Can You Actually Work on Amtrak’s Free Wi-Fi? We Tested It to Find Out
We tested the free Wi-Fi against cellular on three of the most heavily trafficked routes in the nation to see whether you can expect to get work done, or even just relax and watch streaming video.
By Sascha Segan
Americans like to be on the move. The AAA says more than 100 million people traveled for the 2021 holidays, even with the threat of Omicron sweeping the nation. And while you can drive with your phone in hotspot mode or pay through the nose for airline Wi-Fi, buses and trains have one advantage over those means of travel: free Wi-Fi.
Amtrak has offered Wi-Fi on some trains since 2010, and on longer-distance routes since 2016. With some new test software to try out, I thought I’d hop on a train and see whether the notoriously questionable Wi-Fi quality has gotten any better in recent years.
In 2019, before the pandemic dug into business travel, Amtrak served up 32,519,241 rides. More than 38% of those were on the eight-state Northeast Corridor spine, from Boston to Washington. (The next most popular route is the Pacific Surfliner from LA to San Diego, with 8.5% of rides.) About 14% of travelers rode on Amtrak’s long-distance routes, such as the epic New York to Chicago or Chicago to Seattle runs.
So I picked three routes to test. First, train 190 runs from New York to Boston in the morning. It runs through commuter towns and mid-size cities on the Northeast Corridor, along tracks owned by Amtrak and commuter railroads.
Train 448 is part of the long-distance Lake Shore Limited to Chicago, crossing rural Massachusetts partially on a single-track CSX freight line. Amtrak has little control over what happens there, so it’s my representative of the cross-country trains.
Finally, Train 64 runs from Toronto all the way down to New York City; I rode it from Albany to New York, on tracks owned by Amtrak and New York’s Metro-North.
Riding Amtrak during the pandemic felt a little safer than being on a plane, if only because I usually had two seats to myself; people just aren’t as packed in as they are on airplanes. Amtrak says its ventilation system exchanges fresh air every 4–5 minutes. A blower expels air from the train and brings in filtered fresh air. While Amtrak doesn’t give details on the filtration level of its filters, the important point is that the dirty air is going out of the train.
The trains are better equipped for the tech-heavy than airplanes are. Every pair of seats has two 120-volt AC ports, and in my experience, they’ve always worked. The cafe cars have large tables you can spread out on. And every train, at least supposedly, has free Wi-Fi on the “Amtrak_WiFi” network, although as you’ll see, that’s not always true.
How Amtrak Wi-Fi Works
Most of Amtrak’s trains use cellular modems which “aggregate available bandwidth from all the major U.S. carriers,” according to Fierce Wireless. Coupled with big antennas on the outside of the train rather than your phone’s tiny antenna inside a big metal tube, the Wi-Fi system gets better, more reliable cellular signal than your phone can.
On the Northeast Corridor, Amtrak has moved to a different system: a private “ trackside network” with fiber-connected access points about 0.8 miles apart. That new network first came to Acela Express trains, and then to Northeast Corridor trains.
Within the trains, each car has two or three 802.11ac or 802.11n routers, primarily from French firm Acksys, which specializes in on-train connectivity. We also scoped out some equipment from TP-Link and Extreme Networks, although neither of those routers seemed to work. Maybe they were being phased out. (Amtrak did not respond to two requests for comment or answer any of my queries about their systems and technology.)
How We Tested Amtrak Wi-Fi
To uncover the secrets of Amtrak’s Wi-Fi, we brought in some heavy-hitting software. Ekahau makes devices designed to measure enterprise Wi-Fi, and the Ekahau Sidekick lets you map the Wi-Fi cloud around you and use comprehensive statistics to improve your network. The new Ookla Wind tests not just network speed, but also dropped calls, video streaming, and even which frequency bands a network uses in a given place. We’ll be using Wind more throughout the year, including in our flagship Fastest Mobile Networks testing stories.
Editors’ Note: Ookla and Ekahau are owned by Ziff Davis, PCMag’s parent company.
Amtrak Wi-Fi vs. Your Phone
Our tests showed that Amtrak’s Wi-Fi is much more reliable than your phone’s cellular connection, on all three trains. (That is, when it works at all—see what happened on our second ride.)
It’s also completely free to use, with no data limits like your cell phone plan’s hotspot mode may have.
Unfortunately, it’s capped—harshly. Our tests showed a 3Mbps upload and download cap on all three of our trains, with really long latencies of 67–83ms, longer than our cell phones showed. You can Zoom, but it’ll be rough. The system also frequently appeared to block YouTube.
Amtrak’s Wi-Fi is less secure than your phone; it’s not secured with a password or encryption. If that concerns you, use a VPN or your personal hotspot.
Whether the train Wi-Fi or your phone’s hotspot is a better bet is all about the tradeoff between speed and reliability. It’s good to have both—to use your phone’s hotspot most of the time, but be able to fall back to the train Wi-Fi when you enter a dead zone.
Having both free Wi-Fi and the ability to tether to your phone really beats what the airlines offer, though.
Train #1: Train 190, New York-Boston
I rolled out of New York around 7 a.m., headed to Boston on one of the nation’s oldest and busiest commuter lines.
Our Ekahau software showed us two or three Acksys-branded access points in each car, with at least one 802.11ac and one 802.11n. There were also a few TP-Link 802.11n routers mixed in, although our device never connected to them. Signal strength didn’t really change depending on where I was in the car—the cars are covered pretty evenly.
It’s not tough to get work done on this train, no matter what network you’re on. If you’re making phone calls, AT&T, Verizon, and T-Mobile were all basically reliable the whole way.
For data, my T-Mobile and Verizon connections dropped from time to time, but would generally recover quickly. You wouldn’t be able to notice any problems with e-mail or Web browsing. T-Mobile had about 10 minutes of dead time around Greenwich, CT, and both T-Mobile and Verizon had some trouble in the Providence train station.
All of the cellular networks were so much massively faster than the Amtrak Wi-Fi system that the occasional dropout was worth it. AT&T and Verizon averaged more than 50Mbps down and 10Mbps up, and T-Mobile was at 143Mbps down and 25Mbps up. Those kinds of speeds let you download entire Netflix shows or buffer videos while Amtrak’s Wi-Fi just crawled along.
A 5G phone definitely helps with T-Mobile coverage on this corridor. All the way up through the New York and Connecticut suburbs, our T-Mobile phone hung onto standalone band 71 5G, the carrier’s lowest-frequency and longest-distance technology. Older phones might have coverage issues if they can’t reach that frequency.
This was a tremendously productive train for me. Using my own T-Mobile connection, I banged out our exclusive on millimeter-wave 5G at New York’s holiday markets and filed it by the time we got into Boston. I probably wouldn’t have been able to do that on a flight, as the flight involves so much more waiting for things: going through security, waiting to board, and going through takeoff-and-landing procedures. I also had two huge seats to myself.
I got into Boston at 11 a.m., relaxed and comfortable.
Train #2: Train 449, Boston-Albany
For my second trip, I took the Lake Shore Limited train from Boston, MA to Albany, NY. The train continues on to Chicago, often traveling through very rural areas on tracks mostly used for freight.
This train is more representative of Amtrak’s national experience: slow and infrequent, but offering folks in places such as Pittsfield, MA a way out without driving. The trip takes three hours by car, but five by train. You’re only on this one if you don’t have other options.
My train only had two passenger cars: an Amfleet II coach car, where I was mostly joined by college students heading home, and a combined cafe/business class car. The Amfleet II coach has even more legroom than the Northeast Corridor coach car, and I was comfortable for the entire ride.
Train 449 was where I ran into the biggest problems—both Wi-Fi and cellular. The passenger car had two 802.11n routers from Extreme Networks, according to our Ekahau survey software. Looking at the in-car spectrum survey, I found their 5GHz band was off, and they were not giving IP addresses to my devices, even on 2.4GHz. In other words, the Wi-Fi didn’t work in the passenger car for the entire six-hour trip.
An equipment conflict may have been the problem. In the combined cafe/business class car we were back to those reliable Acksys 802.11ac routers, delivering their reliable 3Mbps. The conductors said the train-wide router controls were in the cafe car—so if you’re having trouble with Amtrak Wi-Fi, the cafe car is clearly the place to sit.
That was especially frustrating because while Wi-Fi was much slower than cellular, it was much—much!—more reliable. Our T-Mobile data connection was pretty much dead between Springfield and Pittsfield, and Verizon stuttered on and off through a lot of central and western Massachusetts. Verizon’s flickering connection meant you’d want to cache videos, and that you’d probably end up hitting ‘reload’ on Web pages more often than you’d like. AT&T did better.
For voice calls, AT&T was pretty reliable, but T-Mobile and Verizon both dropped 10% of our call attempts.
It became clear here that there are very much two different Amtrak online experiences—a better one on Amtrak and state-owned passenger tracks, and a much worse one out on the freight lines in the country.
Train #3: Train 64, Albany-New York
The Maple Leaf glides through the vast expanse of the State of New York, connecting the City to Niagara Falls, 480 miles away. (It used to go as far as Toronto, but that was terminated with the pandemic.) And it does it all just about as fast as driving, especially on the 53-mile-per-hour, 141-mile run between New York and Albany.
In daylight, this is one of America’s prettiest train journeys. Sit on the right side heading south, and you spend most of your time watching the broad Hudson River.
It was pitch-black out when I boarded train 64, the Maple Leaf, at Albany back to New York City. Like train 190, each car on the Maple Leaf appeared to have two Acksys routers, one broadcasting 802.11ac and one for 802.11n. And like on the other trains, Amtrak’s Wi-Fi system was capped to 3Mbps.
There was an additional little insult, though—on this train, our YouTube test stopped working on the Wi-Fi. I don’t know if it was YouTube or Amtrak’s network configuration, but our YouTube stopped playing on the Wi-Fi while it still played on all three wireless carriers.
Voice calling on all three carriers was almost completely reliable on this route, but data coverage came and went. T-Mobile was by far the fastest network, but AT&T was once again the most reliable.
The run from Albany to New York carries legislators and commuters to and from the cities at both ends and the smaller towns and cities in the middle. With voice calling the most reliable function, it looks like this is a good train to call your spouse from and tell them you’re coming home.
Conclusions: A Useful Freebie, But Bring a Backup
Amtrak’s Wi-Fi is free and more reliable than cellular, when it works, which it doesn’t always. It’s very slow, though, and that slow speed appears to be artificial—a hard 3Mbps cap on the connections.
Riding three kinds of rails reminds me of Amtrak’s strengths and challenges. Where it controls its own tracks, or other passenger railroads do, the trains are reasonably fast and connectivity is solid. But on the broad expanses where Amtrak must borrow tracks from freight railroads, everything about the experience, including the connectivity, can be an utter mess.
Originally published at https://www.pcmag.com.