A year ago we chaired and spoke at a conference entitled the Gigabit Train. It was a solemn affair which was mostly full of people from the train industry who were laboring under a government ruling that all UK trains should have free WiFi by the end of 2018.
Everyone agreed that the result would be exceptionally poor train WiFi until the target was reached and then it would be quietly shelved as the government changed and something commercial would be put in its place.
The key issues where that a) People would not pay for WiFi on trains, that much had been made clear by early experiment and that b) Backhauling it over LTE would leave train carriers hostage to fortune as cellcos offered low rates for a while and then charged the earth once the train operator could not afford to “cancel” the service. We can vouch for a) because it leads to spending 30 minutes trying to work on a train, where it takes you 20 minutes to sign up for paid WiFi, and by then you have to collect your things to leave, having wasted the money. Also, LTE backhaul is notoriously poor. If LTE falls back to GSM in rural regions, then there is no backhaul and the WiFi does not work.
Self-driving cars may be with us before this is solved, and they will obliterate the problem by offering LTE or 5G based backhaul for in-car WiFi, and could replace trains entirely, if they are embraced at scale.
For now, it is a problem which needs to be solved. We are going to ask you, our audience, how you would solve it and then take a stab at it ourselves.
But first let’s take a look at the US Amtrak solution which was being proposed last week. Amtrak wants to build its own Track Side Network using the same spectrum as WiFi, in the UNII-1 and UNII-3 bands. Of course, UNII-1 is an indoor band and Amtrak wants a waiver on the type of power that it uses for the connection. Presumably this is still a backhaul connection to an on-train router that drives antennas in each carriage. UNII-3 is set aside for outdoors and can go up to 1 Watt power usage, and it wants the same for UNII-1 so that it halves the cost of building this network.
Now power variation is allowed when a beam is focused in a point to point relationship and Amtrak just wants those rules applied. A kind of point to moving point. Up steps the WiFi Alliances pointing out that Amtrak needs to conduct tests to see if it would interfere with other WiFi users along the tracks – almost certainly it would, in some isolated places. Imagine not being able to have WiFi in your home because you lived next to a rail track.
We would have thought that a constant cost relationship could be built between a satellite backhaul supplier and a train operator quite easily. Satellite operator are happy to keep prices for ten years or more and then prices would be likely to go down, not up. But in the past when we have suggested this, everyone has said satellite cannot do the trick. But that’s not strictly true.
New HTS (High Throughput Satellites) may well be able to manage this, and could use different spots beams on different train routes, simultaneously and combine the data allocation. It can, in effect, follow the train.
Now there is a half second propagation delay, and it would disappear when you go into a tunnel. But actually, so would LTE unless the tunnels all had picocells placed inside them.
Older satellites charged something like $100 million for 1 Gbps for a year, HTS is targeting $3 million for that service, but that data can be shared across many trains simultaneously. It is a cost, and nowhere near as cheap a cost as spending all the capex up front on your own network and then having a tiny amount of Opex backhauling that network through an ISP in one or two big fiber connections.
But right now, HTS’ are just being put into space and we’re sure a train service would be able to cut a sweet deal way under that type of data level and price. Having access to train travelers and a confirmation and terms web-page, would provide a chance to sell local service to rural broadband customers. It would give the satellite operators access to a broader public market.
The big issue in the US is that satellite broadband has a terrible name, associated with rip offs and high prices and detestable customer service. Viaplay in the US, who runs the Wildblue home broadband service in the US, is now advising European satellite operators on how to sell satellite broadband to the public.
A hybrid system using satellite for certain parts of certain routes, trackside backhaul in the UNII-1 and UNII-3 bands for parts of the trackside a set distance from any give home, to eliminate interference issues, and LTE or 5G for a premium superfast, but a paid service might well work. And again, that shopfront opportunity on a terms and conditions page is there for cellular operators to say, “We are providing this excellent free WiFi for you on the train, but it is only free to a) Our mobile customers and b) Would you like to become one of our mobile customers.” That might mitigate backhaul costs for the train operators down to zero.
Once 5G arrives it could offer cheap backhaul to trains up to a particular data cap per journey, running it in a virtual network slice – but having people use it for watching videos on a journey over 1 hour will frighten off most train operators from committing to a free WiFi service, so some form of layered service will end up being required. Ask every hotel in the US and Europe, which has already reached this same conclusion.