of contextual data (sensing/controlling).

The vision should be to provide a suite of products and services that enable best-in-class Enterprise IoT solutions that provide:

  • Scale: A Cloud Platform can scale up to manage billions of devices.
  • Security: From device protocol/OS to cloud services, a company can provide an end-to-end secure and updatable model.
  • Data management: Cloud data products provide a cutting edge platform for analyzing and working with IoT data, and can scale to meet the data and processing requirements of the vast majority of enterprise use cases.
  • Machine learning (ML): In the cloud and on the device, advanced analytics/ML and controls to drive better outcomes
  • Faster IoT deployments: Plugin existing devices and other clouds or take a preconfigured “IoT in a box” with defined device semantics, data structure, and analytics and ML services – with clear ROI (Return of Investment)

Companies can really attract and shape the enterprise IoT market once they can apply the learnings and technologies we develop to integrate 1st party devices and IoT interactions with our cloud infrastructure such as:

  • Sorting out the semantic graph of the world of devices, their schemas, and UX (User Experience) interactions
  • Creating and exchanging private sensitive data and building security models and access scopes to support secure data sharing
  • Building a secure updatable device to device, device to cloud application-level communication protocol for things
  • Enabling close integration between IoT devices and the Cloud enterprise data platform, so that device data becomes just another data source an enterprise customer can access for business analytics and product development.
  • Building, tuning, and delivering across a fleet of devices ML models catering to details from the physical world based on above schemas
Related Article:   Blockchain: The Most Useless and Overhyped Technology in History?

The development and subsequent transfer of methodologies, abstractions, and technologies built to develop the company-specific branded IoT experience need not be strictly sequenced. Key home device partners and their hybrid needs, or split consumer/enterprise product lines, can serve as a great substrate to explore this transition with partners through bespoke developments on Cloud along with jointly developed pilots and specific vertical solutions.

There have been multiple attempts in the IoT arena to build consortia or partnerships that intended to create a larger ecosystem than any of the constituent companies could do alone. They were, however, largely blueprints for larger silos, rather than attempts to create truly open ecosystems. To achieve that aim, companies will need to foster partnerships where our peers share the vision of an open ecosystem. They will also need to confirm their intentions by working with open standards organizations to create specifications which are more broadly appealing. This does not mean that companies should just wait for the standards to be created. They should create interoperable protocols, schemas, security methods and UX elements now; the contribution to standards processes can proceed in parallel or follow later.

The key to platform adoption is to showcase valuable user experiences that may be seen as baby steps before the companies can pursue truly magical experiences. Therefore, an approach based on cloud-hosted orchestration (as discussed in detail in the following chapters) is well-aligned with those near-term objectives. However, in the longer term, the distribution of IoT intelligence over ensembles of devices would be highly desirable. To understand the rationale, let’s first categorize the purpose of IoT intelligence into four roles:

  1. Automate the things that users don’t want to do:

    • Turn off the lights at bedtime.

  1. Enable the doing of things that users want to do but cannot:

    • Turn on the thermostat just long enough before I get home.

  1. Suggest the doing of things that users should do or may want to get done:

    • Don’t forget to lock the front door.

  1. Discover new things that users may want to do or get done:

    • Lower the brightness and blue-light levels before bedtime.

The first three roles above roughly correspond to those of a butler, a concierge, and an advisor. Along those lines, the fourth role could be considered that of a guide. Table 1 provides a visual summary of these roles.

Table 1. The Roles Regarding IoT Intelligence

Over the next few years, as all of the above intelligence roles become more ingrained in the daily routines of users, they will come to see some IoT intelligence-enabled features as must-haves and some others as nice-to-haves. Furthermore, users will expect that the must-have features are functional even when their devices/homes are disconnected from cloud-based intelligence services. It will then be critically important that such features be implemented through the distribution of the intelligence locally within the home.

One of the main benefits of distributing intelligence in such a way is spontaneity. Watching a performance of the New York City Ballet is sure to leave you impressed, while the performance of an improv comedy troupe such as the Upright Citizen’s Brigade would probably leave you exhilarated. There is something inherently exciting and human about not being able to predict what will happen in a given circumstance.

Whether or not spontaneity is an explicit goal, in order to give IoT devices a “life of their own,” companies should seek opportunities where they can establish the rules of the game and let the users and their IoT devices play the game. Including human-beings as triggers or decision-makers adds elements of uncertainty in the emergent behavior, which can help keep users engaged. However, the rules of the game should establish constraints that prevent the core requirements of the user – comfort, safety, security, etc. – from being compromised.

The required architecture is akin to an IoT nervous system — a cloud-hosted brain enabled by a plethora of device-resident sensory and motor functions. The goal with distributed intelligence should be to essentially reverse the roles such that the end devices exhibit sophisticated intelligence with the cloud playing a supporting role. One way to visualize the end state is akin to the creation of a family — parents who nurture children until they are knowledgeable enough to become independent.

Related Article:   The Age of Edge Computing

One of the promises of the Internet-of-Things (IoT) is that devices will to be smart. In other words, they can interpret commands, or data, in their current context and modify their behavior accordingly. For example, a traditional ‘dumb’ garden watering system will use a timer to turn on at the same time everyday, rain or shine. However, a smart watering system will receive the same timer command but save water either by querying ‘employ a sensor’ to determine if the grass is already damp enough, or check an Internet weather forecast to figure out if it’s about to rain.

Context is a term that has been used in many disciplines with differing meanings, so before going further we will define what it means here. Context can be thought of as a set of five properties:

  • Location (where),
  • Time(when),
  • Identity (who),
  • Environment (what now), and
  • History (what has passed).

Let’s try to visualize such contextual factors:

The more devices have access to context, the more they are able to interpret their environment, and the intent of user events and actions. However, context information generated by a wide variety of devices can only be shared for maximum effect if there is a common format/representation.

Context provides an IoT system with the ability to make sense of user-generated events and actions. This becomes even more important when the context is aggregated in a cloud service to learn common behaviors and the associated actions expected by users in the general population. The larger the data sample, and the richer the context data, the more likely we will be able to apply machine intelligence to automate actions correctly. It is well known that making a small number of errors among many correct ones, is still a huge annoyance for the user; and this is a challenge IoT must face. Ironically, companies may only be successful if IoT is already at a significant scale. Product excellence may drive companies thereby keeping users in the decision loop for most of this growth, and only transitioning to full automation when products have a suitably large user base.

For IoT devices to be ultimately successful, they must share context between themselves and the cloud. Unfortunately, today IoT is creating an inordinate amount of information that is not represented in a sharable form (silo’ed and proprietary).

If we expect IoT devices to be smart, they must not only adapt to their current environment but also learn about the habits and preferences of their users. The context of every user action is an essential input to a machine learning process and the more accurate and precise that information is, the better the learning will be.

Some tech giants in Silicon Valley already have access to many aspects of a user’s personal context through location and search history (if allowed by the user). This can result in an understanding of 1) current preferences, aspirations, and interests, 2) social context, who companies care about, 3) what others have done in a similar context, and 4) capabilities, what routines they are engaged in.

It might be possible that in less than a decade most individuals will interact with their personal devices hundreds of times per week, and users will expect to be able to get information about, and control, most of the things they interact with. Who knows how the future of IoT might look like?

Previous articleThe Role of Quants on Forex Trading
Next articleThe Truck Driver Shortage
Ayse Kok
Ayse completed her masters and doctorate degrees at both University of Oxford (UK) and University of Cambridge (UK). She participated in various projects in partnership with international organizations such as UN, NATO, and the EU. She also served as an adjunct faculty member at Bosphorus University in her home town Turkey. Furthermore, she is the editor of several international journals, including those for Springer, Wiley and Elsevier Science. She attended various international conferences as a speaker and published over 100 articles in both peer-reviewed journals and academic books. Having published 3 books in the field of technology & policy, Ayse is a member of the IEEE Communications Society, member of the IEEE Technical Committee on Security & Privacy, member of the IEEE IoT Community and member of the IEEE Cybersecurity Community. She also acts as a policy analyst for Global Foundation for Cyber Studies and Research. Currently, she lives with her family in Silicon Valley where she worked as a researcher for companies like Facebook and Google.

LEAVE A REPLY

Please enter your comment!
Please enter your name here