Better Amtrak WiFi: Maybe Not So Crazy

Amtrak wants to fix its weak WiFi in its Northeast Corridor. The train company's idea isn't as improbable as it may seem.

June 12, 2014

3 Min Read
Network Computing logo

Amtrak raised some eyebrows with its announcement this week that the cash-strapped train company is investigating the possibility of bulking up its anemic WiFi services for passengers in its Northeast Corridor. On the surface, getting each train to respectable network speeds sounds pretty far-fetched, but Amtrak has a few tricks up its sleeve that make the effort promising.

Where a lot of riders and pundits slam Amtrak’s current wireless services (about 10 Mbps per train with lots of protocol restrictions), I give the company credit for even trying. If you are not familiar with actual rail routes, you may not realize that they frequently wander well off the beaten path and away from mobile network coverage. With 457 miles to cover in its Northeast Corridor (NEC), Amtrak is doing well to provide even basic wireless service for the millions of riders a year who use the mobile network-sourced WiFi on trains that carry as many as 500 passengers each.  

In simplest terms, the rails frequently stray from cell tower coverage and so WiFi drops. But when it does work, it’s free and better than nothing. Yet it's hardly competition for the WiFi provided on buses and commercial aircraft.

Amtrak has never operated in the black, and each year receives billion-dollar government subsidies that keep it afloat. To reach the stated goal of delivering a highly reliable 25 Mbps minimum throughput for each train that travels the NEC, we’re looking at some serious infrastructure of one type or another. No options will be inexpensive, but skeptics probably don’t know about key assets already in place that will help get a 10-mile proof of concept off the ground in Maryland, at the southern end of the NEC. Those assets could eventually make a hundreds of miles long wireless network not seem like such a crazy idea.

As I read early takes on the Trackside Wi-Fi initiative, it’s pretty clear that most of those ripping on the notion haven’t dug into the publicly available project documents, as they assume any improvements would continue to come from mobile carriers. In reality, Amtrak plans on abandoning the mobile network model and leveraging fiber, power, and structures it already has in place in its rights of way to provide dedicated, high-speed backbone connectivity and mounting points for new wireless cells.

Given that typical WiFi gear doesn’t do so well with high-speed hand-offs, Amtrak would likely use carrier-grade kit along the lines of Proxim Wireless and Strix Systems to allow on-board train routers to not drop session as the train traverses through cells at up to 150 MPH. In the cars themselves, passengers would be oblivious to the trackside backhaul and would connect to one of the WiFi APs on board (typically two per car).  

From the perspective of one who designs WiFi networks and point-to-point links, this is both (mostly) realistic and pretty exciting from the technical perspective. At the same time, high costs are unavoidable when you consider that along with financial, technical, and logistical challenges, the NEC has 17 tunnels, including one almost three miles long.

Still, assuming that Amtrak can get the network built,  things get interesting on many fronts. Amtrak has a number of operational aspects that would benefit from better networking, including IP-based video security, e-commerce and ticketing, inventory and security of rolling stock, and monitoring of IP-based components.

And if the new trackside network is successful, it stands to reason that Amtrak stations would get better WiFi. And what works for the NEC would likely serve as a model for the rest of the Amtrak areas in the US and Canada, and potentially for railroads in other parts of the world.

I’ll admit to being hopeful about the initiative for Amtrak’s sake. Of course, there’s a long way to go before any success can be realized. The proof-of-concept testing in Delaware is likely to start in early 2015 and run for a year, with an NEC build out taking several years after that if early testing proves viable from the technical and business perspectives.

Stay informed! Sign up to get expert advice and insight delivered direct to your inbox

You May Also Like

More Insights