It took over 15 years to get the RL 01800's to stop butchering the RFID readouts to complete self-parody, so this is a longstanding problem. Although funnily enough the identical system on the Green Line seems to nail it 99+% of the time without hiccups despite trawling way more complicated branch routing, and there's no technical reason whatsoever explaining why that line can be so flawless while the same exact thing on Orange + Red is a complete error-city. I don't ride Blue nearly often enough to get much of a sample size on that line's ASA's accuracy, but it's likewise a tried-and-true RFID ping of the same technical generation as Green's installation. The only complaint I've heard about Blue's ASA is that it's way too talky about surplus-info bus connections such that the announcements sometimes bleed into each other at close-spaced stations...though I think they've toned that excess down considerably with firmware updates to the ASA. Like Green, however, it doesn't accumulate many complaints for inaccuracy like has been noticed so far on Orange and has been a 25-year running farce on Red.
You're absolutely right, though. RFID pings are very hard to screw up, so the fact that it happens so regularly is baffling. It's probably not the trackside tags, because all those do is prompt the announcement and a broken tag would simply leave the ASA trigger absent on a station approach (which would be immediately noticeable for a quick overnight fix, as tags are 'semi-consumables' they stock for quick replacement). It's the onboard computers that are somehow still mangling such a dirt-simple readout.
I apologize if this is taken as a diversion -- its just a bit of clarification of the RFID technology -- having spent quite a bit of time in the past few years mucking with such at various levels from Patent Filings to actually testing tags and readers in real-world situations -- including reading tags inside of moving packages on a conveyor
Here's a plausible scenario for the problem:
First a brief primmer on RFID
You have an active component -- the Reader complete with RF components, data acquisition components and a computer as well as some networking
the reader sends out pulses which both power the "Tag" and inquire as to its ID -- the ID in RFID
You have a purely passive component - -the Tag" which needs energy to operate its internal microelectonics and to in effect send the response back to the reader
To comply with the use of the spectrum [avoiding interference of licensed services] the RF [in the so-called ISM band around 900 MHz] is modulated into a complex sequence of sub pulses at changing RF frequencies
Throw in the fact that the train may be moving and changing its aspect with respect to the tag and there are going to be some outgoing inquiries which don't "catch" a perfectly functioning Tag and additional returns from Tags which get missed by the reader
This is where the computer and the logic controlling everything else gets involved --= if the reader tries and fails to get a response -- typically you try again -- until you either get a response which you can "read" run out of time or decide to abandon the "pinging" of that tag
If the re-reads time-out then the computer code needs to have a back-up plan. Otherwise the computer looks up the name of the station which coincides with the tag number and then plays the audio
So my guess for the most common path to failing to announce the station is that a particular reader on the train [either internally] or more likely in its antenna assembly or cabling is either semi-dead or intermittently reading tags. This would lead to the computer to possibly use either the most recent successfully read tag's information to enter the database or more likely take the next entry in the database -- if there were several failed reads -- you could get quite far behind
An alternative failure would be if a tag is replaced and someone makes an error in "linking the tag's ID" to the proper entry in the database -- then of course you could have both ahead as well as behind announcements
Sorry for the length of the above -- its as some science fiction writer said -- a technology sufficiently advanced seems like magic -- but sometimes the spells don't work as intended
