A Talk with Geoff Mulligan of the IPSO Alliance


Date: September 7, 2009

line break image

The IETF Journal sits down with Geoff Mulligan, chair of the newly formed IPSO (IP for Smart Objects) Alliance, to discuss why he believes that IP offers the most promising means for connecting smart objects.

IETF Journal: What was the motivation behind the formation of the IPSO Alliance and what is the organization’s mission?

Geoff: A little more than a year ago, a number of companies were working on smart objects and what is referred to as the Internet of Things. Unfortunately, nothing was being done to promote the use of IP as a viable protocol for connecting smart objects. In fact, the market was fragmenting into many proprietary ad hoc solutions for connecting these devices, with failed attempts to couple the solutions with Band-Aids of gateways. Those actions only created complexity and instability, and they decreased security. This was happening mainly because people didn’t know about the work being done in the IETF and the applicability of the IP protocol suite to this space.

We started the alliance to bring together a critical mass of companies that would create the right ecosystem for this type of work. Then we could say that all of those companies-the chip vendors, the software vendors, and the end-system vendors-believe that IP is the right way to go when it comes to connecting smart objects and sensor networks. Once we established that, we could move toward education by writing white papers, conducting tutorials, and participating in meetings that promote the concept. So, first and foremost, the IPSO Alliance is there to market the idea that IP is the best possible protocol for connecting smart objects.

The second part involves technology demonstrations and technology support. In other words, let’s show how this actually works; let’s actually build it. Let’s show that, for instance, we can build an IPv6 battery-operated sensor the size of a quarter and put it on the Internet so we can dispel the myths about how small these objects can be. On the technology side, we want to help facilitate interoperability testing and interoperability events. We’ve done a couple of interop events to show how objects are plugged together-or, nowadays, unplugged, because a lot of the technology today is wireless.

IETF Journal: Can you briefly describe what you mean by the Internet of Things?

Geoff: I think of the Internet of Things as that edge between human existence and technological existence. There are myriad things that people interact with on a daily basis, and once they have an IP address, they become part of the Internet of Things.

IETF Journal: Is the Internet of Things connected to the Internet that we know today?

Geoff: That’s a really good question. The answer is yes and no. Just because something runs IP doesn’t mean it has to be connected to the Internet. With the right security and management there is no reason why not, if you want or need to interact with other devices on the Internet. Without the right security and management, even though there may be a device that you’d like to connect to the Internet, you may not want to.

On the other hand, the Internet of Things is called that because it uses the Internet. However, there will be smart objects that use IP but are not connected to the Internet, and sometimes they’re not connected for reasons of privacy, safety, or security. For example, the brakes on your car could have an IP sensor, but do you want anyone else on the Internet pinging that sensor?

IETF Journal: What is a smart object?

Geoff: The IPSO Alliance defines smart objects as devices that are a combination of sensors, controllers, or actuators with computational capability and some communication capability that enables them to be connected to a wider array of networks and devices. IP smart objects are those whose communication capability happens to be, or should be, IP.

IETF Journal: How is the work of the IPSO Alliance related to the IETF? And what kinds of work are you personally doing with the IETF?

Geoff: I’m cochair of the IETF 6lowpan working group (WG).
Very early on, when we defined what the IPSO Alliance is, we also wanted to define what IPSO isn’t; and what it isn’t is yet another alliance trying to masquerade as a standards organization. We wanted to make sure it did not look like we were competing with the IETF. In fact, what we are doing is promoting the standards that come out of the IETF along with standards that come out of other standards organizations, such as IEEE, W3C [the World Wide Web Consortium], and IEC [International Electrotechnical Commission]. We want to demonstrate how these standards can work together, and how it actually works-particularly when it comes to very small components, such as temperature and motion sensors in a home or building talking to thermostats to improve energy usage or to smoke detectors communicating with gas appliances to improve home safety. In addition, we want to help provide a translation between end-user requirements and IETF techspeak. That way we can come to the IETF and participate in the working groups and say, here are the issues the IETF needs to work on; these are the holes that need to be filled.

IETF Journal: What protocol work currently being done in the IETF is important to the IPSO, and when would you like to see those specs completed? Did you identify things the IETF needs to be doing that have not yet been done?

Geoff: The IETF working groups that currently are of interest to IPSO are 6lowpan, roll, manet, auto-conf, mipv6, and mobileIP.
The work being done in 6lowpan and roll is critically important. If we’re talking about trillions of devices, we need IP addresses. The IPSO Alliance is promoting IP as the protocol that runs to the very edge of the line that separates the network from actual physical work; there are no gateways and no translation in between. That means that IPv6 is the only viable alternative. The IPv6 Maintenance WG, known as 6man, is important because if you need to manage trillions of devices, this is not something you want to do manually.

We are also interested in transport protocols because we do not want to skip any portions of the stack. Therefore, we need to understand how TCP or UDP really works for these small embedded devices.

What are holes? What are the things we don’t know? We believe that UDP is good enough for many applications, but we don’t know if it’s good enough in all situations. It may well be that we need to look at something between UDP and TCP-something that is a bit more reliable and that provides a bit more functionality. Our member companies are saying that they would like a UDP-like thing, but one that also has sequence numbers and that itself has acknowledgments. Where the Transport Area is concerned, this would be an important thing to have.

Also missing is an application protocol for embedded devices: http isn’t right; TFTP [Trivial File Transfer Protocol] is fine for some things; and SNMP [Simple Network Management Protocol] is close. So, trying to define what works at the application layer is going to be very important in the next phases of our work.

There are a few other things missing in the 6lowpan WG. For example, how do devices get commissioned in a bootstrapping phase and then get on the network without someone having to connect to it and set an SSID and give it a channel? We can’t do that for a trillion devices. The devices need to become self-actualized without human intervention. This is a tough problem. How will a device know what network to join when it comes online if it can join any of a number of networks. This is one of the areas that 6lowpan needs to look at going forward.

IETF Journal: What is the smallest device or object that can be addressed with an IP address?

Geoff: We’re finding that there’s almost no limit to how small a device or object can be. Some of the devices currently on the network include ones that have less than 32 kilobytes of flash and 2 kilobytes of RAM, and they operate on a couple of AA batteries for many years. In order to have a reasonable life span, battery-powered devices aren’t switched on continuously, but you can ping them and they respond. We still find that these extremely small devices appear to be always on, even though they’re sleeping 99.6 percent of the time. The devices have less computing power than the average digital watch, and yet they can still be on the Internet.

Therefore, we can think about putting outlets, smoke detectors, light switches, and lights on the Internet, so if the lights go out, we can ring your phone or we can send a text message. Could it be handled in other ways? Yes, but the Internet infrastructure is already there, so why build another?

IETF Journal: What are the privacy concerns with this kind of technology?

Geoff: People like to measure things that can be measured. We can count people, but for reasons of privacy, some people don’t want to be counted. I’m not so much worried about things being tracked; there are many occasions when tracking is a good idea. For example, I may want to know where my coffee cup is. But for many people or for certain things, tracking might not be such a good idea. The Internet of Things isn’t really changing this-tracking can be done today- but it is accelerating it.

Unfortunately, technologies have a tendency to outpace our ability to understand how to manage them. It’s important to ensure that the security protocols that the IETF has developed can be applied so that if it’s my device, I can control where the data goes and I can choose who gets the data and make sure that no one can intercept it.

IETF Journal: What is the relationship between this work and radio-frequency identification [RFID]?

Geoff: That’s a good question. RFID is a lot of technologies. There is active RFID and passive RFID. Active RFID tags could be part of the Internet of Things that IPSO and the IETF and 6lowpan are enabling. Rather than sending a blob of numbers, those bits could be reformatted to look like an IPv6 packet. Passive RFID tags are a little different in that they have an extremely limited amount of data that they transmit when energized.

I don’t think that with the current technology the data is large enough to look anything like a compressed IPv6 address. Passive RFID could be connected to a slightly smarter device that could relay the data back to the Internet.

IETF Journal: Is there anything else you would like us to know about the work being done by the IPSO Alliance?

Geoff: Mainly that IPSO is dedicated to supporting the work of the IETF and the Internet Society by promoting the use of IP and Internet technologies. I believe it is a symbiotic relationship. We cannot and would not exist without IP.

What we hope we can do is bring the marketing, visibility, and promotion of the phenomenal work that is happening at the IETF to a brand-new set of potential users, builders, engineers, devices, customers and applications. The IETF is of critical importance because IP, the central protocol within the IETF, is the first thing in our name.

Whatever we can do to support the goals of the IETF and the Internet Society, we want to do. We want to find ways to work together, to promote the Internet and IP.