16 thoughts on “Much Talk These Days About “the Internet of Things””

  1. Dearie…not an expert on RR signaling technology, possibly Cold Spring Shops could help. But Automatic Train Control / Automatic Train Stop systems are by no means new, either in the US or other countries.


    Generally I believe these systems communicate with the train via pulse code, using close-range magnetic coupling rather than radio, and train location is determined by which track block it is within. The difference with the current PTC project, I believe, is the use of radio to communicate with the train and GPS (plus wheel rotation count) to determine location more precisely than track block occupancy can. One of the main problems with the current implementation is the failure of government to release adequate radio frequencies, couple with “historical preservation” issues slowing down the placement of required trackside antennae. These would be problems even if the technology has been purchased from an existing operator.

    I’ll ask Stephen at CSS if he would be kind enough to stop by and share some enlightenment.

  2. Article about the signaling and control for the French TGV system here:


    …seems to use very-low-frequency radio over most stretches of track, with somewhat higher frequency at fixed locations. Train location identification is by block occupancy.

    Need to remember that the US system is required to work with freight trains as well as passenger, and will operate over considerably greater distances of track than the typical high-speed passenger system.

  3. Thanks for your interest.

    David Foster’s comments identify the two greatest challenges to implementing positive train control on North American railroads, namely keeping track of the precise location of each train on lines used by a wider variety of trains than those you’d encounter on the fast lines of Europe or Asia.

    The general principle of train protection is that two trains shall not simultaneously occupy the same track. Usually, that’s done by dividing the rail line into sections, which in railroading we call blocks. Then, skipping a lot of details, lineside signals or readouts (cab signals) in the locomotive advise the engineer that the next block is occupied or clear. If the current block is occupied you already have a problem!

    But the blocks are fixed in space. The point of positive train control is to alert engineers to the presence of other trains in a more flexible way, which can permit faster operation or more trains occupying a given stretch of track. It also allows for application of the train’s brakes if the engineer neglects the prime directive, which is to control the speed of the train. It’s that second function that came into play after the Frankford Junction derailment. But keeping track of the trains in continuous time and continuous space is much more challenging than keeping track of trains passing discrete points at or after specific times.

    The fast trains of Germany and England have a form of train control, but it’s based on the block system with sensors at discrete points. And the trains are less heterogeneous, for instance the standard German service braking application required a dead stop from 160 km/h in 1 km (100 mph to 0 in 3300′, doable for electric multiple unit cars.) Try doing that with a 125 car North American coal train, or a container train. Now write algorithms to keep track of the head end and hind end of those trains.

    Even that barely scratches the surface. Perhaps I’ll have something up at my site tomorrow or Friday, depending on how hung over the presidential debate leaves me.

  4. I can’t speak much on the engineering but what we have here is mostly politics because the trouble is set up by the artificial deadline created by government mandate and the fines for noncompliance that kick in with the new year. Was the 2008 mandate to get this done by year’s end 2015 reasonable or unreasonable? What went wrong?

    The brinksmanship of writing such a letter at the present time is a sign of a broken governance system. For any big project like this in a highly regulated arena, it would be normal in the 21st century to publish the Gantt chart in a way that would have let us all know that there was going to be trouble hitting the deadline years ago and it would identify exactly who the guilty parties are. Was this a case of railroad companies spending too little to get the job done? You can’t tell without knowing the scope of what had to be done and the resources they threw at it. Did the government via the FCC, sabotage the project by releasing the needed frequencies to railroad use? You would only know that by knowing when they asked for those frequencies, when they got released, and whether that time frame the FCC took is unusual.

    All of the needed citizen oversight information exists and could be shared with the public at little cost. We just don’t do it and so we’re reduced to climbing on our ideological hobby horses, guessing at the underlying facts, and doing another round of useless kabuki combat that resolves nothing, improves nothing, and changes nothing. What a waste of time.

    Railroaders deeply care about trains. There are train obsessed people out there that could have warned of what was coming and clued the rest of us in years ago. That would be 21st century governance. We don’t have it yet and so we’re going to be in trouble for a bit.

  5. Another example of how clueless our political class is. It would have been better to leave the RR companies alone to implement this if there is an economic incentive to do so. Forcing technology is very inefficient and always leads to wasteful situations like this one.

    Another example – the Senate Launch System for NASA.

  6. Just put up a mini-dissertation on positive train control at my site. There’s a Heisenberg problem: the system must know the location, direction, and mass of each train in the network. Nontrivial.

  7. “For any big project like this in a highly regulated arena,”

    Have you seen Megan McArdle’s column on Healthcare .gov ?

    Same issue. I remember when Dave Durenberger told a bunch of us that he was the only guy in the Senate who knew anything about health care. Nobody else knew anything. That was 30 years ago. They are all lawyers. There are no engineers and the few physicians are all Republicans. They can’t do anything right on any technical issue.

  8. That last comment was me. I guess posting from another ISP stripped the ID. We are in London now, just getting up. I took some photos of “migrant” camps yesterday. I haven’t checked to see if they came out yet. We took the ferry back to UK from Calais.

  9. I rode Maryland Commuter Rail for years. There was a time when a new control system was installed which was run for a while from Jacksonville, Florida. Every time there was a thunderstorm in Jacksonville, not rare in Florida in the Summer, the system went down. They could no longer use the system of stationmasters reporting the locations of trains because they got rid of most of those. Therefore, the whole system was immobilized until they could locate somebody to authorize the trains running whereupon they ran at 20 MPH the whole way (I live in West Virginia, 60 miles from DC. Unforeseen consequences and all that.

  10. Yes, replace timetable and train order operation or centralized traffic control managed from Cumberland, Maryland (Baltimore and Ohio) with centralized traffic control managed from Jacksonville (everything on Baltimore and Ohio and several other railroads) without sufficiently hardened communication systems, and without backup (which doesn’t have to be the legions of station agents and tower operators) and the railroad will get fouled.

    That old Baltimore and Ohio line also figures in the development of positive train control, as it was on that line that the engineer of an eastbound suburban train forgot a signal indication leaving a stop, winding up in a fatal collision with a westbound Amtrak train for Chicago.

Comments are closed.