Non Boaty .. computer clock

[lustyd]

Umm, why? The gps is doing the navigation not the pc. If the pc syncs to the nmea stream a second or two one way or the next makes no difference to anything! The clock on the pc will only drift a few microseconds in the time it takes to do any speed or other calcs.

Depending on the OS and hardware the clock could actually drift by minutes, hours or even days on a passage if the PC is left switched on without internet connectivity. You'd be surprised how flawed the time systems are on computers which are not internet connected but that are switched on 24x7.

 

[lustyd]

Yes but now I'm concerned about being the only kid in the marina without sub-millisecond accuracy, especially after just googling this.

I'm suspecting for work you just *buy* the network timeclocks

hehe if you want to properly geek out, look into the potential attacks on the stockmarket made possible by Canary Wharf (and other financial places) relying on GPS time. Each of the buildings has a GPS receiver in the roof which gives their servers "accurate" time. It's trivial to override the weak sat signals and change the time while you buy or sell shares

P.S. I have not done this - proof being my tiny boat!

 

[VicS]

New battery fitted.

Cannot say without hiccups but after 2 or 3 reboots it's up and running. The clock defaulted to some date and time in 2006 but is now set and synchronized with the internet time.

Fingers crossed it will still be OK tomorrow

ps old battery was down to 1.5 volts

 

[nigelmercier]

Each of the buildings has a GPS receiver in the roof which gives their servers "accurate" time...

They may well have this for backup, but GPS time is not stable enough for the banking system to rely on it. They will have access to a time server, with route correction applied.

 

[nimbusgb]

They may well have this for backup, but GPS time is not stable enough for the banking system to rely on it. They will have access to a time server, with route correction applied.

Other way round. Npt is waaaaay less accurate than gps by a factor of 2 orders of magnitude. NTP servers use gps as their time source. The gps's used have an output pulse whose trailing edge is accurate to less than 10 nanoseconds. Using bsd as the os you can get down to those sorts of accuracy, with Linux you can get down to around 1 microsecond.

NTP Will only get millisecond accuracy in a local network, referencing an external NTP server you could be a whole 10 ms out.

 

[laika]

They may well have this for backup, but GPS time is not stable enough for the banking system to rely on it.

Not only is GPS normally used as the main source of time information (for both internal NTP and PTP) in institutions I'm familiar with, it's often the *only* source. The wisdom of the latter point is debatable though probably less of a risk than I first thought: Not just in terms of likelihood of global GPS shutdown (The boaty interest bit!) but in terms of impact if the network stays sync-ed to itself and only starts to drifting from UTC very slowly. With many big banks outsourcing infrastructure and networks and cutting the costs and quality of the staff they hire, that risk analysis has (unfortunately) probably not been performed in many cases.

Not sure I agree with Dave's statement about fiddling a bank's time being "trivial" using GPS spoofing. Internal time for a global investment bank will be based on a number of stratum 1 clocks at locations around the world. You'd need to know where in which buildings they were in (Canary Wharf or the datacentre in Swindon/Slough/Bracknell? Manhatten or one of the datacentres in New Jersey?), preferably have proximal physical access to the antennae, and you'd have to globally co-ordinate the attack to drift time slowly: jumping the time is easy enough to detect and flag up as unreliable, though whether the majority of commercial time server appliances do this these days I don't know. Not being an expert in such things I can't comment on how much more complicated spoofing becomes without fairly close proximity.

So possible I believe, but far from trivial and I would think that with that degree of insider information and mathematical and technical ability there are easier ways to game a bank. Probably a lot of them are perfectly legal too.

Not only is this not boaty, it's not even related to the original post. sorry.

 

[lustyd]

They may well have this for backup, but GPS time is not stable enough for the banking system to rely on it. They will have access to a time server, with route correction applied.

Rofl. You don't understand GPS at all, do you?nit works by your device receiving the correct time from several satellites and from the offsets of each triangulating the position. The whole thing only works because it has accurate atomic clocks in each satellite. The new system being launched will be more accurate because it includes more accurate atomic clocks!

 

[lustyd]

Not only is GPS normally used as the main source of time information (for both internal NTP and PTP) in institutions I'm familiar with, it's often the *only* source. The wisdom of the latter point is debatable though probably less of a risk than I first thought: Not just in terms of likelihood of global GPS shutdown (The boaty interest bit!) but in terms of impact if the network stays sync-ed to itself and only starts to drifting from UTC very slowly. With many big banks outsourcing infrastructure and networks and cutting the costs and quality of the staff they hire, that risk analysis has (unfortunately) probably not been performed in many cases.

Not sure I agree with Dave's statement about fiddling a bank's time being "trivial" using GPS spoofing. Internal time for a global investment bank will be based on a number of stratum 1 clocks at locations around the world. You'd need to know where in which buildings they were in (Canary Wharf or the datacentre in Swindon/Slough/Bracknell? Manhatten or one of the datacentres in New Jersey?), preferably have proximal physical access to the antennae, and you'd have to globally co-ordinate the attack to drift time slowly: jumping the time is easy enough to detect and flag up as unreliable, though whether the majority of commercial time server appliances do this these days I don't know. Not being an expert in such things I can't comment on how much more complicated spoofing becomes without fairly close proximity.

So possible I believe, but far from trivial and I would think that with that degree of insider information and mathematical and technical ability there are easier ways to game a bank. Probably a lot of them are perfectly legal too.

Not only is this not boaty, it's not even related to the original post. sorry.

While you are correct that drifting by several minutes would be detected quickly, the stock market requires only seconds of advance notice to make lots of money. If you can see 5 seconds into the future that is sufficient. To target GPS you only have to broadcast on the right frequency at a pretty low power to override a satellite because by the time it hits the earth the satellite signal is very very low power. Knowing where the antenna is is easy - they are on the roof because they need line of sight to the sky you're right though, this is so far off topic we may as well discuss my little pony.

 

[laika]

you're right though, this is so far off topic we may as well discuss my little pony.

yes too offtopic. Though if you have a boat *and* a pony I'm wondering if your plot may be more advanced than I suspected...

 

[lustyd]

Sorry I meant My Little Pony but couldn't be arsed with the capitals on the iPad. I don't in fact have a pony and don't feel I need one.

For anyone "official" reading this, I do in fact not have a master plan for stealing anything. I'm just quite technical and have worked with quite a few financial institutions and so know their kryptonite to be honest if I wanted to do anything mean I'd use their wifi which was installed at the request of the CEO when he bought an iPad...much easier!

 

[nigelmercier]

Rofl. You don't understand GPS at all, do you?

I understand the system enough to know that it doesn't cater for leap-seconds, and therefore drifts relative to UTC.

 

[laika]

Actual Boaty relevance!!!!

I understand the system enough to know that it doesn't cater for leap-seconds, and therefore drifts relative to UTC.

UTC drifts from GPS time (and TAI) as leap seconds are added (as you correctly say), but the offset from GPS time (currently 16s) is transmitted with the GPS signal, so receivers can work out UTC and *hopefully* (if correctly programmed) everyone's plotters are still showing the correct time after leap seconds are added.

 

[BruceDanforth]

Whose internet time server?