Tuesday, January 20, 2009

WiFi on Wheels

NOTE: We've moved! Visit us at the California High Speed Rail Blog.

First off, Happy Inauguration Day everyone! I suspect somewhere north of six billion people all around the world may breathe a collective sigh of relief at 12:01pm EST when the Dubya era is finally well and truly over. We'll have to wait and see if President Obama can bring about the fundamental change he has promised, but kudos to the American people for giving the first nominee of African descent ever this opportunity.

For transit enthusiasts, the logistics of the inauguration extravaganza appear promising: the Obamas and the Bidens chose to arrive by train, millions of spectators are using the Washington subway and buses to attend this historic event, cars are banned and there's a strict no-fly zone over the capital. The fly in the ointment is that these decisions were taken for their historical significance and security reasons, respectively, rather than environmental or energy security considerations. So far, the incoming administration is paying only lip service to HSR.

A new post on Treehugger brings the scale of the challenge into sharp relief. Analysis of the lucrative business travel market has shown that at almost 1.4 million trips, San Francisco-Los Angeles was the 7th busiest route in the world in 2004. Given that the alternatives for travel within California are time-prohibitive at present, that 7th place is actually not surprising. Even five years later, corporate etiquette still requires that meetings with important customers and partners should be conducted face to face. Trade shows and skills development are another major driver for corporate travel. Perhaps by now some of the following city pairs have moved further down the list thanks to HSR:

5. Milan-Rome
8. Amsterdam-London
11. London-Paris
13. Marseille-Paris
16. Boston-New York

For business travelers, perhaps even more important than raw line haul speed are fast transportation to and from downtown stations, punctual service and especially, improved productivity en route. A year ago, SNCF began introducing reliable broadband internet access on Thalys and TGV Est, with plans for expanding it to other parts of the TGV network. SNCF's Eurostar service will cut journey times from London to Amsterdam to less than four hours in 2009, though WiFi is still limited to hotspots at selected stations unless you bring your own 3G wireless router.

CHSRA should definitely seek commercial partners to provide no-fuss WiFi connectivity to its customers, preferably based on low latency terrestrial relay networks (802.16e-2005 or later) already tested by Caltrain, albeit at lower train speeds. Videoconferencing and shared desktops are fast becoming good enough for business use, but these applications tend to be both bandwidth-intensive and horribly sluggish over links based on geostationary satellites. CHSRA is not yet doing a good enough job of anticipating the kind of on-board services HSR customers will be expecting a decade from now. Reliable broadband data networking for fast-moving vehicles is a global industry segment that Silicon Valley ought to dominate. However, the initial focus should be on adding value for passengers rather than the even more demanding infrastructure applications (cp. ERTMS Level 2).

For corporate travel policies to take full advantage of HSR, it needs to offer a significant cost advantage over airline travel. For a business, the largest single line item in that equation is the opportunity cost of having someone in transit while he or she is on the clock and racking up expenses. Therefore, boosting productivity is even more important than boosting line haul speed or reducing fares. Perhaps more relevant is that WiFi would allow operators to offer fewer express and more semi-express trains, boosting economic development in the Central Valley - a macroeconomic benefit for the state that the airline industry cannot afford to deliver.

Still, the core business travel market in California will still be SF-LA and, HSR can expect to capture between 50% and 80% of that after a ramp-up period of several years. The exact figure will depend primarily on how the price of jet fuel develops. Note that airlines will figure out ways to offer WiFi as well, but that trains will always offer more seat and desk space. Also, you spend a larger fraction of total travel time actually sitting down in a train, so there's more opportunity to take advantage of any WiFi services on offer. You can use your cell phone while waiting in line to have your shoes inspected at the airport, but fine-tuning your Powerpoint slides? Not so much.

Note that leisure trips were not included in the analysis the Treehugger post refers to. However, if you want cheap HSR fares for the general public, you absolutely have to poach business customers from short-haul commercial aviation. Fortunately, the same networking technology that persuades the suits to choose clean, all-electric trains over dirty, oil-guzzling planes will also attract leisure travelers and tourists - even if they have to pay a little extra for it. Marrying WiFi and HSR could well be the killer application that supercharges ridership between the Bay Area, the Central Valley and SoCal, creating new jobs throughout the state. Now that's change we should all believe in.


Anonymous said...

Two things: first off, Caltrain did try a WiMax-based system but has chosen not to pursue this any further, I guess because of the lack of maturity of the technology. They need a system that will work, do so reliably, and continue working for a decade or two regardless of things like the vendor going under. Onboard WiFi would be a useful bonus, but at the same time, the focus should be on the service itself.

Secondly, ETCS Level 2 is a system that should absolutely NOT be implemented on new lines. That's not what it's for. Basically, the idea is that you remove the wayside signals but keep the track circuit. Instead of a fixed signal, you have a wire going to the central control center, where the central computer processes all this information and radios the trains to tell them how far ahead the line is clear. It makes sense if you already have all the fixed-block infrastructure and are on the way to full ERTMS level 3. But unfortunately, there is no ERMTS level 3 standard yet, not even in any kind of preliminary form, and for all that effort you might as well save on the cost of wires to the central control center, and use coded track circuits instead for wayside-to-train communication, for which standards already exist and which have many systems using them, including of course the Acela, but also the TGV, many freight lines in the US, the entire Russian railway network, and so on.

Rafael said...

@ anon -

of course WiFi would be a service on top of the core high speed transportation. My point is that it can boost ridership without the much greater expense of jacking up line haul speeds ever further. To support a top speed of 220mph, the motors, power electronics, transformer, pantograph, catenaries and brakes all have to be rated at 3.2 times as much electrical power as those on Amtrak's Acela Express.

I never suggested that ECTS level 2 be implemented on the California system. Broadband internet access should be used for applications that are not critical to the safety of train operations.

Alex M. said...

By the time the trainsets are ready to be built and the route is fully under construction in around 2013, 4G wireless connectivity will be fully developed. LTE (GSM 4G) and WiMax will both offer broadband speeds (ability to be as fast or faster than a home internet connection, although those speeds will not be achieved as quickly), significantly faster than satellite internet. Antennas could be placed along the route about every 5-10 miles and the trains could be equipped with receivers and wifi access points for train riders. This system wouldn't be difficult to deploy, as long as it is planned from the beginning. This would also have the benefit of constant cell phone reception too.

Rafael said...

@ Alex -

"This system wouldn't be difficult to deploy, as long as it is planned from the beginning."

Exactly, it shouldn't be an afterthought.

Anonymous said...

Not very difficult at all, only 80 base stations along a 400 mile ROW. And of course there are going to be tunnels, and you'd better hope that WiMax can work with leaky coax antennas, or else you'll have lots of dead zones. It will be, like everything else in this project, a considerable engineering challenge with few analogues anywhere in the world. But I trust our wonderful authority of nine politicians and six employees can manage one of the world's biggest, most complex construction projects.

Alex M. said...

@ Anonymous -

Stop being so cynical about every hill the project must climb. The CHSRA will hire the right people, and there is plenty of expertise in every aspect of the project. None of the problems we will face haven't been met and overcome elsewhere in the world.

Anyways, back to internet access. Offering internet access on the trains is vital to the project. A massive percentage of the people traveling on HSR will be business people who will require internet access to remain productive while traveling. Right now, Aircell Gogo wireless internet is being deployed on airlines across the world. Even domestic airlines like Virgin America and JetBlue have signed up for the system and have already deployed planes with this technology installed. By the time HSR opens, I'm sure every airline in the US will have wireless internet access, including Southwest (the biggest airline rival to CHSR). If internet access is not offered on the train, they will simply take Southwest, prohibiting HSR from gaining important market share.

Anonymous said...

I hope that when one says "internet access" that one means full IP packet connectivity and not some stripped-down (or worse, filtered) http/web-only kind of access as that TGV commercial suggests.

CHSR net access ought to be able to handle the full set of internet protocols, including VoIP, and try to avoid forcing people to operate via proxies and relays. (I have a suspicion that the NAT horse is already too far out of the barn.)

(I won't even ask about IPv6.)

Anonymous said...

"None of the problems we will face haven't been met and overcome elsewhere in the world."

That, by and large, is true, but high speed rail remains on the leading edge of what's possible in rail technology, and it requires a lot of expertise to deal with. Expertise that I am worried doesn't exist in this country, not even to the point where they would have enough competence to hire competent people from abroad. The only reason I seem cynical about every little hill is because together they add up to a freaking huge mountain. All the gung-ho "Yes We Can!" in the world is no substitute for competence and preparation. And you have to beware, when you've got a a 40 billion dollar project, with little in-house expertise and lots of contractors wanting in on the mega-project action and wanting to help themselves to as much of the spoils as possible.

Rafael said...

@ anon @ 2:10pm -

the SNCF system is seamless, you don't lose internet connectivity in long tunnels because they switch to straight WiFi inside them. I'm not sure what kind of cabling they use for the back-end there but I'm pretty sure it isn't coax. Fiber optic, more likely.

As for the number of base stations, I don't see why that should be an insurmountable problem. Any given train car will only "see" a few of them at any given time and choose to connect to the one with the strongest signal. Access authorization will keep out any passengers (or nearby residents) trying to connect to the base stations directly using their own WiMax gear.

@ Karl -

the main reason SNCF decided to use web proxy servers is that the satellite hookup is limited to 10Mbps per train. That's also why they stress mail over other applications like Skype. Another complication of the SNCF system is that it has to switch to 3G or even straight WiFi in stations and tunnels. The handshaking takes a second or two, good enough for TCP services but UDP applications do glitch.

Terrestrial connectivity could supply more bandwidth, lower latency and substantially shorter handshaking delays.

IP v6 is mostly a question of getting customers to adopt it. Chances are, we'll be stuck with IP v4 for quite a while. However, given that each train will have its own firewall, that isn't really a problem. DHCP will give your laptop its address on the train car's LAN.

@ anon @ 7:50pm -


OMG, OMG, OMG! We're spending over $40 billion on a single project! Disaster! Panic! O woe! It's too big! It's too hard! It's too complex! It will never work! We are doomed!

You know, we really should send astronauts to Mars instead.


Anonymous said...

If NASA had a staff of SIX PEOPLE and did the rest of its work by contracting out to Boeing and Lockheed, I'd be just as concerned about it. There are more people discussing the project on THIS BLOG than there are working for the HSRA. It's not the fact that it's a $40 billion project that I'm worried about, it's the fact that it's a $40 billion project run by an agency that hasn't done even a $400 million construction project, in fact hasn't done any construction at all.

As for the base stations, the only reason I mention them is to give an idea of the scale of the work. Leaky coax, incidentally, is a form of antenna often employed in tunnels, where a partially-shielded coax cable is strung along the tunnel and radiates the signal along its length, which works better in a long confined space than having a conventional antenna at the base station. It's all doable, I suspect, but might require resolving a few as yet unresolved technical issues, and of course quite a lot of construction work. It's probably still worth it anyway, as there's plenty of good uses for internet access on a train just in terms of the needs of the railroad (e-ticketing and credit card authorizations being two that come to mind).

Rafael said...

@ anon @ 6:23am -

CHSRA is a planning body. By the time any real money gets spent, additional people - from Caltrans and/or outside consultants - will be brought in to formulate and evaluate the tenders and later, to hold the general contractors accountable during construction.

The only reason CHSRA is not expanding staff and/or consultants is that the State of California doesn't have a budget and can't/won't sell any of the prop 1A bonds right now.

As for the wiring used in short tunnels, it's something SNCF has already figured out. Not sure about long tunnels, the California project won't have any longer than six miles. I'm sure cable IP providers operators have long since figured out how to deal with that kind of distance using amplifiers.

Optical transmission would be preferable, if only because bandwidth demand will grow very quickly. A full-length TGV Duplex has 1090 seats, imagine if 85% were occupied and half the passengers were connected to the Internet via multimedia apps that consume 1Mbps each on average -> 463Mbps! That's well beyond today's average requirements, but then again we made do with 28.8kbps modems just ten years ago.

Plan for success!

無名 - wu ming said...

wifi would be cool, but on a ride that takes 2.5 hours at most, and far shorter than that for regional passengers, i suspect getting some decent coffee, microbrews, wine and snacks will be of roughly equivalent importance WRT sheer comfort of travel.

after all, the trains are so smooth, you don't live in fear of spilling your drink on your pants. might as well have a tasty one in hand as you chat with a friend or just watch the landscape fly by.

James said...

Re: Contracting HSR

Contract execution is critical to keeping the project on schedule and budget. There are a few contractors capable of handling such a scale of project. Bechtel Engeering and Jacobs Engineering would be able to handle parts of the project. Does anyone know of others? CHSR will need to rely heavily on experts from HSR around the world.




eel said...

Is there any hope of getting WiFi service on existing Amtrak trains? I am specifically thinking of the San Joaquin and the Surfliner for my own needs, but I think it would be hugely helpful on all the routes. It is absolutely the case that I could and would travel more frequently if I could do so while staying online for most of the journey. I wouldn't even mind paying in an extra couple of hours of total travel time.

As it is, being able to work on my laptop continuously is part of the reason I am choosing Amtrak for my north-south travel, even with the annoying 'bus bridge' between Bakersfield and LA.

I agree that it is absolutely essential for the HSR line.

Alon Levy said...

Some airlines already have wireless technology on board. I'm not sure what kind it is, but it does allow people to be connected using their own laptops, and use bandwidth hogs like videoconferencing.

Anonymous said...

Can we have a thread about food service?

Alex M. said...

@ Alon Levy -

Like I mentioned before, it is called Gogo by Aircell and it is already offered on some JetBlue and Virgin America planes and will be coming to American Airlines and United Airlines.

Anonymous said...

@ anon from 2:34 pm -

I would love to have a dining car that's a call back to the "golden age" of rail travel. Too bad that isn't possible in today's fast-paced world. :-(

Rafael said...

@ anon @ 2:34pm -

I think food service is one of several opportunities for competing operators to differentiate themselves. Most HSR trains offer a cafe car serving drinks and snacks. Dedicated restaurant cars are offered on many rapid rail intercity trains in Europe, but typically not on HSR.

SNCF's Eurostar service to the UK offers freshly prepared at-seat hot meals for "leisure select" and first class customers. Reportedly, they are much better than airline fare.

Anonymous said...

Rafael: what are your thoughts about the whole "competing operators" thing? It seems like in the rest of the world it's more an aspiration than actual reality, with most HSR services run by their national rail companies: the TGV by SNCF, the AVE by RENFE, the ICE by DB, and so on. The UK does have multiple private operators, but they basically compete for the right to operate a given franchise for a certain time, with government subsidies and some protection from competition. And in that sense, I suppose even California has competing commuter rail operators: Amtrak, Veolia, and Herzog. Japan somewhat an exception to all this, since the JR companies that operate the Shinkansen are indeed nominally private, though I think they're still getting various benefits of former state ownership, and they are mostly operating their own trains on their own track.

Rafael said...

@ anon @ 5:37am -

the situation in Europe is a result of history: once railroads became a vital tool for military logistics, each nation created its own national railroad. Technical parameters such as gauge, signaling and overhead electricity were often deliberately chosen to be incompatible to prevent foreign invaders from leveraging captured railroad tracks.

Japan created one specifically for the shinkansen network, created to overcome the limitations of legacy narrow-gauge lines.

The story is different in the US, where only a very small but important portion of the network (the NEC) was ever nationalized. Amtrak, the national passenger rail service, was created when the freight rail operators decided to discontinue their own - the combination of civilian aviation and the motor car had rendered it unprofitable.

Some corridors and/or trackage rights on corridors were purchased by local and regional public entities. For example, the Joint Powers Board representing San Francisco, San Mateo and Santa Clara counties purchased the Caltrain ROW. Other county-funded passenger rail services include BART, LA Metro, Metrolink, NCTD and (soon) SMART. The state of California has partnered with Amtrak to deliver regional service in the Pacific Surfliner, Capital Corridor and San Joaquin corridors.

In other words, heavy rail passenger service in California is a patchwork of individual services administered as bureaucratic fiefdoms. There are few interconnections, no unitary fare structure, no tickets covering multiple services within a geography.

HSR differs in that it will run on dedicated tracks, with very few exceptions. These tracks (and other fixed infrastructure) will belong to a public-private partnership. In theory, this PPP could also be the monopoly operator of all trains, but there is no need for that: there are no foreign invaders to worry about and, no sections on the HSR network are expected to require operating subsidies. Unlike Amtrak and commuter rail, HSR should be profitable.

In general, profits attract competition, which is a good thing for customers. That means the PPP should auction off slots on the timetable to private companies that lease or purchase trainsets from a shortlist of products that the PPP has pre-qualified. In principle, it would even be possible for train operating companies to act as wholesalers of HSR seats, leaving the business of retail sales to a third level of companies.

Trainset maintenance services could be provided directly by the PPP - which is responsible for overall safety - or by third party technicians certified by the PPP. Those technicians could work for the train operators or specialized companies.

Proceeds from the (annual?) slot auctions would pay for maintaining and expanding the infrastructure. Billing arrangements for the electricity are another issue.

The one reason why this scenario may not come to pass is that the State of California and individual counties may well prefer to use profits from HSR to cross-subsidize existing commuter rail services. If the PPP ends up as the monopoly train operator, it will be able to charge higher fares and offer lower service quality.

The big question is if the higher margins would make up for the inevitable loss in ridership, relative to competing train operators. My hunch is that it wouldn't. The biggest cost center for any railroad is constructing and maintaining the infrastructure, so the way to maximize profits to the state is to maximize ridership. This is essentially why SNCF is buying more TGV Duplex trainsets with top speeds of 186mph, rather than single-level AGVs rated at 224mph.

Anonymous said...

An interesting proposal, and just for the record, I think that's more or less where things should be headed in the rest of the world. Unfortunately I haven't heard anything from HSRA saying that this is how it will work, and I don't know of anywhere in the world that uses that model.

James said...

@ Rafael

" This is essentially why SNCF is buying more TGV Duplex trainsets with top speeds of 186mph, rather than single-level AGVs rated at 224mph."

Would it be practical to have boardings platforms to match doors on both the first and second level of duplex trainsets?

Rafael said...

@ james -

in principle, I suppose, it would be possible to design trainsets and stations that could support simultaneous boarding and alighting on both levels of a bi-level train.

In practice, however, it has never been done - not even in Japan. Sixteen-car E4 Max trains offer 1634 seats, though usually only eight-car trains are deployed. JR uses wide-bodied trains featuring five (narrow) seats abreast in economy class.

Dwell times at shinkansen stations are usually just 50 seconds and drivers are expected to follow schedules to the second. This is necessary as there may be as many as 50 bullet trains moving on a single shinkansen line at one time.

During rush hour, shinkansen passengers must make seat reservations and are expected to form orderly queues as marked on the platforms. I'm not sure if JR has already gone one step further and enforced one-way traffic in the aisles of the individual cars. SNCF also requires reservations on all TGV trains, including the ones based on two Duplex trainsets.

Basically, the experience in France and Japan appears to have been that a single level of doors combined with flights of stairs to reach the upper level is sufficient for HSR service, provided the pedestrian flow between platform and train is well managed.

Local commuter trains in Japan are an entirely different matter, they could use the extra capacity of a second level. My guess is that trains are already as long and headways between them as short as possible and, that tunnels along some routes cannot accommodate bi-level cars. Stretching out rush hour to avoid such pile-ups is probably infeasible in Japanese corporate culture.

Alon Levy said...

My guess is that trains are already as long and headways between them as short as possible and, that tunnels along some routes cannot accommodate bi-level cars.

My understanding is that in Japan the limiting factor is the speed at which a train can be loaded and unloaded; bi-level cars reduce the number of doors available, creating congestion even as they increase the capacity of each individual train.

Rafael said...

@ Alon Levy -

as the video I referenced shows, dwell times are indeed an issue for local commuter trains during rush hour in some parts of Japan. However, this is because the trains are already full by the time even more passengers want to board, not because of any bottleneck at the doors themselves.

There just isn't enough standing room, never mind seats. It's called crush capacity for a reason.