Introduction:

 In this second article on TiECon 2014, we examine the key issues, technologies, standards and technical challenges for the Industrial Internet and related Industrial Internet of Things (IoT) as discussed in the Infrastructure for IoT panel session.  Note that the IoT is referred to as IoE by Cisco and also by Qualcomm.

Infrastructure for (Industrial) IoT Session Abstract:

The key components required for the Industrial IoT were detailed in this panel session.  Many questions were raised:

• Should all data be sent from devices to the cloud or is some form of intelligent decision making needed at the network edge?  
• What Physical layer interfaces are needed in IoT devices?  Note that there are many physical media types, from various types of Ethernet on shielded and unshielded twisted pair to IEEE 802.11 WiFi, IEEE 802.15.4 Zigbee, Z-Wave and EnOcean.  Will this “alphabet soup” of acronyms be dwindled down to a few Physical layer interfaces that most IoT endpoints will support?  
• Similarly, there are many services layers, from various flavors of ZigBee profiles, to legacy industrial protocols like LonWorks, BacNet, SCADA, DALI and many more.  And there are many management platforms in the works. 
• What’s needed for IoT security?  
• Given the diversity in the technical ecosystem, do we need to get to a common set of Industrial Internet standards, or is that not possible?  

Moderator:   Paul R. Teich, Moor Insights & Strategy 

Panelists:

• Carl Stjernfeldt,  Shell Technology Ventures
• Joseph J. Salvo,  Complex Systems Engineering Laboratory at GE Global Research
• Nate D’Anna, Cisco Systems

A few introductory remarks worth noting:

• Shell is using the IoT in oil drilling operations where “last mile” connectivity in the ocean is quite challenging –  9K feet under the surface of the ocean.
• GE is buliding out an “Industrial Internet,” which is a virtual network overlay on a physical network.  
• A consortium of companies (AT&T, Cisco, GE, IBM and Intel) was launched in late March 2014 to foster interoperability in the Industrial Internet.  Goals are to identify requirements for open interoperability standards and define common architectures to connect smart devices, machines, people, processes and data. 
• Cisco has a $100M venture fund to invest in the IoE and move it forward.
• There’s a huge need to get real time data from a variety of complex industrial sub-systems (e.g. a jet engine) via satellite, microwave, fiber optic long haul or short range WiFi/Zigbee wireless networks.
• Trade-off between bandwidth vs reliability is very important for the Industrial Internet and ruggedized IoT applications.

Several important questions need to be addressed for the Industrial IoT:

• There’s a lot of IoT data collected for compute/storage.  How much is actually needed and how should data decision making be made so that not all data collected is sent from industrial premises to the cloud?
• Where do IoT standards take over from customized, vendor specific solutions?  
• How will minds and machines be connected in the Industrial Internet? 

Here are the panelists opinions on these and related issues.

Each layer of the protocol stack will need flexibility to accomodate measuring, monitoring, analytics, automated decision making in the field, and other tasks.

Shell colects a tremendous amount of data in the field, but uses “intelligent pushback” to get large amounts of that data processed, analyzed with the results coveyed to a decision maker in the field (presumably, a human operator).

GE sees a strong trend of “data commoditization and lower level knowledge.”  Today, everyone has access to lots of data and information.  Future market and technical differentiation will be centered around decision making, learning what data to forget/discard and what to retain/remember.  

Managing the onslaught of incoming real time data (monitoring status, alarms and measurements) will be critical in the industrial world.  Smarter and faster edge devices that monitor complex real time systems and make decisions what data to retain and what to push back (to the cloud/data center) for number crunching.

Cisco sees a huge industrial customer demand for collecting large amounts of data.  They believe decision making is needed at the network edge – to only send relevant data to the cloud for processing.  “Make decisions locally, by smart devices at the edge of the network,” said Cisco’s Nate D’Anna.

GE sees the context of data becoming more important than the data itself, said Joseph Salvo of GE’s Complex Systems Engineering Laboratory.  There’s a need for data storage centers to only retain relevant data from connected devices with most of it discarded at the network edge.  This seems identical to Cisco’s view expressed above.  

Mr. Salvo sees a hyper-connected world with a changing role for computation and big data.  Analyzing vast amounts of data, predicting and changing system performance are the key attributes of this new computational model.  “We are at an inflection point. The next wave of productivity will connect brilliant machines and people,” he said.

Security is certainly a critical issue.  It must take into account the hardware, software, network, devices/components as well as the human factor, according to Salvo.  It must recognize the context of what’s happening in the industrial systems being monitored/controlled.  Note:  No security standards, methods or procedures were identified.

Standards help market adoption as they facilitate interoperability, mass production and lower costs.   However, there are a plethora of wireless and wire-line connectivity standards which might have to be supported for different IoT market segments.  This may create a “matrix of pain,” said one panelist.   It could be that IP (Network Layer protocol) will be the lowest common denominator that all Internet connected “things” support.  Speaking for GE, Salvo said: “The Industrial Internet wants interoperability (over vendor specific solutions).  Like USB I/O devices that auto-configure and work.”  Joe suggested that the (Industrial Internet) market will determine the best standards.  

Surprisingly, there was no discussion of the need for a robust network (for the Industrial Internet/IoT)  with lightening speed fault isolation and failure recovery.  Something that the public Internet often doesn’t provide.

GE’s Jeff Immelt:  GE believes that the Industrial IoT will accelerate the push towards Big Data.  “Industrial data is not only big, it’s the most critical and complex type of big data,” said Jeff Immelt, GE Chairman and CEO. “Observing, predicting and changing performance is how the Industrial Internet will help airlines, railroads and power plants operate at peak efficiency.”

Note on Future TiECon 2014 articles:   

Two TiECON IoT keynotes will be summarized in future articles are Proximity’s Role in the Internet of Everything (IoE) by Qualcomm and What Really Matters for IoT by IBM.