Class 345 progress

[SteveOfTheStow]

The whole point of open-source is to have exponentially more eyes and brains looking for security holes before any bad actors find them. That benefit far, far outweighs the drawback (and an attacker can find flaws without source anyway, while a researcher needs source to fix holes).

What you're describing is "security through obscurity" which is a recognised fallacy — it just means that any exploit which is discovered enjoys the cover of secrecy.

OpenSSL is an open source software library for securing communications over networks, and is used by many applications to facilitate encrypted networking.

It's had a few vulnerabilities over the years, but a particularly serious one appeared in 2014. It was realised that because the project was open source, all its consumers thought the project had many eyes, but it turns out that everyone was just relying on everyone else to find the vulnerabilities. No one was actually looking. Why would you commit valuable time to that, when someone else should do it for you for free? (unless they don't)

Open source is no panacea. It has its place, but not for systems that can't afford to fail.

 

[ijmad]

The whole point of open-source is to have exponentially more eyes and brains looking for security holes before any bad actors find them. That benefit far, far outweighs the drawback (and an attacker can find flaws without source anyway, while a researcher needs source to fix holes).

What you're describing is "security through obscurity" which is a recognised fallacy — it just means that any exploit which is discovered enjoys the cover of secrecy.

I agree that security through obscurity is a bad idea as your primary defence, but it works as a last bastion - particularly for software I can't download and run, or access over the internet, to probe with vulnerability finders. Additionally, there is very limited benefit in open sourcing train control software because no-one in public has the equipment to run it and thus the open source community would have very little interest and engagement and the rigours of testing for safety critical systems makes it unlikely patches from interested developers could ever be accepted.

I'd say that for most software you'd run on your computer, or runs on Internet-connected servers, yes, the scales are tilted in the favour of open source for maximum security. However in this case, I believe the scale is tilted the other way. You don't see a lot of nuclear plant control software out in the open, after all.

I speak here as someone who made these judgements regularly in the civil service for some years in software teams, which now operate 'open by default' but closed when it makes sense.

 

[matt_world2004]

I think obscurity through obsfucation benefits train system software because any hacker who attempts to hack it is only going to get a rather limited opportunity to look at the software. A malicious actorplugging in a device to a software port on a train is only going to have a limited time to probe around before getting caught. Probably too little time to identify a flaw.

Releasing the code may show them where to look and write malicious code.

 

[ijmad]

I think obscurity through obsfucation benefits train system software because any hacker who attempts to hack it is only going to get a rather limited opportunity to look at the software. A malicious actorplugging in a device to a software port on a train is only going to have a limited time to probe around before getting caught. Probably too little time to identify a flaw.

Releasing the code may show them where to look and write malicious code.

Exactly. Without access via the Internet the surface for opportunistic attacks is so small that obscurity can work on balance. This is not true of OpenSSL, a piece of software installed on 90% of every computer and smartphone in the world. You could spend hours trying to dissect that, whether the source is open or closed.

That is rather worrying !. Best to avoid all these remote connection features all together is my feeling. That is despite the advantages.

Interesting aside - while he was Vice President, Dick Cheney had the wireless control features disabled in his pacemaker, due to similar concerns

 

[rebmcr]

I think obscurity through obsfucation benefits train system software because any hacker who attempts to hack it is only going to get a rather limited opportunity to look at the software. A malicious actorplugging in a device to a software port on a train is only going to have a limited time to probe around before getting caught. Probably too little time to identify a flaw.

Releasing the code may show them where to look and write malicious code.

Without access via the Internet the surface for opportunistic attacks is so small that obscurity can work on balance.

I doubt the ERTMS > ETCS > GSM-R chain is easily classified as standalone — don't a fair amount of GSM-R masts use BT's public 21cn as backhaul? It's a very public attack surface, with a huge amout of easily-predictable messages to gain understanding from (e.g. at Airport Junction, one could watch for trains passing, analyse their GSM-R transmissions, and start to piece together what's happening internally without any source code).

OpenSSL is an open source software library for securing communications over networks, and is used by many applications to facilitate encrypted networking.

It's had a few vulnerabilities over the years, but a particularly serious one appeared in 2014. It was realised that because the project was open source, all its consumers thought the project had many eyes, but it turns out that everyone was just relying on everyone else to find the vulnerabilities. No one was actually looking. Why would you commit valuable time to that, when someone else should do it for you for free? (unless they don't)

Open source is no panacea. It has its place, but not for systems that can't afford to fail.

I would say that's not specifically a failing of open-source, but of due diligence in general — i.e. not so different than similarly failing to validate a closed-source solution. It is, however, absolutely a good example of the dangers of complacency, which the "no security-through-obscurity" paradigm is all about preventing.

the rigours of testing for safety critical systems makes it unlikely patches from interested developers could ever be accepted.

Definitely, but independent security auditing is a great advantage that needs no patch acceptance.

 

[kevin_roche]

I think obscurity through obsfucation benefits train system software because any hacker who attempts to hack it is only going to get a rather limited opportunity to look at the software. A malicious actorplugging in a device to a software port on a train is only going to have a limited time to probe around before getting caught. Probably too little time to identify a flaw.

Releasing the code may show them where to look and write malicious code.

Unfortunately, while it might work against some guy in his bedroom it will eventually fail when disgruntled trusted employees or state actors decide to attack.

 

[dm1]

Unfortunately, while it might work against some guy in his bedroom it will eventually fail when disgruntled trusted employees or state actors decide to attack.

But if disgruntled trusted employees are the problem - then it doesn't matter whether the source code is open or closed, the employee will have access anyway.

 

[JonathanH]

Back to the trains, I note that a unit (likely to have been 345043) had to return from Heathrow to Old Oak Depot as a class 8 move yesterday evening.

It appears that 9T03 0502 London Paddington to Heathrow Airport Terminal 5 ran without delay from Paddington to Heathrow but on its return, 9P05 0552 Heathrow Airport Terminal 5 to London Paddington didn't run. The unit doesn't appear to have been able to move from Terminal 5 until 0725, causing the next three services all to be turned short at Heathrow Central (and run to Terminal 4 empty to reverse). It went from Terminal 5 to Terminal 4 at a slow pace, arriving at 0755. Then, having resided in platform 1 all day, finally running as 8Z05 1933 Heathrow Airport Terminal 4 to Old Oak Depot, taking around an hour to cover the fifteen or so miles back to the depot.

Seems to have been fairly good today until 9P65 1322 Heathrow Airport Terminal 5 to London Paddington was just terminated at Hayes & Harlington. The unit appears to have been put in Hayes & Harlington loop to resume on 9P77 1452 Heathrow Airport Terminal 5 to London Paddington.

 

[rd749249]

There's been a bit of disruption last couple of days. It made my down road journey through Hayes a bit interesting. Always nice to get a junction indicator 4 when you're not expecting one and go down the up relief to avoid a stranded unit. Another small move ticked-off the list lol.

 

[kevin_roche]

New Civil Engineer reports:

Crossrail is to begin systems integration dynamic testing (SIDT) on the central section of the network today.

The testing involves running up to eight trains through the central section of the railway.

Crossrail begins train testing after blockade gets central section up to speed | New Civil Engineer

The testing involves running up to eight trains through the central section of the railway. It comes after an 11 day blockade to get the central section

www.newcivilengineer.com

You can see the trips on open train times.

Trains at Abbey Wood (Crossrail)

All trains at Abbey Wood (Crossrail) on 2020-12-03 around 11:46

www.opentraintimes.com

 

[SeaKing]

Anyone know what units have been involved with this.

 

[JonathanH]

Anyone know what units have been involved with this.

It will come from the allocation of tunnel units - 345024/25/31-33/36/41/50/67/69. Whether that list now includes 345045 and 345046 would be interesting to know.

There have been quite a few late night moves off and on Old Oak Depot to the Tunnel (one or two a day) this week - presumably the units in use stable at Plumstead.

 

[FlippyFF]

It will come from the allocation of tunnel units - 345024/25/31-33/36/41/50/67/69. Whether that list now includes 345045 and 345046 would be interesting to know.

There have been quite a few late night moves off and on Old Oak Depot to the Tunnel (one or two a day) this week - presumably the units in use stable at Plumstead.

Using the RTT link above suggests the tests were starting at Abbey Wood around 8am and finishing back there around 10pm.


Simon

 

[Snow1964]

Using the RTT link above suggests the tests were starting at Abbey Wood around 8am and finishing back there around 10pm.


Simon

Is this the multiple train dynamic testing that was originally scheduled for early 2018, about 9 months before the line was to open

Anyone know how many weeks these tests take to pass, and then there is the obvious follow up question, once passed is there anything technical holding back the trains from being used (and I mean trains not incomplete infrastructure).

 

[kevin_roche]

once passed is there anything technical holding back the trains from being used (and I mean trains not incomplete infrastructure).

I think the update to the train software has to be delivered. There are two software deliveries due and I think one is the TCMS (Train) and the other the signalling. I don't know which is which, but they are both delayed.

In order to successfully complete some of the tests, both have to be in place and working.

I think the update to the train software is needed for 9 car trains to run to Reading but there is likely to be some interaction there with the commissioning of the ETCS Signalling between Acton and Airport Junction. I have no update on when that is due or if it has happened now. Anyone else know?

 

[JN114]

but there is likely to be some interaction there with the commissioning of the ETCS Signalling between Acton and Airport Junction. I have no update on when that is due or if it has happened now. Anyone else know?

“Spring 2021” is the best answer we have at the moment

 

[43096]

“Spring 2021” is the best answer we have at the moment

Informed Sources third law applies: “Always mistrust schedules based on the seasons.”

 

[JonathanH]

Using the RTT link above suggests the tests were starting at Abbey Wood around 8am and finishing back there around 10pm.


Simon

Yes, but none of those schedules actually reported anything so I suspect they didn't actually run - indeed clicking on them shows that they were all cancelled.

RTT shows that, once this starts running, two units stable at Abbey Wood overnight and that six units are scheduled to come from Plumstead Carriage Sidings.

Realtime Trains | Departures from Abbey Wood (Crossrail) between 0200 and 0159 on 07/12/2020

Train information at Abbey Wood (Crossrail) between 0200 and 0159 on 07/12/2020. From Realtime Trains, an independent source of train running info for Great Britain.

www.realtimetrains.co.uk

On another Class 345 matter, there appear to be overnight runs to Langley Reception Siding off Old Oak Common at the moment, including tonight. It seems an odd place to be running test trips to.

There seem to be quite a few different test / mileage accumulation projects in place at the moment on the west side:
* Daytime / evening trips to Heathrow 5Z01-5Z06 (afternoon), 5Z07-5Z12 (evening)
* Overnight trips to Reading 5Z21-5Z26, 5Z51-5Z56 one week, 5Z13-5Z15, 5Z16-5Z19 the next
* Friday / Saturday overnight possession on the Heathrow branch requiring a locomotive drag to Hayes & Harlington Loop
* Friday / Saturday overnight trips to Heathrow on weeks when the above doesn't happen
* Overnight trips to Langley

One of the webinar presentations linked above appears to suggest that the Heathrow branch possessions are Bombardier's chance to test new software. Any ideas on the others?

The daytime trips to Reading and Maidenhead that were presumably mileage accumulation or driver training (5G50-5G55, 5G56-5G61) haven't run since June other than when the paths have been used to test 7-car units after works.

 

[Siemens Staines]

I have seen a quote that the 9-cars with the latest software rev (Y0.545) are;

345021
345023
345028
345030
345060
345065
345068

I think that this means that, for now, only these units should be in use on the Heathrow services. I do not know how accurate the quote is, so perhaps those who view these trains regularly could comment upon whether these are the units in use this week?

 

[kevin_roche]

I have seen a quote that the 9-cars with the latest software rev (Y0.545) are;

345021
345023
345028
345030
345060
345065
345068

I think that this means that, for now, only these units should be in use on the Heathrow services. I do not know how accurate the quote is, so perhaps those who view these trains regularly could comment upon whether these are the units in use this week?

Is the version for trial running still planned to be Y0.603?

 

[JonathanH]

I have seen a quote that the 9-cars with the latest software rev (Y0.545) are;

345021
345023
345028
345030
345060
345065
345068

I think that this means that, for now, only these units should be in use on the Heathrow services. I do not know how accurate the quote is, so perhaps those who view these trains regularly could comment upon whether these are the units in use this week?

Ah, that is interesting - 345021 was out on Wednesday, 345030, 345065 and 345068 yesterday and 345028, 345030 and 345065 reported today. Units not on the list were out earlier in the week.

345023 isn't yet in passenger use but has been a (mainline) test unit for a while, presumably for this software. 345021 and 345028 have also been on those test runs. First passenger work for these two since June.

345060 appears to have done the Langley trips overnight.

 

[Siemens Staines]

Is the version for trial running still planned to be Y0.603?

I am afraid I do not know that. The information I received was probably fourth hand, so I was not even certain it was correct, but JonathanH observations would indicate that it is

 

[JonathanH]

I have seen a quote that the 9-cars with the latest software rev (Y0.545) are;

345021
345023
345028
345030
345060
345065
345068

I think that this means that, for now, only these units should be in use on the Heathrow services. I do not know how accurate the quote is, so perhaps those who view these trains regularly could comment upon whether these are the units in use this week?

345070 reported in service on both Wednesday and today so I suggest that this one should also be on your list.

Friday looks like it was an almost exemplary service with trains almost all on time into Heathrow Airport Terminal 5 (other than an early cancellation and 9P93 1652 Heathrow Airport Terminal 5 to London Paddington delayed by 15 minutes between Terminal 5 and Central).

Today looks good too other than 9P01 0522 Heathrow Airport Terminal 5 to London Paddington delayed 15 minutes after Tunnel Junction, 9P57 1222 Heathrow Airport Terminal 5 to London Paddington which left 10 minutes late and 9T07 1802 London Paddington to Heathrow Airport Terminal 5 which appears to have been delayed after Airport Junction. Looks like there was also a swap between 9P69 1352 Heathrow Airport Terminal 5 to London Paddington and
9T79 1432 London Paddington to Heathrow Airport Terminal 5 as well. Maybe not so good.

Still seems to be an improvement.

 

[Siemens Staines]

That is interesting. I wonder if these were ETCS resets, or other train faults which the software fix would not have addressed?

 

[rebmcr]

That is interesting. I wonder if these were ETCS resets, or other train faults which the software fix would not have addressed?

That's the £18.7bn question!

 

[kevin_roche]

That is interesting. I wonder if these were ETCS resets, or other train faults which the software fix would not have addressed?

It is!

In his monthly report to the London Assemby, Mark Wild said:

Further software upgrades are being introduced in early December for the 9-Car Class 345 trains that operate to Heathrow which should improve the reliability of this part of the fleet, with a further major software release due at the beginning of February 2021.

Having read the last few reports from Jacobs and the responses from Crossrail again (much more carefully this time), I think both these updates are to the TCMS software on the train. My earlier assumption that one was for the signalling system was wrong.

 

[iphone76]

ETCS training for East based drivers is really ramping up now. By Christmas a lot should be FLU / ETCS trained. It is reassuring to know that these issues should be resolved by the time we learn the route out West.

 

[JonathanH]

ETCS training for East based drivers is really ramping up now. By Christmas a lot should be FLU / ETCS trained. It is reassuring to know that these issues should be resolved by the time we learn the route out West.

Does that ETCS training for east based drivers get done on the west side or is this classroom stuff? Is that what the daily Heathrow trips 5Z01-5Z12 are about? I haven't noticed any reports of 345018 out and about on the east side.

 

[iphone76]

Does that ETCS training for east based drivers get done on the west side or is this classroom stuff? Is that what the daily Heathrow trips 5Z01-5Z12 are about? I haven't noticed any reports of 345018 out and about on the east side.

Yes. I believe the first 2 stages are carried out in the training rooms and simulators at Old Oak Common Depot. Then the third module is practical handling on the Heathrow Branch. I am not sure whether the 5Z runs are for that, although I would imagine we would use normal passenger services. I will update later this month when I do the course.

 

[rd749249]

ETCS 3 is degraded ops at OOC, ETCS 4 is practical (3 day course plus assessment).