Expert Interview: Marc Weiss, PhD

April 3 - 6, 2017
San Jose, CA
DoubleTree San Jose

The ATIS Interview

IoT and Timing as a Service

Marc Weiss, PhD, Time & Frequency Expert Consultant, NIST


Marcella Wolfe, ATIS: With more machines talking to machines, sensors communicating with utility substations, manufacturing processes using distributed control, and robotic surgeries being conducted, timing has a central and increasingly critical role. Tell us more about it.

Dr. Marc Weiss: The change on the horizon is stunning and unprecedented in history. We can view the Internet of Things (IoT) or the "Industrial Internet," as some of it is also known, as a revolution that brings together the Industrial Revolution and the Internet Revolution. Companies such as GE and Cisco predict that the magnitude of change coming is massive. Distributed industrial networks are predicted over large areas. These will need precision timing.

People talk about getting their household appliances online. This will not require precision timing, but rather time to a second or a minute at best. For things like the power industry, however, microsecond timing or better will be needed. Perhaps significantly better.

The medical industry is a key example of where precision timing is critical. Robotic surgery options exist now. These typically take place in an office with a surgeon controlling a machine and a patient in the same room or next door. The network used between the two is currently the same local area network. There are no network elements between the controller and the actuator, the term that refers to the cyber-physical systems used when a computer is doing something in the physical world.

If robotic surgery could be performed over the network over a distance of hundreds of miles the Industrial Internet will have created an amazing revolution. Currently nothing restricts this, either technically or physically. This potential means you could have one surgeon operating over a large rural area reaching people in underserved areas in the United States or ,with perhaps more impact, in a developing country as in parts of Africa or Asia, for example. You would only need nurses to prepare patients and stay with them in the clinic during the surgery.

When a surgeon makes an incision it is critical that there is minimal and predictable latency. Currently provides web pages that hesitate when they load because of large or unpredictable latency, or the delay through the network. In some cases, webpages or apps load and then stop. That's not good enough for robotic surgery. You want latencies that are stable and consistent. You need guarantees, and such low-latency is technically possible but the telecom industry will need to support it. For this to happen there needs to be investment in new equipment. Somebody will have to pay the telecom industry to create these options. The capabilities that are possible, however, are beyond imagination.

ATIS: How can timing as a service be cost-effective? How can telecom operators sell timing as a service?

Dr. Weiss: That's a critical question. If the telecom companies are going to provide precision timing as a service, they need to recover their costs as well as make a profit on their services. This will require significant operator investment.

Currently, with the delays in getting time through network elements now, it is as if I wrote you a letter and said "It's now two o'clock. Set your clock." And I dropped it in a mailbox and you got it day or two later. That's not particularly useful to you, unless you know exactly how long it took my message to get to you. That's exactly the parallel for sending time through the network—or sending time anywhere.

GPS only works because we send time from satellites, we know the satellite's positions and time, which gives your position and what the delay is. There's a speed of light distance. Years ago nobody in the telecom industry considered having minimal delay through the network. The network was designed for maximizing bandwidth, so a lot of data could go through quickly, but not with low latency.

In the finance industry high-speed traders want to be as close as possible to the actual markets. This means they're looking to get as low latency as possible, and pushing for special dispensations from the telecom operators. In some cases, they're building their own networks. For the larger public, secure timing through telecom networks is going to require that operators invest significantly in putting in equipment that tracks timing packets as they go through the network.

A set of standards to make this happen already exists. It's called Precision Time Protocol (PTP). It's being used widely in China, where they have made a big effort to measure everything and put in PTP-aware nodes everywhere in their network. This is also taking place in Europe. In the U.S., it is used less frequently because there is a cost involved, and the U.S. networks trust GPS.

The Industrial Internet has the potential of generating tens of trillions of dollars. For it to work, the telecom network needs to invest a tiny fraction of that amount. I think they will be willing to do that. It's a matter of instilling awareness in both camps. The industries involved and the telecom operators need to understand that there's an opportunity here.

ATIS: Why is the Workshop on Synchronization and Timing Systems important? We know it's the first to cover the issue from such a broad perspective and bring in the industry verticals. Why are you excited to be part of the event? Why should attendees be?

Dr. Weiss: I've been with WSTS since its inception. This is its 26th year. At the same time, it's a brand new event because of the potential of the Industrial Internet. Innovative new applications are being developed. Many industries are out there trying to do new things that are more timing-dependent than ever. That's why we're expanding—beyond the telecom community—to bring in experts from the many areas where timing is increasingly having an impact.

Historically, timing has been invisible. It's like the physical highway you're driving on. If it's perfect, has no potholes, and is well built, you drive and don't think about the pavement. Your focus is on your destination and what you'll accomplish there. You only notice the highway when you start hitting potholes.

We want to get the timing problem fixed before the multiple new industries that depend on timing technologies start hitting potholes. WSTS is where people can get both the broad picture and some of the details on what's going to happen, what's needed, why is it needed and how is it needed before problems occur.

If you're an equipment vendor, if you're a user building applications involved with the IoT or the telecom network, you want to know what the timing implications are before you start building. You don't want to discover things don't work and have to back up and redesign everything. Historically, that is the way it's been. With SONET, timing wasn't considered. The equipment was designed and built and timing considered later. This adds significant costs and limits timing accuracies. There's an important need for both the industry and the telecom network to be aware of the importance of timing, and how it can facilitate or block developers if it's not considered properly. At this event, we're trying to solve problems before they happen. That's why I'm excited about WSTS—and why, if you're reading this, you should attend.

ATIS: Thank you Marc. We look forward to seeing you at The Revolution in the Need for Time: WSTS 2017, April 3 - 6 in San Jose, CA.