Driverless trains - why limited progress on the national rail network?

[jon0844]

Can just have the computer go directly to emergency braking if signal loss is identified.

I would imagine everything fails safe if the 5G network goes down, which means stopping. Which means chaos.

We are still going to be people, with legs and arms who will seek to find ways to keep moving. So what happens then? People deciding to jump off trains (who will be there to stop them?) and people walking along roads (again, who will stop them)?

All real issues that technology can be aware of, but can't necessarily cope with in a satisfactory way. Well, unless we change laws so if you get near to a driverless car/train and it hits you, tough luck. Wait until there's a chance for a driverless vehicle to come sweep up the body and things carry on.

 

[NSEFAN]

Being claimed on another thread ... so more lies and damned lies

If you're referring to GSMR, then it is supposed to provide 100% coverage at all times to the rail network, however there have been cases where coverage drops out, or the GSMR handsets try directing calls to the wrong signalling centre. This is somewhat disappointing, given how much was spent on rollout (around £1bn I believe).

 

[The Ham]

Why do you think that? Highway driving is the easiest to automate: all the vehicles are going the same direction, at roughly the same speed and there are relatively few obstacles.

The easiest way to power an automated car is by electricity, if the plug isn't connected right the power is unlikely to flow. If the automation messes up on the refilling of a hydrocarbon powered vehicle then the results are likely to be more of a problem and if nothing else likely to be more costly.

As such, although long distance automation would work for programming purposes it could mean bringing back forecourt attendants or result in very large battery packs.

200 miles is likely to be OK for a lot of journeys, but there are still significant journeys where people wouldn't want a 40 minute delay on a 4 hour journey, when a 3 hour journey wouldn't have such a delay, just to top up the batteries.

 

[najaB]

The easiest way to power an automated car is by electricity...

Most of the self-driving test cars on the roads at the moment are powered by internal combustion engines. Why that should suddenly change I don't understand.

If the automation messes up on the refilling of a hydrocarbon powered vehicle then the results are likely to be more of a problem and if nothing else likely to be more costly.

The US Air Force has demonstrated automated in-flight refuelling - that has to be orders of magnitude more complex than filling a stationary vehicle at a stationary pump.

 

[NotATrainspott]

The easiest way to power an automated car is by electricity, if the plug isn't connected right the power is unlikely to flow. If the automation messes up on the refilling of a hydrocarbon powered vehicle then the results are likely to be more of a problem and if nothing else likely to be more costly.

As such, although long distance automation would work for programming purposes it could mean bringing back forecourt attendants or result in very large battery packs.

200 miles is likely to be OK for a lot of journeys, but there are still significant journeys where people wouldn't want a 40 minute delay on a 4 hour journey, when a 3 hour journey wouldn't have such a delay, just to top up the batteries.

Not quite.

Charger prototype finding its way to Model S

Cables capable of significantly higher power delivery are going to be heavier and more cumbersome than current ones, making it even more important that the technology exists for them to plug in automatically. With this automatic plug-in there's no need for wasteful wireless charging for range. It probably will have a future but only to allow trickle charging or enough power for AC or heating when commercial vehicles will be waiting around but not somewhere suitable for fast chargers.

Most of the self-driving test cars on the roads at the moment are powered by internal combustion engines. Why that should suddenly change I don't understand.The US Air Force has demonstrated automated in-flight refuelling - that has to be orders of magnitude more complex than filling a stationary vehicle at a stationary pump.

Automation and electrification will go hand-in-hand. Many of the problems of electric vehicles are at worse mitigated and at best solved completely by autonomy. Tesla were very clever in targeting their first vehicles at wealthy drivers, as not only do they have the money to spend on the technology but they are also much more likely to have the means to charge their cars at home or work. As EVs get more affordable, more city dwelling types without their own parking or charging facilities will think about having one, but it'll never be worthwhile nor practical to fit every street parking bay or small domestic home with a charger. With autonomy, you need only provide a small number of ultra-fast charging stations in central locations where high-wattage power supplies are already available.

 

[najaB]

Automation and electrification will go hand-in-hand.

I agree, but neither is dependant on the other. The trend towards electric and hybrid drivetrains was going to happen simply for efficiency.

 

[ComUtoR]

If you're referring to GSMR, then it is supposed to provide 100% coverage at all times to the rail network,

There are places where the GSMR drops out on every trip.

or the GSMR handsets try directing calls to the wrong signalling centre.

Again, there are places where it will always call the wrong box. The most common, is calling the underground box.

 

[Tim M]

I have only just come to this thread so some of what follows may repeat others comments. There are distinct stages in defining automation of train operation:

- Manual driving with the driver responsible for safety in accordance with lineside indications, signals or telephone orders

- Manual driving with additional protections such as AWS and TPWS in the cab, but not to full fail safe (SIL4) standards

- Manual driving with intermittent SIL4 Automatic Train Protection (ATP), e.g. GW main line and Chiltern

- Manual driving with continuous SIL4 ATP, e.g. HS1, Channel Tunnel

- Automatic Train Operation (ATO) (SIL2) with continuous ATP and driver managed, e.g. LU Central, Jubilee, Victoria lines etc.

- ATO with ATP attended, e.g. DLR

- ATO and ATP unattended, no current examples in the UK apart from airport people movers at airports

The last one is the true driverless railway, but requires some thought to understand the implications, something I have done when quoting for mass tnsit signalling in Singapore (new) and Helsinki (upgrade). There are two major issues:

1. obstacles on the track, particularly people falling off platforms for which the solution is Platform Screen Doors per Jubilee Line Extension;

2. recovery from failure, how to drive the train with no trained staff on board. This is possible but can be expensive and probably needs forward looking CCTV with 'joystick' driving from the Control Centre as a minimum.

One of the primary reasons for Singapore and Helsinki wanting unattended operation was the ability to bring extra trains into service without having to find train crew, but this does require fully automating depot operation.

Mention has been made of ATO slip & slide and weather. Building on extensive experience in Hong Kong, Singapore and Madrid, the good chaps at Chippenham (yes, that's where I worked) rose to the challenge to make it work, and used their knowledge on Central Line, altogether nearly 40 years experience.

Unattended operation on the main line is more problematical for the two reasons mentioned. The problem with objects such as trees on the line is not just knowing how to avoid them but what to do when the object is hit? For example just consider a person being hit, fencing only works up to a point.

 

[HSTEd]

There are places where the GSMR drops out on every trip.

That is actually not a huge issue - in areas where GSMR performance is always unavailable it can be fixed with gap fillers.

The problem is if we got all sort sof horrible transient drop outs that were not so simply (if not cheaply) fixed.

 

[ComUtoR]

That is actually not a huge issue - in areas where GSMR performance is always unavailable it can be fixed with gap fillers.

When the GSMR drops out coming into London Bridge its not an issue ?

Ever since the introduction of GSMR there has been black spots. One of our biggest was at a point where there was no communication available with the Signaller other than getting out and using the SPT.

I do agree that it seems to be easily fixed but my concern is that those fixes take months to appear.

I give NR its due and you do see the dude with the radio pack on his back checking reception etc and they are addressing the black spot issue but we have spots appearing that were never there before. Personally, I see it getting worse, not better.

 

[Sunset route]

When the GSMR drops out coming into London Bridge its not an issue ?

Ever since the introduction of GSMR there has been black spots. One of our biggest was at a point where there was no communication available with the Signaller other than getting out and using the SPT.

I do agree that it seems to be easily fixed but my concern is that those fixes take months to appear.

I give NR its due and you do see the dude with the radio pack on his back checking reception etc and they are addressing the black spot issue but we have spots appearing that were never there before. Personally, I see it getting worse, not better.

I agree before they use GSMR as any sort of safety critical carrier it needs to be sorted out. There have been a national outage, two south east England outages, loss of local cells, miss directed calls, wrong areas, black spots still and this is only in the last year. This system has been running for 3 years now and it's barely good enough for voice comes let alone signalling and automation.

 

[The Ham]

Most of the self-driving test cars on the roads at the moment are powered by internal combustion engines. Why that should suddenly change I don't understand.The US Air Force has demonstrated automated in-flight refuelling - that has to be orders of magnitude more complex than filling a stationary vehicle at a stationary pump.

That's perhaps because the current self drive cars so require a human supervisor, who can be relied upon to do the refiling.

If the air forces (as it's not just the US air force that does it) spil some fuel then it can be a major problem and so have some very clever safety features, the likes of which are not filtered to standard fuel pumps. Clearly and filing is possible, in that hydrogen pumps have been developed to do so, but why bother when electricity is an easier system to deal with.

It also has to be remembered that currently we are talking about costs to the end user comparable to taxi fares, which would mostly limit prior to shorter journeys (fairly likely to be sub 50 miles) unless people have deep pockets.

 

[The Ham]

Deleted

 

[quantinghome]

What it the point of a driverless train? From a safety perspective, surely ATP + driver is the best option, having the advantages of human eyeball plus a safety system to remove elements of human error.

I can't see it reducing fares significantly either. The investment needed for a truly driverless operation would mean it could only be done on intensively used services - busy intercity and commuter routes. The cost per passenger journey attributed to the driver, even including overheads, would be rather small for these services, so where's the gain?

 

[jon0844]

If you're referring to GSMR, then it is supposed to provide 100% coverage at all times to the rail network, however there have been cases where coverage drops out, or the GSMR handsets try directing calls to the wrong signalling centre. This is somewhat disappointing, given how much was spent on rollout (around £1bn I believe).

5G is going to be reliant on a network of sites that is unthinkable right now. With mesh networks (where one site can feed another) the idea is that you'll be able to offer pretty much 100% coverage with loads of small/tiny cells.

As 4G rolls out, small cells are already being used to give good/fast/reliable connections in previously hard to cover locations. Vodafone for example as used small cells (or is in the process of doing so) to fully cover London Victoria station and Gatwick Airport (both terminals and station). It has also told me that it now has contiguous 4G coverage from Victoria to Gatwick for trains.

As time goes on we'll see blanket 4G coverage that 2G and 3G could never manage. GSM-R is a bit different as it's a network purely for the railway, but in the future the sites could be used for an IP-based replacement, similar to the network being rolled out by EE for the emergency services.

It won't mean we won't have outages, especially as in the future the reliance on everything being connected makes all those far-fetched movie scenarios where hackers bring down a city from a computer isn't actually going to be that far-fetched. The next wave of terrorism will almost certainly be based on breaking into networks to cause problems and bring people to a standstill.

For what it's worth, 5G and 4G aren't that dissimilar. Both are IP based. It's just that 5G will probably be distributed in a different way, allowing faster speeds and more resilience. For the end user, it will work pretty much just the same - just faster. Rather like going from ADSL to FTTC to FTTH broadband. It isn't the step change that 2G to 3G was for data, or 3G to 4G for voice.
--- old post above --- --- new post below ---

What it the point of a driverless train? From a safety perspective, surely ATP + driver is the best option, having the advantages of human eyeball plus a safety system to remove elements of human error.

I can't see it reducing fares significantly either. The investment needed for a truly driverless operation would mean it could only be done on intensively used services - busy intercity and commuter routes. The cost per passenger journey attributed to the driver, even including overheads, would be rather small for these services, so where's the gain?

I doubt it will be about reducing fares, but there are clear benefits if a train can go anywhere because a driver doesn't need the same route knowledge. I would assume that you'd have a standardised system so a driver could sit in any train without additional training.

I'd still hope that we kept drivers who maintained a safety role and were able to take over during disruption, or when certain networks that the train relies on may go down. After all, our future will rely on connectivity - but it WILL fail from time to time, whether by accident or maliciously.

I am sure unions would be fighting for this to be the case, and I think the general public would probably also want to know there's someone in a huge juggernaut coming down their road - although who knows whether someone who isn't hands-on would be able to step in quickly.

I think for roads, we have much bigger issues than rail. On a train, a 'driver' just needs to observe and the only time to get involved is an emergency situation or after the train stops and can't proceed by itself. At that point there would be a procedure to get the train moving (likely in a sort of limp mode) until the network is back up again.

I think a lot of fans of the future forget that there will be a massive potential for problems in this utopian future. Problems that, if bad enough, could cause civil unrest. Imagine a hack that brings down a network that controls all trains and road vehicles, or brings factories and distribution centres to a standstill.

If we get rid of all the people today, who steps in when we go back to 'pen and paper' style operation? As we know that air traffic control currently does. No good if we assume that in the future we just let computers get on with it entirely on their own.

 

[squizzler]

To my mind it only seems worth attempting totally automatic trains if we were looking at a complete rethink of how the railway runs. This article on something called "Seamless Interchangeability"seems to be the most interesting work on the subject: not that I see it happening by 2040...

 

[najaB]

That's perhaps because the current self drive cars so require a human supervisor, who can be relied upon to do the refiling.

It's because self-driving does not mean electric. There will be self-driving hybrid and pure IC vehicles too.

If the air forces (as it's not just the US air force that does it) spil some fuel then it can be a major problem and so have some very clever safety features, the likes of which are not filtered to standard fuel pumps.

What's the obsession with 'standard' fuel pumps. If we're talking about changing the world it's not hard to imagine changing the petrol station. In the future there may well be no petrol station at all - fuel may be delivered by roving tankers, or cars may drive themselves to fueling points in downtime.

It also has to be remembered that currently we are talking about costs to the end user comparable to taxi fares, which would mostly limit prior to shorter journeys (fairly likely to be sub 50 miles) unless people have deep pockets.

Or the vehicles are in shared ownership syndicates.
--- old post above --- --- new post below ---
This showed up in my Google+ feed (they are watching!): NISSAN’S PATH TO SELF-DRIVING CARS? HUMANS IN CALL CENTERS

 

[jon0844]

Car stops.

Press assistance button.

Music plays. 'Your drive is important to us, please hold'.

10 minutes later, call is answered by 'Kevin from Customer Services' with strong Indian accent who asks what your problem is and repeatedly says 'I understand' while somewhat missing the point of the problem.

After establishing the problem and speaking to his manager, he takes control of the car to steer around the traffic cone in the road, and you proceed on your journey 30 minutes late, trying to remember a time when you could actually drive yourself.

 

[bramling]

Car stops.

Press assistance button.

Music plays. 'Your drive is important to us, please hold'.

10 minutes later, call is answered by 'Kevin from Customer Services' with strong Indian accent who asks what your problem is and repeatedly says 'I understand' while somewhat missing the point of the problem.

After establishing the problem and speaking to his manager, he takes control of the car to steer around the traffic cone in the road, and you proceed on your journey 30 minutes late, trying to remember a time when you could actually drive yourself.

That's not a million miles different to how ATO drivers feel on the Underground, particularly the Jubilee and Northern lines where the system is not user-friendly, nor conducive to keeping things moving naturally when there is some kind of failure.

 

[Tim M]

That's not a million miles different to how ATO drivers feel on the Underground, particularly the Jubilee and Northern lines where the system is not user-friendly, nor conducive to keeping things moving naturally when there is some kind of failure.

Any chance you could expand on why the systems on these lines are not user friendly say compared to the Central and Victoria lines?

 

[bramling]

Any chance you could expand on why the systems on these lines are not user friendly say compared to the Central and Victoria lines?

Where does one start!

For a start, the impression one gets is it's designed by a bunch of software computer programmers, rather than railway engineers. Secondly London Underground has had to fit round it, rather than the other way round. Nonetheless it's reasonably reliable when it works, but ...

... the architecture is very centralized, so when things fail you can lose large areas. In the worst case the entire line can go down. On the Central Line (and I think the Vic, although I'm not 100% on their local procedures) the driver can continue in Restricted Manual, on their own authority if necessary, unless or until they reach a semi-automatic area - which you'll only find in places where there are points.

On the Jubilee and Northern Lines the train can only work in Restricted Manual under the signaller's specific authority.

To work a train in RM across points, the signaller has to:
1) Identify what the move will be, and ensure there will be no conflicting moves.
2) Ask the train to select RM and await further instructions, then stop that call.
2) Cancel the Auto Manual Route Reservation which the system will automatically generate when it detects a train has gone into RM.
3) Generate a new Manual Route Reservation plus a Route Secure Reservations for any points.
4) VCC Technician confirms (although I believe this requirement has been dropped *providing* the train operator can see an indication conforming points are set).
5) Instruct train to proceed in RM to agreed limit of move.
6) Once completed, the signaller has to cancel all of this, then re-assign the train to the timetable.
7) If the train encounters an "RM Hold" board or station platform in the interim, the driver has to be further authorized, although it doesn't normally require all the mouse work to be re-done.

Bear in mind if you get a loop failure or an intersig failure *every* train through the affected area will have to be authorized in this way over a considerable distance, one can see it is an enormous workload for the signaller, plus the process is so time-consuming that you can barely run any trains through the section. Under conventional signalling, a signal failure on plain track could be run through relatively simply with minimal direct workload for the signaller.

As regards manual driving, any Jubilee or Northern line driver will tell you that the system's full-speed Protected Manual driving mode is a highly unpleasant experience, for various reasons. Having said that, the ATO is so rubbish that - despite the PM being non-user-friendly - it's still possible for a human to better the ATO's running timings. Meanwhile, the system struggles to cope with certain gradient profiles or any sign of poor adhesion, so platforms that used to be entered at 35 or 40 mph are now entered at around 22 mph. Don't you just love progress?!

 

[notverydeep]

Most of the self-driving test cars on the roads at the moment are powered by internal combustion engines. Why that should suddenly change I don't understand.

While an Autonomous Road Vehicle (ARVs) can be electric or internal combustion powered, or hybrid, it is the way that these vehicles will probably be owned and operated is likely to mean that many more will be electric. The current model could be described as mass individual ownership, that is most cars are owned by one household and are exclusively used by that household. Even fleets often allocate each car to a particular employee. There are exceptions of course, you can take a taxi or make use of car clubs or longer term car hire.

This ownership model builds in very significant inefficiency into the way cars are used. Train operators expect a peak utilisation of their fleet around 90% or more. That is 9 out of every 10 units they have will be in service at the busiest times. The average for cars is barely 10%. That is, even when the roads are at their most congested, 90% of cars are parked up somewhere! If you don’t believe this, walk down the street where you live at 1 am and count the cars and then on a weekday repeat the count at 08:30.

In addition to this many cars are sized for the largest number of people and luggage an individual wants to be able to carry. Despite this, most of the time, these cars do short commutes, with just one person. My car doesn’t even commute, usually I just do short weekend journeys and quite a few of those with only one occupant, but to facilitate a couple of holidays with the children in self-catering accommodation and all the clobber that implies, I have a Ford Focus estate.

The alternative model that many envisage for ARVs would be analogous to that of taxis. These cars would be hired for an individual trip and then go onto another with another use, and another trip etc. As the hiring would likely be by app, the user would specify in advance what sort of vehicle was needed (at least the broad category). For a single occupant going a short distance this would be a small electric vehicle in most urban areas. Even with current electric vehicle range, say 150 kms on a single charge, such a vehicle could be in service for five or six trips, before taking itself to a charging point. This would not be enough to take my family from Welwyn Garden City to the Lake District, so the ARV I would summon for that trip would still be either hybrid or internal combustion engine with today’s battery limitations.

This change of models is possible, because without the labour cost of the driver, a taxi would cost about the same per trip as your car does. It is possible that ARVs implemented this way would cost even less as some of the fixed costs you have will be spread over much more utilization of each vehicle.

If we woke up in this world tomorrow, a good majority of vehicles could be those small, electric cars and only a minority would need to be good for long distances with bigger loads. In practice, this world will only come about gradually over the next couple of decades and hopefully, battery technology will improve to allow a higher proportion of vehicles to go without their internal combustion engine. I would guess that large commercial vehicles will be the most difficult to electrify.

Even with all the advantages outlined above, mass transit of some form is likely to be necessary for dense urban flows. ARVs will improve the efficiency of how road space is used (particularly once driven vehicles reduce), but not to the degree that would allow every London commute to be done this way – even if every last piece of rail infrastructure were converted into dedicated ARV roads. For railway routes carrying fewer than perhaps an average of 1000 - 2000 passengers per hour, the above - if and when it comes to pass - will be very bad news indeed.

 

[najaB]

While an Autonomous Road Vehicle (ARVs) can be electric or internal combustion powered, or hybrid...

Didn't want to quote the entire post, but wanted to say that we are in complete agreement as to what the future looks like.

 

[squizzler]

The discussion of driverless cars raises interesting points, although I suspect most of what we know about these stem from press releases and investor prospectuses, and is therefore what the developers what us to know rather than independent engineering analysis.

While autonomous cars are very much a threat to secondary and branch lines, this all assumes railways survive in their current form.

If you look at this inherent advantages of traditional rail (speed, comfort and capacity) it isn't impossible with automation and other emerging technologies some of these advantages will not be so pronounced in future.

Speed, comfort and capacity. Pick two. e.g. light "pod" type cars going at a low constant speed will release capacity without discomfort. Try thrashing these"pod cars" and you regain speed at the expense of comfort. And so on.

For example. rail capacity advantages could be eaten away by "virtual coupling" or automatic coupling of road vehicles (buses and trucks) run in platoons. Convert a few railways to busways, running at higher speeds, with vehicles coupled together run automatically and you have a serious threat to traditional rail. Maybe, these could be 'road railers' with or without Paris style rubber/steel tyre running? Automatic ( road/rail?) buses could traverse roads, then be (virtually) coupled together over old rail routes and run at much higher speeds. Ditto trucks.

If highway capacity will be unlocked by automation, surely there is no need to turn old rail-beds into motorways. If these things do what they say on the tin we will more likely see a maintenance crisis on underutilised highways.

Arguably motoring is facing threats to its dominance due to environmental factors and technology to a much greater extent than the railways did in the 1955 modernisation plan. And at least the railways did have a plan, however badly conceived!

As for those who assume firms successful in the field of computing will inherently be also successful introducing a radical new type of motorcar, I'd point out the Sinclair C5.

 

[notverydeep]

As for those who assume firms successful in the field of computing will inherently be also successful introducing a radical new type of motorcar, I'd point out the Sinclair C5.

I suspect that the developers of Autonomous Road Vehicles are much, much better funded and resourced than Sinclair was...

 

[NSEFAN]

5G is going to be reliant on a network of sites that is unthinkable right now. With mesh networks (where one site can feed another) the idea is that you'll be able to offer pretty much 100% coverage with loads of small/tiny cells.

As 4G rolls out, small cells are already being used to give good/fast/reliable connections in previously hard to cover locations. Vodafone for example as used small cells (or is in the process of doing so) to fully cover London Victoria station and Gatwick Airport (both terminals and station). It has also told me that it now has contiguous 4G coverage from Victoria to Gatwick for trains.

As time goes on we'll see blanket 4G coverage that 2G and 3G could never manage. GSM-R is a bit different as it's a network purely for the railway, but in the future the sites could be used for an IP-based replacement, similar to the network being rolled out by EE for the emergency services.

It won't mean we won't have outages, especially as in the future the reliance on everything being connected makes all those far-fetched movie scenarios where hackers bring down a city from a computer isn't actually going to be that far-fetched. The next wave of terrorism will almost certainly be based on breaking into networks to cause problems and bring people to a standstill.

For what it's worth, 5G and 4G aren't that dissimilar. Both are IP based. It's just that 5G will probably be distributed in a different way, allowing faster speeds and more resilience. For the end user, it will work pretty much just the same - just faster. Rather like going from ADSL to FTTC to FTTH broadband. It isn't the step change that 2G to 3G was for data, or 3G to 4G for voice.

Although mesh networking would help with resilience, I'm not sure how useful mircocells would be for safety-critical railway comms, given the railway is largely linear and trains can travel quite fast. I also can't see the thought of putting safety-critical traffic on the global internet going down too well either; as you've said it'd be more vulnerable to attack. I would therefore expect the railway to use its own allocated frequencies and physically private network, but using some of the more useful features of 5G where appropriate.

 

[najaB]

I also can't see the thought of putting safety-critical traffic on the global internet going down too well either; as you've said it'd be more vulnerable to attack.

IP ≠ The Internet.

 

[jon0844]

Although mesh networking would help with resilience, I'm not sure how useful mircocells would be for safety-critical railway comms, given the railway is largely linear and trains can travel quite fast. I also can't see the thought of putting safety-critical traffic on the global internet going down too well either; as you've said it'd be more vulnerable to attack. I would therefore expect the railway to use its own allocated frequencies and physically private network, but using some of the more useful features of 5G where appropriate.

Yes, the railway would have to use its own network like now.

By then, chances are the sites may be shared with public networks. Heck, the Government of the day may have changed the rules on the use of all sites and the railway will have no more rights in site locations/heights as anyone else.

 

[NSEFAN]

IP ≠ The Internet.

I know that. I was (poorly) addressing the point that just because the railway is surrounded by GSM networks doesn't mean it should use them, given the safety-critical nature of the comms. As much as possible the network should be physically private, as VPNs and tunnelled traffic can be remotely compromised far more easily.

By then, chances are the sites may be shared with public networks. Heck, the Government of the day may have changed the rules on the use of all sites and the railway will have no more rights in site locations/heights as anyone else.

Site sharing is less of an issue, as I would expect phone companies to already keep a close eye on who has access to base stations, and the GSM-R base station would probably have its own additional security anyway.

 

[WelshBluebird]

IP ≠ The Internet.

Something doesn't have to be on the internet to have a risk of malware, hacking, infiltration or have other weaknesses.

GSM itself is pretty damn insecure. I am aware that GSM-R is more secure but I don't know the details well enough to know if that actually means it is secure or not.