Every Second Counts

Time has always been important to humans. The concept of annual cycles in farming and other activity, the celebration of an event such as a New Year, the principle of a week, aligning with cycles of rest or alternative activity to punctuate the normal activity of farming, building, writing, or whatever. A way to measure progress.

And we need to divide the day up into smaller units, to measure how long we are spending on work, to agree a rendezvous with someone else, to coordinate our mealtimes, and so on. Technological progress and the increasing sophistication of society worldwide have required these divisions to become ever smaller. Earlier civilisations didn’t need accurately measured seconds. The concept of widely-used national common time only came with the railways in the 19^th century, and on a worldwide basis with transport and electronic communications.

You may be wondering about the connection between time standards and radiocommunications? We will not take space here to explain exactly why the issue of worldwide time standards and measurements is on our agenda, but we can explain what the issue is and why it is of more than academic importance.

Today’s use of time

The world now runs with an extremely accurate, constant and reproducible definition of a second of time; these days it is taken from a very precise atomic clock reference. And having defined it so precisely and needing it to remain as a physical constant we now want to keep it that way. Surely that is a good thing? Well, it is good for our technologically supported way of life. But thankfully the planet is part of nature and not so regimented. Its rotation on its own axis – the basis of our day and the sub-parts which we divide it into – is very gradually and erratically slowing down ¹.

Therefore, the measurement of time by using the scientifically based second, used for synchronising all manner of processes including communication networks, slowly gets out of step with the natural world of which we are a part.

Until now the solution has been to introduce a 'leap second', in other words to stop 'official/scientific' time (Co-ordinated Universal Time, 'UTC'), for one second every so often. The slowdown in earth rotation is not constant, so the leap second has been applied when required, on average about every four years. This maintains UTC’s link to solar time, where we have 3600 seconds per hour, and 24 hours in a day. This link is consistent with ITU-R Resolution 653 (agreed at the World Radiocommunication Conference in 2012 (WRC-12)), which identifies the need for a timescale linked to earth rotation. However, the Resolution also calls for ITU to study the feasibility and implications of establishing a continuous reference timescale, i.e. one without correction by leap seconds.

So what is the problem: why not just keep the practice of a 'leap second'?

The problem arises because so many digital systems have difficulty in coping with an interruption to a continuous time reference, for example, receiving two identical consecutive time stamps, or not being able to respond to the advance notification that is broadcast before a leap second is applied. This can impact, inter alia, internet timing, global satellite positioning systems (e.g. GPS, GLONASS), mobile phone networks, and high speed financial trading. The last leap second, in 2012, did cause problems.

This in turn is starting to lead to the adoption of limited private time standards used just for one system or group of systems, which in turn will increasingly lose the benefits of those systems being able to communicate with each other on the basis of a common time reference.

Figure 1: how leap seconds have tracked the increasing difference between the continuous count of the atomic clock (TAI), and the Coordinated Universal Time (UTC)

This issue has been debated for a long time in ITU-R, over 15 years now, without reaching any agreement and consensus. Although not a specific radio-spectrum issue, it is again on the agenda of WRC-15.

The ECC is currently studying various options to find a solution and we have provided contributions to ITU-R on the development of the Conference Preparatory Meeting (CPM) text and the methods to satisfy this agenda item.

One such option we have studied would remove the use of leap seconds, with UTC then becoming a continuous time scale, thus breaking the link between civil time (using UTC) and the Earth’s daily rotation (which is slowing down). Several concerns have been raised on this solution, such as the potential impact to one global navigation satellite system (GNSS), the impact to civil and legal time keeping, and also the uncertainty and impact on existing systems and software which are designed to operate on the UTC using leap seconds.

Therefore, the ECC is also giving further consideration to the option of disseminating (i.e. broadcasting) two time scales in parallel, which might solve the differences between the two opposing camps and which may accommodate the differing needs. Hence, one study concluded that it is worth keeping the insertion of leap seconds in UTC, as done presently and which avoids ambiguity for civil users, and then allowing experts to extract from such dual broadcast the continuous pseudo-reference time scales required for the specific more automated applications.

The next Conference Preparatory Group project team (CPG PT-A) meeting this January is planning to consider the option of retaining UTC as currently defined (i.e. which introduces leap seconds as and when required) and modifying the Radio Regulations to recognise, on an equal basis, a continuous reference atomic time-scale with an offset which can be derived from the UTC broadcast. This method could be acceptable as a compromise and possibly the only way forward in satisfying different users of the international time scale reference.

Tony Azzarelli, Chairman of the ECC's CPG - PT-A
Mark Thomas, Director of the ECO

¹ The Earth slows down due to its interactions with the moon; the slowing is erratic because of the redistribution of mass within the Earth, in the mantle layer.