Time after Time (part 1)

Time is critical to many operations. The statement rings particularly true when the processes occur over an environment composed of distributed, interconnected, and heterogeneous nodes.

Often overlooked is that the 'correct' time for all these nodes should be second to the essential aspect of process synchronization.

In its simplest form, the goal is: "everyone and everything must operate 'using the same' time." Obviously, a confident and precise timing reference is necessary to achieve this.

Setting aside - for the moment - the possibility of 'spoofing' and 'jamming' the existing Global Navigation Satellite System (GPS, Galileo, Beidu, Glonass, IRNSS, etc.) provides an acceptable first-tier level of time-of-day confidence around which to build a system capable of providing sub-microsecond-accurate, worldwide distributable Time-of-Day (TOD) information and synchronization.

We will achieve this by combining satellites, disciplined reference oscillators, and – for good measure – the most accurate network protocol capable of synchronization across separate domains, the Precision-Time-Protocol or PTP.

Later, we will enhance this system to operate accurately in a signal-deprived environment where relative accuracy is maintained during the eventual loss of any reference time source.

First things first, to implement this system, we first need to understand a bit about the GNSS networks, and for this writing, we shall focus on the US-owned constellation known as the Global Positioning System or GPS.

There is no need to go into the fine details just yet; suffice it to say that:

• Each GPS satellite broadcasts radio signals with their location, their status, and precise Universal Coordinated Time (or UTC) information derived from their onboard 'atomic clocks' (stable within three nanoseconds)

  • Regardless of their superb accuracy and stability, these onboard clocks receive two daily updates to correct for their natural drift via ground-based atomic clocks that are fifty times more stable than those on the spacecraft, with a natural drift of no more than one second every 100 million years.

• A typical GPS receiver processes these radio signals and uses them to calculate the distance from each satellite its antenna can 'see.'

• Once the GPS receiver knows the distance from at least four satellites, it uses a form of geometry (called trilateration) to determine its exact location on Earth.

Of course, environmental factors influence every receiver to some degree; therefore, the overall accuracy of the time sourced from any one receiver is universally accepted to be within 100 nanoseconds of the valid UTC' second-by-second rollover.

That's all for now. In upcoming installments, this blog will provide a step-by-step, open-source approach to a GPS-based time and frequency delivery system capable of +/- 20 nanoseconds accuracy, a 10MHz lab-grade sine-wave frequency reference, and the necessary network interfaces to distribute this data via the PTP, NTP, Sonet E1, and IRIG B protocols.

Below, there is a simplified block diagram of the final project. The architecture was selected to overcome the individual shortcomings of the individual components, mainly:

• A time-of-day based on GPS is "noisy" in the short term but very accurate in the long term.

• A compensated (stable) oscillator ensures continuous frequency reference in GPS-deprived environments.

• A real-time clock continuously updated by a valid GPS, using a disciplined oscillator as source, can provide long-term holdover TOD operation in the absence of GPS

Next: Time-after-Time Part 2… A spoofing/jamming hardened GPS design…