Big Data // Hardware/Architectures
Commentary
6/26/2014
11:06 AM
Cristian Borcea
Cristian Borcea
Commentary
50%
50%

Internet Of Things: New Architecture Needed

Today's Internet architecture isn't ready for the scale of what the IoT should become. Here's a proposal for something better.

The vision everyone keeps touting for the Internet of Things, where billions of devices are sharing information and completing tasks to improve the efficiency of daily life, relies on access to a planetary-scale Internet. Unfortunately, today's architectures can't handle the needs that the Internet of Things will demand.

In the near future, the number of devices connected to the Internet of Things (IoT) will reach 26 billion devices, according to Gartner, and produce $300 billion in revenue. Some of the devices that will participate in the IoT are sensors and actuators in smart cities, smart tags on many familiar objects, wearable health monitoring sensors, smartphones, intelligent cars, and smart home appliances. In the not so distant future, IoT will incorporate many types of robots -- domestic, flying drones, and even bee-size flying robots. The changes to our daily life will be immense. For example, IoT will lead to a more precise and greener management of power and water distributions in smart cities and improvements in healthcare such as stopping the spread of epidemic diseases.

However, today's technology isn't ready for the massive scale and highly dynamic nature of the future IoT, the huge amounts of data streamed from the physical world, and the new communication patterns it creates. We need novel programming, content delivery, and network management approaches. What follows is a proposed, still-developing framework for such a global, Internet of Things architecture.

One major problem with the current IoT architectures is they're designed for relatively small scale IoT islands -- closed-looped networks, such as a power plant operator pulling data from a turbine -- under proprietary protocols. Densely deployed "things" can't collaborate dynamically across these "islands" to execute distributed tasks that involve sensing, actuating, and computing.

So, how can we better facilitate the current state of IoT to reach its full potential? We must rethink the intelligence embedded in the IoT architecture.

[What about security? Read IoT: Get Security Right The First Time.]

In collaboration with the National Institute of Informatics (NII) in Japan, I've been researching how to leverage cloud computing technologies and software defined networks (SDN) to promote effective and efficiently distributed IoT services. This investigation has led me to believe that the future of IoT should evolve organically on top of the existing Internet and implement a novel distributed intelligence architecture called 3DIA -- a three tiered "distributed intelligence architecture."

3DIA is based on three design principles:

  1. Distributed execution of IoT services
  2. IoT networking inspired by SDN principles
  3. Pervasive cloud-based support for IoT

By implementing 3DIA, Internet of Things devices will be able to execute distributed sensing and computing services with help from the cloud, which can address the problems related to availability, resource limitation, bandwidth, latency, and management.

3 tiers of a gloabl network
3DIA is broken down into three tiers: global intelligence, regional intelligence, and local intelligence. Existing efforts have tackled Internet of Things issues at the regional and local tiers.

Global: The top tier provides global intelligence in the cloud for scalable services, network management, service programming, and device interaction. For example, this tier would allow for global coordination among IoT devices in a large city (e.g., assets belonging to public utilities, various city agencies, and individual users) and for efficient distributed service execution. Currently, this is not possible because different IoT networks don't interact with each other. Furthermore, the global tier would simplify the interactions between users and services by providing a highly available service end point in the cloud. The problems inherent to the distributed nature of services, such as device failures or disconnections due to mobility, would be transparent to users who interact with this service end point.

Regional: The middle tier provides regional intelligence to effectively handle the IoT dynamics, network traffic engineering, and wireless resource allocation, which is implemented at wireless access networks that are enhanced with computing and storage resources to form a new type of pervasive cloud infrastructure We call these "stratus clouds." This tier would, for instance, provide effective management of location dependent resources. One typical example of such resources is scarce radio resources, which have a strong dependence on location. In this case, the stratus clouds will run software defined radio (SDR) protocols that help with radio spectrum allocation. For example, this could be useful in a disaster scenario where many users will crowd certain regions trying to evacuate and efficient spectrum allocation is necessary to avoid interference. The stratus cloud would enable the deployment of this type of regional IoT management software.

Local: The bottom tier provides local intelligence, which consists of software running on IoT devices to let them interact with the stratus clouds and with other nodes to deliver scalable IoT services as well as to interact with the whole cloud infrastructure for efficient networking, content delivery, and wireless resource allocation. This last tier consists of elements in the physical environment, such as smartphones, intelligent vehicles, and computers. One example of its functions is intelligently forming ad hoc networks among IoT devices when such networks are needed in situations such as natural disasters.

We're currently working to build a prototype of 3DIA. To succeed, this architecture needs SDN controllers in the global cloud and the stratus clouds to provide networking support. SDN controllers will handle all decisions related to device to device communication -- such as routing, traffic scheduling, and ad hoc vs. Internet communication decisions. Because SDN controllers are daemons/processes running in virtual machines in the cloud, they can be easily scaled up or down and moved between different cloud locations to follow the potentially mobile IoT devices or users. Ideally, IoT scalability could be preserved even for a substantial number of devices.

Questions we can't answer
In my research, there are still questions that we haven't answered.

For example, how can the three architecture tiers function seamlessly across different radio access technologies, while minimizing interference and power consumption?

Can we create smarter ways for devices to spontaneously connect over IoT architectures, so that dynamic IoT services could be created on demand?

How can we share devices concurrently among multiple services while giving priority to real-time services without delaying the others indefinitely?

And how do we handle privacy in a massive IoT, composed of devices belonging to many organizations and individuals?

Though we're still searching for answers to fulfill the proposed vision of the 2020 IoT, where billions of devices are communicating seamlessly and solving many of the inefficiencies of daily life, it's clear that a more intelligent IoT architecture is needed. Our approach is to embed more intelligence in the architecture in the form of cloud resources and software defined networking.

Like the smart cities we hope to enable through IoT, 3DIA can be incrementally deployed over the current Internet. Thus, it could be readily prototyped and widely adopted by industry. We believe this goal is within reach.

In its ninth year, Interop New York (Sept. 29 to Oct. 3) is the premier event for the Northeast IT market. Strongly represented vertical industries include financial services, government, and education. Join more than 5,000 attendees to learn about IT leadership, cloud, collaboration, infrastructure, mobility, risk management and security, and SDN, as well as explore 125 exhibitors' offerings. Register with Discount Code MPIWK to save $200 off Total Access & Conference Passes.

Cristian Borcea is an Associate Professor and the Associate Chair of the Department of Computer Science at New Jersey Institute of Technology. He is also the program director of the online MS in Computer Science at NJIT (http://computerscience.online.njit.edu/). Cristian also ... View Full Bio
Comment  | 
Print  | 
More Insights
Comments
Threaded  |  Newest First  |  Oldest First
zaious
50%
50%
zaious,
User Rank: Ninja
6/26/2014 | 1:36:31 PM
Another Change for us
IoT is the new wave that will make us undergo another change. I liked the 3DIA -it's design priciples are few yet to the point. I am also worried about the traffic that will flood the networks.
Stratustician
50%
50%
Stratustician,
User Rank: Ninja
6/26/2014 | 1:54:30 PM
Re: Another Change for us
Nice to see some thought going on about how to better structure IoT communications.  I think this is a great start to build more specific standards around.  Perhaps data priority could also be utilized as a factor, so that critical infrastructure would get different priority levels than less critical data flows (might help with the sheer volume of traffic as well!).
Hospice_Houngbo
100%
0%
Hospice_Houngbo,
User Rank: Strategist
6/29/2014 | 8:41:46 PM
Re: Another Change for us
Will the existing protocols be able to protect the new attack possibilities brought about by IoT? Is there any best approach based on today's technology to protect IoT or new strategies should be implemented for IoT to be viable?
ADDC
50%
50%
ADDC,
User Rank: Apprentice
6/26/2014 | 9:26:15 PM
Why continue flogging a dead horse?
It is interesting to see the continued obsession in the commercial IT world with attempts to expand and implement communication models based on philosophy increasingly removed from the operational principles of the natural world. The complexity of interaction required for support of an efficient IoT will rapidly break the systems currently being proposed as suitable.

No amount of "cobbling together", redrawing, or creation of "levels" will transparently resolve this problem. What is required are models which better match reality and its requirements. Centeralized hierarchical cloud computing technologies are not a solution as they lie within an aging and nearly defunct maladaptive model and framework (that clearly does not cope). Only a fundamental (philosophical) reconfiguration will lead to the auto-adaptive technology of distributed content management systems the IoT requires.
ChrisMurphy
50%
50%
ChrisMurphy,
User Rank: Author
6/27/2014 | 10:52:42 AM
Re: Why continue flogging a dead horse?
Very interesting ADDC, though I'll confess I can't picture what you envision. I felt like Cristian here is getting at this kind of auto-adaptive, distributed content management that you describe. Could you offer any examples how the reconfiguration you envision differs from today's?
ADDC
50%
50%
ADDC,
User Rank: Apprentice
6/27/2014 | 9:48:18 PM
Re: Why continue flogging a dead horse?
Thanks for noticing although nothing in my comments is particularly new. But to elaborate a little in a more abstract domain . . .

It is not unreasonable to suggest that we relate to the world in terms of models of the world that we create cognitively, based on what we have experienced and learned. The better the match between the cognitive model and reality, the greater the efficiency, more rapid the evolution, and greater the possibility of survival and growth. In short, the better the model, the better the predictibility and the flow-on of advantages, whether unforeseen or planned.

Computation and its relation to the internet has reached a complexity which suggests it should be considered more seriously for what it is or rather what it will become, particularly when conceived of as "infrastructure"; rather than being ad hoc expansion based on corporate interest and its drive for product advantage. This drive creates a situation where reality is not well matched to any particular model or theory (if one can even exist within such a "reality bubble"). This in turn drives reduced integrability and its many negative consequences.

One reason for such problems is our hierarchical interpretation of the structural organization of the world. This is founded on and has arisen through a long interaction between theology, philosophy, and (subsequently) science, which provides a deeply embeded model of the world, its contents, and relationships. It has served well through history while populations and their technological and social sophistication existed in relatively isolated pockets. However, our technical ability to create is beginning to far outstrip the applicability of this model and its ability to organize and control ever-increasing complexity. Particularly as populations and their technological underpinnings become more integrated.

One of the most powerful drivers of the hierarchical world view or model is likely the perceived benefit of simplicity of control. It is for this reason that the current developments beyond the single computer to the implementation of cloud computing has taken the path it has. But it is the highly centralized, layered, and in principle hirarchical organization of cloud computing and its associated technologies that is precisely the reason why it's uptake is impeded. It does not provide a good model of the world, people intuitively know it, and consequently are reluctant to participate. One significant reason for this is that in acquiescing to the hirarchical model, control must be relinquished to a higher level of the hierarchy. So, for example, in adopting the cloud, data and/or software control is typically lost and legalities blurred. Whether real or not, there is at least a psychological sense of loss of control for the client/user.

The problem for the IoT is that the orgainzational philosophies required for its successful implementation are even more strongly opposed to a paradigm of control through hierarchical organization than those employed to create and implement cloud computing functionality. Currently, the solution of truely distributed interactional nodes independently responsible for data retention and interpretation is likely still a step too far. (We already seem to have enought trouble with our cell phones, social networking, and personal computers.)

Once a model has been developed to give a better match with reality, appropriate solutions either present themselves or are readily identified (remembering they can only be as good as the model).

Although it is possible to say much in considerably more detail about solutions and implementation, I guess we must leave it here for the moment . . . only to note Cristian's approach as representative of IoT appears to originate from what might be thought of as an increasingly maladaptive (hierarchical) philosophy and for this reason alone is likely to be either highly inefficient or, worse(?), doomed to failure . . .

 
ChrisMurphy
50%
50%
ChrisMurphy,
User Rank: Author
6/30/2014 | 9:28:42 AM
Re: Why continue flogging a dead horse?
Thanks for the perspective, ADDC.
Joe Stanganelli
50%
50%
Joe Stanganelli,
User Rank: Ninja
6/27/2014 | 6:30:16 PM
Iot
Good luck with that.  We can't even get IPv6 deployment -- a necessary for IoT -- going years after the doomsday date.

Personally, I am not particularly bothered by this stuff anymore because the architectural and infrastructural walls that stand in the way of an IoT world will maybe slow IoT development down enough to get the idiots who make IoT products thinking about security -- because clearly, right now, they don't.
Hospice_Houngbo
50%
50%
Hospice_Houngbo,
User Rank: Strategist
6/29/2014 | 8:59:57 PM
Re: Iot
@Joe: I am not that pessimistic because IoT is already a reality despite the slow pace of IPv6 adoption. When IPv4 addresses are gone, Internet service providers will not have another choice, but to turn IPv6 capability on.
Joe Stanganelli
100%
0%
Joe Stanganelli,
User Rank: Ninja
6/29/2014 | 10:48:34 PM
Re: Iot
@Hospice: IoT has already started, sure, but if and when doubling-up (and worse) of IPv4 addresses begins to occur should adoption outpace IPv6 deployment, serious security, privacy, and bandwidth problems can result.
In A Fever For Big Data
In A Fever For Big Data
Healthcare orgs are relentlessly accumulating data, and a growing array of tools are becoming available to manage it.
Register for InformationWeek Newsletters
White Papers
Current Issue
InformationWeek Tech Digest, Dec. 9, 2014
Apps will make or break the tablet as a work device, but don't shortchange critical factors related to hardware, security, peripherals, and integration.
Video
Slideshows
Twitter Feed
InformationWeek Radio
Archived InformationWeek Radio
Join us for a roundup of the top stories on InformationWeek.com for the week of December 14, 2014. Be here for the show and for the incredible Friday Afternoon Conversation that runs beside the program.
Sponsored Live Streaming Video
Everything You've Been Told About Mobility Is Wrong
Attend this video symposium with Sean Wisdom, Global Director of Mobility Solutions, and learn about how you can harness powerful new products to mobilize your business potential.