Time Synchronisation

Time is one of the least understood aspects of our universe. We know it exists yet we have trouble grasping exactly what it is. Time can be viewed in two ways, it is a man made concept used as a tool to describe to explain the sequence of events, comparing the durations and intervals between them.

Time is one of the fundamental quantities which also includes distance, velocity, mass, momentum, energy, and weight and thanks to the work of Einstein and others we know time also makes up the very fabric of our Universe.

Here are ten facts you may or may not have known about time.

10. Time is not a constant; time is relative to different observers. The only constant in the Universe is the speed of light which means no matter how fast you are travelling the speed of light will remain the same although time will slow down.

9. Time can be described as a dimension and along with the other three dimensions we are aware of (up/down, left/right and forward/backward) forms a four dimensional ‘space-time’.

8. Time always moves forward yet many theoretical physicists believe that backwards time travel could be possible.

7. Gravity can warp space-time making time slow down the stronger the gravitational force. Experiments with atomic clocks show the higher above sea level they are (and therefore under less gravitational influence) the faster they run (although the difference is very small).

6. As the speed of light is the only constant in the Universe no matter how fast you travel, light will always seem to be the same speed, this is because time will slow down. A journey at close to the speed of light may seem like a few seconds for a traveler but to an observer it would have taken thousands of years.

5. Time has not always existed. Time started with the big bang and will end if the Universe does.

4. Time can be perceived differently by our brains depending on our activities. A boring day will ‘drag’ on whilst if we are enjoying ourselves time will seem to ‘fly’, this phenomenon is referred to as ‘temporal illusion’ by psychologists.

3. Time appears to accelerate the older we get. Some (including Stephen Hawking) suggest the reason for this is that when we are ten years old a year is a tenth of our whole life and seems a long time, yet for a sixty-year-old a year is just a 60th of their life and therefore perceived as a shorter period.

2. Some modern atomic clocks are so accurate they can lose less than a second in 400 million years.

1. A universal time scale has been developed called UTC (Coordinated Universal Time) which is based on the time told by atomic clocks but compensates for the minute slowing of the Earth’s rotation (caused by the gravity of the Moon) by adding Leap Seconds every year to prevent day from creeping into night (albeit in a millennia or two).

Thanks to atomic clocks and UTC time computer networks all over the world can receive a UTC time source over the Internet, via a national radio transmission or through the GPS network. A NTP server (Network Time Protocol) can synchronise all devices on a network to that time.

The worst part of a power cut is running around the house setting all the clocks and timers back to the correct time, it can take ages and you will always forget one, however, as long as you have a wrist watch it should be quite easy to get your clocks all telling the same time. But what time is your wrist watch set too and who regulates that time?

Complete precision and accuracy in time telling is not essential for our day-to-day lives and neither is synchronisation, our computer may be a few minutes slower than our wall clock but it will make little difference when we send an email.

However, what if the person we sent the email to has a computer clock that is even slower? They may end up sending a reply before they have technically received it. Computers are easily fooled if timestamps run backwards – remember the millennium bug!

For this reason it is important for computers, particularly those that deal with time sensitive or financial applications, to be telling the same time; otherwise global stocks could be bought whilst already sold-out or an airline seat, already purchased could be bought again by a buyer with a slower computer clock.

The regulation of time didn’t start until after the development of atomic clocks when the oscillation of the caesium atom became the standard definition of a second (9,192,631,770 a second).

The time told by these atomic clocks was so accurate a new timescale was developed called International Atomic Time (TAI). However, it was discovered that the traditional method of telling time, based on the revolution of the Earth (ie 24 hours in a day) and this new timescale soon became out of sync with each other as the gravity from the moon alters the revolution of the Earth, slowing it down.

This difference in the Earths spin is only minute but enough people argued (mainly astronomers) that if it was not compensated for, night would eventually creep into day (albeit in many thousands of years) and it would be difficult to keep track of the celestial bodies.

A compromise was called for and the new timescale, Universal Coordinated Time (UTC) was developed that accounted for the slowing of the Earth’s spin by adding leap seconds every year or so.

UTC has meant that modern technologies and applications such as the Global Positioning System, satellite communication, live television broadcasts and global trading have become possible.

Computer networks can receive UTC time and keep all their devices synchronised to it by using a NTP server (Network Time Protocol). NTP servers can receive UTC time from an atomic clock source via the Internet, a national radio transmission or through the GPS network.

This article discusses the development of atomic clocks, why accuracy is so important, how they developed and the next generation of atomic clocks that offer increased accuracy.

Atomic clocks have been with us for over fifty years now and most people have heard of them and know they are very accurate, but how accurate are they and why do we need such accurate clocks?

Atomic clocks are used by many of us even if we are not aware of it. The time they tell is relayed around the world and picked up by time servers using the protocol NTP to synchronise networks, they are vital for lots of technologies, such as global satellite navigation, and TV signal timings.

Before the development of the atomic clock the most precise timekeeping devices were electronic clocks which would lose a second or two every week. These had largely replaced mechanical clocks which were less accurate still.

Mankind has always had a fascination for keeping track of the time but knowing the precise time has never been too important. A second or even a minute’s difference does not affect our day-to-day lives.

However, as technology has advanced the need for more precise timekeeping has increased. Satellites that have to be navigated and communicate with the Earth from hundred, thousands and even millions of miles away require exact timing. Light and therefore radio waves can travel 300,000 km every second so slight inaccuracies in time can have massive differences.

The first accurate atomic clock was built y Britain’s National Physical Laboratory in 1955 by Dr Louis Essen who based his clock around the oscillation of the caesium -133 atom. The idea was actually first conceived as far back as 1879 when Lord Kelvin proposed that time-keeping based on how atoms behaved would be a better way to count time intervals than anything else.

The first generation of atomic clocks (also known as caesium oscillators) used the frequency of this atom which oscillates 9,192,631,770 times every second. Essen’s model was accurate to a second every 300 years but developments of the caesium oscillator mean they can now achieve accuracies of one second every 80 million years.

Yet as technologies get more advanced, scientists strive to make better and more accurate clocks. Rubidium standard clocks offer no better accuracy than caesium models but are smaller and cost less (caesium oscillators are generally only to be found in large-scale physics laboratories).

Clocks using just a single atom have been developed that offer even more accuracy. A clock based on a single mercury atom has achieved accuracies of one second in 400 million years and it is expected that a new type of strontium clock that uses light will go even better.

The future for atomic clocks is ever increasing accuracy combined with scaling down the size and cost of them. The American National Institute of Standards and Technology (NIST) have unveiled a chip-sized atomic clock that boasts millisecond accuracy.

Atomic clocks are now part and parcel of our lives without the time signals they transmit to the world that are picked up by NTP servers modern communication from Internet shopping and GPS and technological advances such as satellite navigation would become impossible.

Summary: This article gives a step-by-step guide in configuring LINUX to act as an authoritative time server using NTP (Network Time Protocol).

Computer time synchronisation is highly important in modern computer networks, precision and time synchronization is critical in many applications, particularly time sensitive transactions. Just imagine buying an airline seat only to be told at the airport that the ticket was sold twice because it was purchased afterwards on a computer that had a slower clock!

Modern computers do have internal clocks called Real Time Clock chips (RTC) that provide time and date information. These chips are battery backed so that even during power outages, they can maintain time but personal computers are not designed to be perfect clocks. Their design has been optimized for mass production and low-cost rather than maintaining accurate time.

For many applications, this is can be quite adequate, although, quite often machines need time to be synchronised with other PC’s on a network and when computers are out of sync with each other problems can arise such as sharing network files or in some environments even fraud!

Network Time Protocol (NTP) is an Internet protocol used for the transfer of accurate time, providing time information along so that a precise time can be obtained. As NTP was originally written for LINUX many LINUX based operating systems already have a version of NTP installed. However the source code is free to download from the NTP website (NTP.org) the most recent version being v 4.2.4.

NTP (version 4) can maintain time over the public Internet to within 10 milliseconds (1/100th of a second) and can perform even better over LANs with accuracies of 200 microseconds (1/5000th of a second) under ideal conditions.

NTP works within the TCP/IP suite and relies on UDP, a less complex form of NTP exists called Simple Network Time Protocol (SNTP) that does not require the storing of information about previous communications, needed by NTP. It is used in some devices and applications where high accuracy timing is not as important.

The NTP background program is configured with the file ‘NTP.conf’. this may contain a list of public NTP server references that can be used to synchronise time. NTP time servers are specified using the ’server’ command, any characters after the ‘#’ symbol are comments:

Example
server time-a.nist.gov # Public NTP server: Maryland
When configured, NTP can be controlled using the commands ‘ntpd start’ ‘ntpd stop’ ‘ ntpq –p’ (displays status)

NTP can also authenticate timing resources Note: It is strongly recommends that you configure a time server with a hardware source rather than from the internet where there is no authentication. Authentication codes are specified in the ‘NTP.keys’ file.

Specialist NTP servers are available that can receive transmissions from either GPS or national time reference broadcasts. They are relatively cheap and the signal is authenticated providing a secure time reference.

Authentication for NTP has been developed to prevent malicious tampering with system synchronisation just as firewalls have been developed to protect networks from attack but as with any system of security it only works if it is utilised.

We have all heard of a leap year – that extra day added to the calendar every four years. It may give us a longer February but it is also essential in keeping our calendars and seasons accurate. If the extra day is not added to a leap year then eventually (admittedly after over a century) the Winter will begin in July and the summer will start around Christmas (and vice – versa in the southern hemisphere) because the Earth takes an extra six hours longer than the 365 days of a year to circle the sun.

A leap year may be a bit of a fudge but the alternative would be to have a quarter day at the end of the year which would of course throw our days and nights out of sync with each other (and could you imagine just having a six hour day – some of us struggle to get things done in 24!).

We have of course always measured time in relation to the movement of the Earth – a day being an entire revolution, a year an orbit of the sun. However, as our way of measuring time became more and more accurate it soon became apparent that there were more irregularities in the Earth’s rotation than just the extra six hours in a year.

GMT (Greenwich Mean Time) was developed because there was a need for a time scale where the mean position of the sun at noon, averaged throughout the year, is above the Greenwich Meridian (zero longitude) and daylight saving hours are added or taken away depending on the time of year.

However, in 1955 the first atomic clock went into operation following the discovery of the stability of the caesium-133 atom which vibrated at an exact rate (9,192,631,770 a second). Impressed with this accuracy, The International System of Units of Measurement (SI) decided that a second should be defined as this number of oscillations of the caesium-133 atom.

Following the SI second a time scale called International Atomic Time (TAI – from the French Temp Atomique International) which was a simple count, in seconds, for the 24 hours of our day. Conversely as TAI is not related to the movement of the Earth, it was soon discovered that TAI and atomic clocks were far more stable and reliable than the Earth itself (in fact an atomic clock is 1,000,000 times more accurate than the Earths rotation).

Generally the Earth is continually slowing in its rotation (although, inexplicably, every now-and-then it seems to speed up) so TAI is of little use for those that wish their clocks to be in step with the Earth (astronomers being by far the most vocal of these).

So another time scale was developed called Coordinated Universal Time (UTC – again from the French – Temp Universel Coordonne). This was based on atomic time (TAI) but small adjustment are made to keep it in step with GMT (which incidentally is now commonly referred to as UT1 or depending on time zone UT+1 UT+2 UT+3 etc)

UTC is adjusted by the insertion of extra seconds, called leap seconds, as necessary to keep it within a second of GMT (or UT1). It is possible a second may have to be removed in the future but that hasn’t happened as yet. UTC is essential in modern industry and technology where computers are synchronised to UTC time, usually through a NTP server (Network Time Protocol) – to allow international time sensitive transactions.

A leap second is normally inserted at the end of December in the last hour (although occasionally it has been done in June, March and September). The decision as to whether a leap second is required is taken by the Earth Orientation Centre of the International Earth Rotation and Reference Systems Service (IERS), who monitor the Earth’s rotation and suggest the adjustment about six months in advance.

When a leap second is added there becomes 61 seconds in that final minute of the year. The familiar ‘six pips’ radio signal gains an extra pip and even London’s famous Big Ben is held back a second before it bongs (but not an extra bong as they are meant to represent the hours)

There have been 33 leap seconds added to UTC since 1972 (although the first ten were added retrospectively) but as the Earth’s rotation is continuing to slow it is estimated that over the next millennia or two leap seconds will have to be added each month.

All PC’s and networking devices use clocks to maintain an internal system time. These clocks, called Real Time Clock chips (RTC) provide time and date information. The chips are battery backed so that even during power outages, they can maintain time. However, personal computers are not designed to be perfect clocks, their design has been optimized for mass production and low-cost rather than maintaining accurate time.

These internal clocks are prone to drift and although for many application this can be quite adequate, often machines need to work together on a network and if the computers drift at different rates the computers will become out of sync with each other and problems can arise particularly with time sensitive transactions.

Network Time Protocol (NTP) is one of the Internet’s oldest protocols still used, invented by Dr David Mills from the University of Delaware, it has been in utilized since 1985. NTP is a protocol designed to synchronize the clocks on computers and networks across the Internet or Local Area Networks (LANs).

NTP (version 4) can maintain time over the public Internet to within 10 milliseconds (1/100th of a second) and can perform even better over LANs with accuracies of 200 microseconds (1/5000th of a second) under ideal conditions.

NTP works within the TCP/IP suite and relies on UDP, a less complex form of NTP exists called Simple Network Time Protocol (SNTP) that does not require the storing of information about previous communications, needed by NTP. It is used in some devices and applications where high accuracy timing is not as important.

Many operating systems including Windows, UNIX and LINUX can utilize NTP and SNTP  and time synchronisation with NTP is relatively simple, it synchronises time with reference to a reliable clock source. This source could be relative (a computer’s internal clock or the time on a wrist-watch) or absolute (A UTC – Universal Coordinated Time – clock source that is accurate as is humanely possible).
All Microsoft Windows versions since 2000 include the Windows Time Service (w32time.exe) which has the ability to synchronise the computer clock to an NTP server.

There are a large number of Internet hosted NTP servers that synchronise with external UTC references such as time.nist.gov or NTP.my-inbox.co.uk but it must be noted that Microsoft and others recommend that an external source is used to synchronise your machines, as Internet based references can’t be authenticated. Specialist NTP time servers are available that can synchronise time on networks using either the MSF (or equivalent) or GPS signal.

The most widely used are the GPS time servers which use the GPS system to relay accurate time. The GPS system consists of a number of satellites providing accurate positioning and location information. Each GPS satellite can only do this by utilising an atomic clock which in turn can be can be used as a timing reference.

A typical GPS receiver can provide timing information to within a few nanoseconds of UTC as long as there is an antenna situated with a good view of the sky.

There are a number of national time and frequency radio transmissions that can be used to synchronise a NTP server. In Britain the signal (called MSF) is broadcast by the National Physics Laboratory in Cumbria which serves as the United Kingdom’s national time reference, there are also similar systems in Colorado, US (WWVB) and in Frankfurt, Germany (DCF-77). These signals provides UTC time to an accuracy of 100 microseconds, however, the radio signal has a finite range and is vulnerable to interference.

The Global Positioning System (GPS) is now a familiar tool in helping motorists to navigate but GPS has more uses than merely triangulating a position for direction finding, it can be utilized to provide time and frequency information worldwide.

Developed by the United States military, GPS incorporates at least 24 communication satellites in high orbit, all of which contain precise timing equipment to enable the satellite to triangulate positions with accuracy.

However, each satellite’s highly accurate atomic clock timing reference can also be used by NTP (Network Time Protocol) servers to synchronise computer networks using the highly accurate GPS time signal as an external reference.

GPS is an ideal time and frequency source because it can provide highly accurate time anywhere in the world using relatively cheap components. Each GPS satellite transmits in two frequencies L2 for the military use and L1 for use by civilians transmitted at 1575 MHz, Low-cost GPS antennas and receivers are now widely available.

The radio signal transmitted by the satellite can pass through windows but can be blocked by buildings so the ideal location for a GPS antenna is on a rooftop with a good view of the sky. The more satellites it can receive from the better the signal. However, roof-mounted antennas can be prone to lighting strikes or other voltage surges so a suppressor is recommend; installed inline on the GPS cable.

The cable between the GPS antenna and receiver is also critical. The maximum distance that a cable can run is normally only 20-30 metres but a high quality coax cable combined with a GPS amplifier placed in-line to boost the gain of the antenna can allow in excess of 100 metre cable runs.

A GPS receiver then decodes the signal sent from the antenna to a computer readable protocol which can be utilised by most time servers and operating systems including, Windows, LINUX and UNIX.

The GPS receiver also outputs a precise pulse every second that GPS NTP servers and computer time servers may utilise to provide ultra-precise timing. The pulse-per-second timing on most receivers is accurate to within 0.001 of a second of UTC (Coordinated Universal Time)

GPS is ideal in providing NTP time servers or stand-alone computers with a highly accurate external reference for synchronisation.

Even with relatively low cost equipment, accuracy of a hundred nanoseconds (a nanosecond = a billionth of a second) can be reasonably achieved using GPS as an external reference.

Network Time Protocol (NTP) is one of the Internet’s oldest protocols still used, invented by Dr David Mills from the University of Delaware, it has been in utilized since 1985. NTP is a protocol designed to synchronize the clocks on computers and networks across the Internet or Local Area Networks (LANs).

NTP (version 4) can maintain time over the public Internet to within 10 milliseconds (1/100th of a second) and can perform even better over LANs with accuracies of 200 microseconds (1/5000th of a second) under ideal conditions.

NTP works within the TCP/IP suite and relies on UDP, a less complex form of NTP exists called Simple Network Time Protocol (SNTP) that does not require the storing of information about previous communications, needed by NTP. It is used in some devices and applications where high accuracy timing is not as important.

Time synchronisation with NTP is relatively simple, it synchronises time with reference to a reliable clock source. This source could be relative (a computer’s internal clock or the time on a wrist-watch) or absolute (A UTC – Universal Coordinated Time – clock source that is accurate as is humanely possible).

It is strongly recommended by Microsoft and others, that external based timing should be used rather than Internet based, as these can’t be authenticated. Specialist NTP servers are available that can synchronise time on networks using either the MSF (or equivalent) or GPS signal.

Atomic clocks are the most absolute time-keeping devices; however, they are extremely expensive and are generally only to be found in large-scale physics laboratories. However, NTP can synchronise networks to an atomic clock by using either the Global Positioning system (GPS) network or specialist radio transmission (MSF in Britain).

The MSF national time and frequency radio transmissions used to synchronise an NTP server is broadcast by the National Physics Laboratory in Cumbria which serves as the United Kingdom’s national time reference, there are also similar systems in Colorado, US (WWVB) and in Frankfurt, Germany (DCF-77).

A radio based NTP server usually consists of a rack-mountable time server, and an antenna, consisting of a ferrite bar inside a plastic enclosure, which receives the radio time and frequency broadcast. The antenna should always be mounted horizontally at a right angle toward the transmission for optimum signal strength. Data is sent in pulses, 60 a second. These signals provides UTC time to an accuracy of 100 microseconds, however, the radio signal has a finite range and is vulnerable to interference.

A radio referenced NTP server is easily installed and can provide an organization with a precise time reference enabling the synchronization of entire networks.

Network Time Protocol (NTP) is one of the Internet’s oldest protocols still in use. Invented by Dr David Mills from the University of Delaware it has been utilized since 1985. NTP is designed to synchronize the clocks on computers and networks across the Internet or Local Area Networks (LANs).

NTP (currently version 4)  is actually three things in one; a software program that runs in the background of Windows or UNIX; a protocol that exchanges time values between servers and clients; and a suite of algorithms that process the time values to advance or retreat the system clock.

NTP uses an algorithm (Marzullo’s algorithm) to synchronise time on a network using a time reference. Although networks can be synchronized with internal clocks or Internet based timing references, it is highly recommended by Microsoft and others that an external timing reference should be used to guarantee authentication. An absolute timing reference should use UTC (Coordinated Universal Time or Temps Universel Coordonné) which supports such features as leap seconds – added to compensate for the slowing of the Earth’s rotation.

NTP works within the TCP/IP suite and relies on UDP, a less complex form of NTP exists called Simple Network Time Protocol (SNTP) that does not require the storing of information about previous communications, needed by NTP. It is used in some devices and applications where high accuracy timing is not as important, it is also included in most Windows operating systems but more recent versions have the full NTP already installed, which is also free to download via the Internet.

Synchronisation with NTP is relatively simple, it synchronises time with reference to a reliable clock source such as an atomic clock, although these are extremely expensive and are generally only to be found in large-scale physics laboratories, however NTP can use either the Global Positioning system (GPS) network or specialist radio transmission to receive UTC time from these clocks.

NTP uses timestamps to represent the current time of the day each timestamp is ephemeral, in other words it is always greater than the previous timestamp as time never runs backwards. NTP analyses the timestamp values including the frequency of errors and the stability. A NTP server will maintain an estimate of the quality of its reference clocks and of itself.

The distance from the reference clock is known as the stratum levels and they exist to prevent cycles in the NTP. Stratum 0 are devices such as reference clocks connected directly to a computer. Stratum 1 are computers attached to stratum 0 devices, while Stratum 2 are computers that send NTP requests to Stratum 1 servers. NTP can support up to 256 strata.

NTP timestamps are in two formats but they relay the seconds from a set point in time (known as the prime epoch, set at 00:00 1 January 1900) The NTP algorithm then uses this timestamp to determine the amount to advance or retreat the system or network clock.

The NTP program (known as a daemon on UNIX and a service on Windows) runs in the system background. NTP refuses to believe the time it is told until several packet exchanges have taken place, each passing a set of tests. Only if the replies from a server satisfy the test, known as protocol specifications, the server is considered. It usually takes about five minutes (five good samples) until a NTP server is accepted as a synchronization source.

A typical GPS time server can provide timing information to within a few nanoseconds of UTC as long as there is an antenna situated with a good view of the sky.

There are also a number of national time and frequency radio transmissions that can be used to synchronise a NTP server. In Britain the signal (called MSF) is broadcast by the National Physics Laboratory in Cumbria which serves as the United Kingdom’s national time reference, there are also similar systems in Colorado, US (WWVB) and in Frankfurt, Germany (DCF-77). These signals provides UTC time to an accuracy of 100 microseconds, however, the radio signal has a finite range and is vulnerable to interference.

Time servers are like other computer servers in the sense they are usually located on a network. A time server gathers timing information, usually from an external hardware source and then synchronises the network to that time.

Often time servers are synchronised to a UTC (Coordinated Universal time) source which is the global standard time scale and allows computers all over the world to synchronised to exactly the same time. This has obvious importance in industries where exact timing is crucial such as the stock exchange or airline industry.

There are various sources that a time server can use as a timing reference. The Internet is an obvious source, however, internet timing references from the Internet such as nist.gov and windows.time can not be authenticated, leaving the time server and therefore the network vulnerable to security threats.

There are authenticated alternatives to the Internet, the most common being to use the GPS network. As the Global Positioning System is reliant on knowing exactly what time it is to ensure reliable location information, this information can be utilised by a time server.

A simple GPS antenna connected to the time server will allow the GPS timing reference to be regularly checked by the time server. A GPS time server will be accurate to within a few hundred nanoseconds (a nanosecond = a billionth of a second).

There are also a number of national radio broadcasts such as the WWVB signal from Colorado in the US, the MSF signal from Cumbria in the UK and the DCF-77 signal from Frankfurt in Germany.

These radio signals are limited in their range though and even in major cities such as London it can be difficult to receive a decent enough signal.

Most timing servers use NTP (Network Time protocol) there are other protocols available but NTP is predominately used and is thought of as the standard for timing protocols. NTP has been around for over 25 years and is currently on version 4 but is always being updated which is probably why it is by fat the most common timing protocol.

NTP time servers work within the TCP/IP suite and rely on UDP (User Datagram Protocol). A less complex form of NTP – Simple Network Time Protocol (SNTP) is used in some devices and applications where high accuracy timing is not as important and is also included as standard in Windows software (although more recent versions of Microsoft Windows have the full NTP installed and the source code is free and readily available on the Internet from NTP.org).