On the Internet of Things

April 25th, 2011 Leave a Comment

Editorial note: The following guest post is by John Yopp (@johnyopp) of Oracle’s Retail Applied Research team. John and his team have been doing cool work internally, including experimenting in areas like the internet of things, which is a new favorite topic of mine. You can read more about Oracle Retail and some of the Retail Applied Research team’s escapades over on David Dorf’s (@dordav) blog. Enjoy.

The possibilities of an “internet of things” are becoming more and more relevant to consumers and retailers by the day.  The concept originated with the MIT AutoID Labs and refers to an interconnected web of sensors and chips in everyday items.  In this idea, most objects around us would have chips that could at least identify themselves, and in more complex instances could provide information about the item they are on, the item’s current state and environment, and even some historical state and alerts to changes.  These chips would be able to communicate with each other in a network, likely mesh, and share information between the objects and with other receivers.  As you would expect, radio frequency (RFID) tags are often at the center of these discussions, since those have the most potential to be included in the supply chain and eventually on the items in the store.  However, any technologies that provide item level identification, sensing, and interaction (such as Rubee) could also take part in this network.  Programs like AutoID have taken steps in the supply chain to provide identification and tracking, and are moving towards tags that also sense their environment and can store information.  Once tags are on most items that are sold, while there are significant privacy concerns once the item leaves the store, there are also significant opportunities before that, and we will discuss a small example we prototyped in a moment.

One of the most pressing examples of how the internet of things is gaining momentum is the current market rush to support Near Field Communication (NFC) for tap and go, or chip and pin, payment via mobile phone.  The push is for having NFC, which is a high frequency, low range RFID chip, in your mobile phone for payment, interaction with displays, customer service and loyalty interactions, digital receipt upload, coupon usage, and other functions.  Mobile carriers and banks are positioning themselves to roll this out this year or next, with Isis in the lead so far. What is interesting about the NFC rollout in the context of the internet of things is that the phone would now have an RFID chip, and likely an RFID reader as well.  With those in place the phone can interact with the items on the shelves, or with digital displays and “smart posters”.  The phone can not only identify itself and its user to the network, but gather information from the objects around it.  In retail this translates to personalized promotions on nearby displays, downloading coupons from smart posters with a tap of the phone, filtering items on the nearby shelves based on preferences or on information from your own internet of things (e.g., food allergies, green sourcing, cables that work with your home products, food to replace items that are low in your fridge, etc.), and many other use cases.  Much of this interaction can be done without the consumer doing any significant setup; the functionality and tracing would be automated if they opted in.

One prototype the Oracle Retail Applied Research Team did with the internet of things was using Sun Labs SunSPOT product.  SunSPOTs come with a base station and two sensors, with the base station providing communication between the server and the two sensors.  The sensors have radios in them and are connected to each other and the base station through a mesh network, so sensors not only communicate to the base station but to all other sensors.  Programs run on the SunSPOT are done with standard Java, thus developing for them does not require a specialized skill set.  The sensors come with standard environment sensors as well as open inputs for adding new sensors as needed (for example GPS, moisture, altimeter).  The standard sensors included are accelerometer (telemetry position in 3-D space), temperature, and light. They also have a line of LEDs on them that can shine in virtually any color combination and frequency.  While there are use cases in retail for using these devices for loss prevention and detailed shipment tracking, we will focus on the warehouse pallet tracking use case that our team prototyped.  This prototype mimicked monitoring the movement and temperature of pallets or cases in a warehouse, and assisted in raising pick accuracy.

For this project the concept was to place the SunSPOT sensors on either the outside of pallets or inside high value packages.  The monitoring screen provides feedback per SunSPOT, with each side of the screen monitoring one of the two SPOTs.  The gauges on the top show the movement in G (earth’s gravity).

In the prototype all pallets or boxes were monitored for the following:

In addition, the LEDs on the SunSPOT sensor would flash in matching patterns and colors for pallets that need to be picked together.  In the demo monitoring UI each sensor had a dashboard showing the current state of the pallet.  If it was supposed to be moving to the dock and was not, an alert would show.  If it was moving and was not supposed to be, another alert would show.  If the item was dropped, or opened, or exposed to temperatures outside of the allowed range, other alerts would display.  The SPOTs themselves ran the code to track their environment and how it changed.  They could tell whether movement occurred by establishing their own base line for g forces and then watching for changes.  Brief, high percentage change was considered rough handling, with a graduated rating of how rough.  Prolonged low percentage change was rated as regular movement.  Other projects than this one used the mesh network to have the sensors share this movement data, so that they could tell when packages were being treated differently, and may be either stolen or having other issues.  The power in the SunSPOTs was in having not only an ability to identify an item, but to layer on top of that information about the environment around the item, and then, most importantly, layer on intelligence to track trends and respond to the information it is gathering.  It is the final step of supporting the upload of regular Java code that makes the SunSPOTs so compelling to demonstrate what is possible in the internet of things:  one has the ability to make it an active member of the environment it is in.

The SPOTs could be used for a variety of use cases, including adding sensors for time and motion studies of shopping carts, warehouse forklifts, or trucks in a shipping yard, adding robotics so a single SPOT can be used to check various zones in a warehouse, or add an RFID sensor to expand the number of items for which it can provide sensor information.  Even if out of range of a computer the SPOT could record a full trip within a truck, indicating where it stopped and for how long, when the doors were opened, the variance of temperature within the truck, and where and when and how extreme any rough travel occurred.  All of these use cases are under consideration for future prototype efforts.

The barriers to the further spread of the internet of things are mainly issues relating to cost and RFID limitations.  While RFID tags have come down significantly in price, they still need to reduce further to gain item level penetration.  Even then, managing and sifting the data from so many tags would be a significant task.  RFID also still faces complexity in implementations and problems with signals being unable to penetrate water and steel, as well as signals bouncing.  It remains to be seen whether RFID will move beyond high-value or high velocity items and case level tracking, and if it does not, then what technology may step in.

SPOTs help demonstrate the future potential of the interactivity provided by the internet of things. Concepts like this often seem far away, particularly for consumers, but the imminent advent of NFC in the mobile space could accelerate its reach and impact.  Oracle Retail is actively researching the current state and future of RFID and NFC, within the broader lens of the potential that the internet of things provides an RFID implementation.  Potential future and current prototype and research projects on this topic from the Applied Research Team include an RFID whitepaper, augmented reality views of the store, and further implementations of SunSPOT functionality.  Watch this space.


Possibly Related Posts

Leave a Reply