Paris Metro GSM Coverage Revisited

The Paris metro remains an interesting research location for me as especially during rush hour you can easily experience what congestion does to a network. Even with Opera Mini it's difficult to use the network at such times as you can almost see individual bits trickling in. Anyway, so I was wondering for a long time just how many cells they have in the metro to get an idea of the effort invested and the capacity available.

Recently I had the time and opportunity to dig a bit deeper. One line number 12 I've been looking, each metro station is covered by its own cell. By extending the coverage into the tunnels, handovers for voice calls and uninterrupted web browsing works most of the time. In other words, once cell has to absorb the load generated by all people currently waiting for a metro plus the traffic generated by the people in the metro trains currently in the stations or the tunnels.

That's an important first indication of the capacity deployed but a second parameter to come to a conclusion is more difficult to obtain: The number of carriers deployed. Unfortunately, that can't be seen with publicly available methods such as using AT commands that give you the current cell-id whenever it changes. So that part remains a mystery for the moment.

But capacity seems to be sufficient for voice calls, as all call attempts I made on the trip resulted in a connection. So no blocking here. The GPRS / EDGE side of the network is another story though. While speeds are good during non rush-hour times, the PS side is totally overloaded when people go and come from work. So either there are not enough timeslots available or some of them are taken for voice calls. Again, difficult to tell.

But one thing is good to know: With one base station per metro station, it is much easier for network operators to increase capacity if they wanted to compared to a situation in which a single cell was used for much longer parts of the underground network. But please don't bother adding a GSM TRX or two, just add 3G to the mix as other cities have already done to fix the issue for good.

One thought on “Paris Metro GSM Coverage Revisited”

  1. If it was SFR network you were camping on, you were connected to Nokia GSM equipment. In Nokia, it works like this: you can define a dedicated and a default GPRS territory. The default GPRS territory includes the dedicated GPRS territory and both of them have a size defined in number of Timeslots. For example, we can set 1 TSL for the dedicated GPRS territory and 3 TSL for the default GPRS territory. With this configuration the total amount of pre-defined GPRS resources in the cell is 3 TSL whose 1 TSL is dedicated. The 2 of the 3 TSLs of the default territory that are not dedicated can be pre-empted by voice at anytime if all TSL outside the GPRS default territory are busy. Indeed, voice has priority over GPRS in the default territory excluding the dedicated. The dedicated territory is as stated by its name dedicated exclusively to GPRS. In the case of low voice traffic, the PCU can configure an additional territory to GPRS beyond the default territory if the PCU capacity and voice traffic allows it and if required by the GPRS traffic. In general, the operators use to dedicate the minimum amount of TSL to GPRS (e.g. 1 TSL), that is why in busy hour period you experience very bad GPRS quality. GSM is still optimized for voice by the operators!

Comments are closed.