Cellular Phone Forum / General / GSM / February 2004
Seperate carriers sharing GSM infrastructure?
|
|
Thread rating:  |
Donald Newcomb - 02 Feb 2004 21:53 GMT I want to ask about the options for multiple carriers to share GSM infrastructure. The two extremes seem to be 1) no sharing of any infrastructure 2) a virtual network running on the back of a host network. What's available in-between the extremes? Clearly, carriers can easily share towers, back-haul, emergency generators, etc. But how closely can they actually be intertwined and remain separate? Can carriers share the entire BSC/BTA system (antennas, cables, tower, transmitters) and still transmit separate signals? Are there any examples where competing carriers have developed a modis vivendi for this type of infrastructure sharing, without trying to cut each other's throats on the deal?
Donald Newcomb DRNewcomb (at) attglobal (dot) net
Jer - 03 Feb 2004 13:56 GMT > I want to ask about the options for multiple carriers to share GSM > infrastructure. The two extremes seem to be 1) no sharing of any [quoted text clipped - 9 lines] > Donald Newcomb > DRNewcomb (at) attglobal (dot) net Your #1 would be the winning answer. There is no such thing as a virtual network.
Firstly, nobody shares infrastructure (radios). Nobody. Ever. The antennas you see are exclusively owned and operated by a carrier, and an antenna is connected to a radio - owned by the same carrier. Those radios are connected to only one switching system, also owned and operated by the same carrier. Cellular radios operate on only one frequency - whichever frequency the owner has designated for that site. Cellular carriers don't share any BSC. Ever.
A co-locate tower site is one where the tower is shared, supporting more than one set of antennas. Each set of antennas is connected to a different set of radios, housed in a different hut, and connected to different switching systems.
A wireless call is always handled by a single carrier (on the RF side), the same one that owns all their other RF pieces - whether the caller is home or a roamer. A cellular radio in a cell site is never connected to more than one MSC. If two carriers want to be able to service wireless calls from the same tower site, they'll either have to hammer out a co-locate contract (assuming there's enough real estate at the tower base for more than one hut), or arrange for inter-switch trunking to handle inter-switch traffic (for roamers).
The above applies regardless of the service type, i.e. AMPS, TDMA, CDMA, GSM, IDEN.
 Signature jer email reply - I am not a 'ten' ICQ = 35253273 "All that we do is touched with ocean, yet we remain on the shore of what we know." -- Richard Wilbur
michael turner - 03 Feb 2004 17:37 GMT >> I want to ask about the options for multiple carriers to share GSM >> infrastructure. The two extremes seem to be 1) no sharing of any [quoted text clipped - 12 lines] > Your #1 would be the winning answer. There is no such thing as a > virtual network. Here in the UK we have two virtual networks. Virgin Mobile which is carried on the T-mobile GSM network, and Tesco Mobile carried on the O2 GSM network. Neither Virgin or Tesco have their own infrastructure.
http://www.virginmobile.com/mobile/ http://www.tesco.com/mobile/
 Signature Michael Turner
Gordon Brown - 03 Feb 2004 21:00 GMT > Your #1 would be the winning answer. There is no such thing as a > virtual network. Don't know about in your country, but here in the UK there are several Mobile Virtual Network Operators or "MVNO". Virgin is perhaps the biggest MVNO in the UK, and Virgin use T-Mobile infrastructure (i.e. Airways, BTS, BSC and MSC).
> Firstly, nobody shares infrastructure (radios). Nobody. Ever. The > antennas you see are exclusively owned and operated by a carrier, and an [quoted text clipped - 3 lines] > frequency - whichever frequency the owner has designated for that site. > Cellular carriers don't share any BSC. Ever. Again there are Network Operators sharing antennas in the UK due to planning restrictions and congested areas. There was a programme on the TV (the BBC IIRC) which discussed the clever technology involved in combining the signals from different BTS's for transmissions via a single antenna. Unfortunately I didn't watch the whole programme, so didn't see how the Received signals were split to the various BTS's. It was a very interesting programme for me because they featured a cell site not very far from where I work and commute past every day.
Jer - 04 Feb 2004 02:11 GMT >>Your #1 would be the winning answer. There is no such thing as a >>virtual network. [quoted text clipped - 20 lines] > programme for me because they featured a cell site not very far from where I > work and commute past every day. Okay, yes, my post was predicated on U.S. carriers, and this is the first I've heard of combiners and splitters being used for cellular in any network. Frankly, I don't even want to think about the insertion losses and noise inherent in such a design. But, I suppose some compromises are available for desperate designers.
Like you, the whole premise of this is intereting and I'll start poking around to see what's up. Thanks to you and Michael for the feedback.
 Signature jer email reply - I am not a 'ten' ICQ = 35253273 "All that we do is touched with ocean, yet we remain on the shore of what we know." -- Richard Wilbur
Olivier Boudot - 04 Feb 2004 16:09 GMT > Okay, yes, my post was predicated on U.S. carriers, and this is the > first I've heard of combiners and splitters being used for cellular in > any network. Frankly, I don't even want to think about the insertion > losses and noise inherent in such a design. But, I suppose some > compromises are available for desperate designers. There are at least two more examples of aerial sharing networks in the world I've heard about, aimed at providing indoor coverage : in the Paris Subway (France) and in the Chigago subway in the US, the latter still being under construction IIRC.
I know the Paris network very well. It's based on specific underground network shared by the three French GSM operators. Each operator is responsible for its own BTS, but the RF distribution network (including optical links and repeaters) is operated by the transportation authority, responsible for the whole RF design and antenna implementation. Aerials are made of either omni or panel Kathrein dualband indoor products for stations, corridors and short tunnels coverage, and Jaybeam dualband yagis in longer tunnels. No radiating cables are used (due to supposed money saving issues), the 900/1800 feed being carried through the same coaxial cables (antennas have internal diplexers).
Two of the three operators, Orange and SFR, operate a dual band network (900 & 1800 MHz, dual BCCH) on this infrastructure. The third one (Bouygues) only operates at 1800 MHz for the time being.
The global RF design is, as you can guess, very poor. Insertion losses are of course an issue, but mainly for Bouygues, whose RF outpout power is strictly limited as all three network use a 1800 MHz component. The two others cope better on their 900 MHz layer in term of RXlevel and cell overlapping.
These issue are not as critical as the high level of intermodulation products during heavily loaded hours (when mostry all TRX are "on air"), probably the worst problem they have to cope with.
Moreover, some antenna location choices have been made against the basic rules of RF design and planning. It's hard being a transportation company but still thinking one can easely turn into a RF specialist ;-)...
 Signature A+,
Olivier
Jer - 05 Feb 2004 12:51 GMT >>Okay, yes, my post was predicated on U.S. carriers, and this is the >>first I've heard of combiners and splitters being used for cellular in [quoted text clipped - 35 lines] > rules of RF design and planning. It's hard being a transportation company > but still thinking one can easely turn into a RF specialist ;-)... Holy Cow! What a mess!! :))) I do find all this interesting, even a little amuzing. Terrain issues will always offer challenges to the RF system engineer and the above is clearly an indication of where technology can lead the pack. And your comment on the intermod is well noted. And I can't even imagine what that must look like on a spectrum analyzer, but it'd be a giggle to take a peek. :)
 Signature jer email reply - I am not a 'ten' ICQ = 35253273 "All that we do is touched with ocean, yet we remain on the shore of what we know." -- Richard Wilbur
Olivier Boudot - 06 Feb 2004 00:49 GMT > Holy Cow! What a mess!! :))) Simply imagine all the people trying to get on with their calls on moving trains with a HO failure rate of nearly 50% during rush hours, and regularly moaning on their cell phones ;-)...
> I do find all this interesting, even a little amuzing. I'd better give you a practical exemple : Chatelet les Halles commuter train station, platform level (7 tracks, 3 lines, 2 million passengers/day). Two dualband indoor panels at the West end of the station, two others at the East end. Platform length : 250 meters (
Orange : 48% market share (BSS = Motorola M-6 Cell, EFR only). Two dedicated cells targeted at tracks coverage. 900 MHz cell : 8 TRX 1800 MHz cell : 7 TRX (might be eight now, haven't checked for a while). Both use East and West end aerials.
SFR : 35 % market share (BSS = Nokia Ultrasite, EFR/HR, AMR expected during forthcoming spring). Two dedicated cells targeted at tracks coverage. 900 MHz cell : 5 TRX 1800 MHz cell : 5 TRX Both use East and West end aerials.
Bouygues Telecom : 17% market share (BSS = Nortel S8000, EFR only), 1800 MHz only, split in two cells (still talking about the platform level only). 7 TRX (East end aerials cell), and 5 TRX (West end aerials cell).
These TRX (one per cell support a BCCH, of course) are connected to FOUR (two as far as Bouygues is concerned) Kathrein 738 573 indoor dualband panels. Aerial gain : 4.85 dBd, max. overall imput power per panel : 25 W. LOL.
No information about VSWR nor intermod rating in dBc on the manufacturer's catalog, of course... LOL once again.
Apart from that, all three network are regularly congested in this station between 4 and 7 PM. You're not even sure to get a SDCCH in order to send or recieve a SMS. But how could the operators even think about installing extension cabinets as the antenna systems already seem close to overload ?
The operators are clearly loosing money (but not the transportation company of course, as the annual fee it receives from the three networks is known to be quite high, even though no precise figure has ever been publicly disclosed).
> And I can't even imagine what that must look like on a > spectrum analyzer, but it'd be a giggle to take a peek. :)
:-))))... AFAIK, the only obligation of the transportation company is to provide > - 90 dBm signals in each band everywhere, including in moving trains. I regularly measure RXlevel in the whole underground system, and - 105 dBm figures are quite frequent. Thus, SFR has recently moved is RXlev access min from - 101 to - 106 dBm on its 900 MHz underground layer ;-)...
If they want to do a serious job, they'll have to overlay their existing system with a radiating cable based network IMO. Allgon, for instance, has developped very good engineering solutions. And they're not the only ones. Of course the best way to deal with all the above mentionned problems would be to let each operator build its own network, but this can't reasonnably be achieved, as the network company already refused such a scheme.
They could keep their existing network on a 1800 MHz only basis in order to cope with traffic issues especially in main station platforms, subway entrances and lobbies, and use a 900 MHz only system based on radiating cables in tunnels and long corridors. This is the only way to achieve good handoff figures in moving trains (up to 60 MPH) as cell overlapping is then easyer to implement in a cost-effective way with a reasonable QOS (the existing system is based on optical repeaters feeding multiple antenna systems in long tunnels, so I doubt it is globally cheaper). Each system being independant from the other, the max number of TRX could be higher for a given QOS, load sharing between bands being achieved by fine tuning the C2 criteria, as it already works on existing dualband networks.
Fortunately, Orange, SFR and Bouygues all operate in the underground network on a single LAC, so signalling issues should take place where the old and the new network would still operate simultaneously.
 Signature Olivier
Jer - 06 Feb 2004 14:08 GMT >>Holy Cow! What a mess!! :))) > > Simply imagine all the people trying to get on with their calls on moving > trains with a HO failure rate of nearly 50% during rush hours, and > regularly moaning on their cell phones ;-)... 50%? Get a rope! That's a hanging offence on the west side!
>> I do find all this interesting, even a little amuzing. > [quoted text clipped - 19 lines] > MHz only, split in two cells (still talking about the platform level > only). 7 TRX (East end aerials cell), and 5 TRX (West end aerials cell). Not an Ericsson among it. :) And no AMPS.
> These TRX (one per cell support a BCCH, of course) are connected to FOUR > (two as far as Bouygues is concerned) Kathrein 738 573 indoor dualband > panels. > Aerial gain : 4.85 dBd, max. overall imput power per panel : 25 W. LOL. 25W!!?? Underground? Do the panels have a blue corona? (oh wait, maybe you shouldn't be looking directly at them)
> No information about VSWR nor intermod rating in dBc on the > manufacturer's catalog, of course... LOL once again. [quoted text clipped - 27 lines] > can't reasonnably be achieved, as the network company already refused > such a scheme. Does radiating cable = leaky coax? (not to be confused with 'leeks')
> They could keep their existing network on a 1800 MHz only basis in order > to cope with traffic issues especially in main station platforms, subway [quoted text clipped - 12 lines] > network on a single LAC, so signalling issues should take place where the > old and the new network would still operate simultaneously. Olivier, I'm zonked with all this.
 Signature jer email reply - I am not a 'ten' ICQ = 35253273 "All that we do is touched with ocean, yet we remain on the shore of what we know." -- Richard Wilbur
Olivier Boudot - 06 Feb 2004 14:58 GMT > 50%? Get a rope! That's a hanging offence on the west side! It's an offence to me, too ;-) ! But during off peak hours, it works (close to 100% where both cell overlapping and HO margin are OK - but it's not the case everywhere, especially in some stations where aerial implementation was decided by some guys I'd willingly hang with your rope ;-).
> Not an Ericsson among it. :) And no AMPS. No E/// on the BSS side, but both Orange and Bouygues have E/// switches (MSCs). An I personally use a R520m TEMS ;-). No AMPS in this part of the country (there is an AMPS network in some French Caribean islands due to the high number of North American roamers there).
> 25W!!?? Underground? As in some stations some aerials get connected to more than 30 different TRXs, I believe the overall field can be very high, especially in the main beam during heavily loaded hours. The (funniest ?) thing is you could nearly touch some of them, as they are wall mounted at 7 feet only...
> Do the panels have a blue corona? No ;-).
> (oh wait, maybe you shouldn't be looking directly at them) Sometimes, when you're stuck among others passengers on a crowded platform waiting for your train to come, you have no other choice than standing by them ;-).
Here many people feel very concerned about health issues regarding cell phone macrocellular antennas, even where they live more than 150 m (500 ft ?) from them, and of course not in the direction of the main beam which is quite narrow on the elevation side, but they don't know anything about indoor antennas, they don't even know what these kind of aerials look like, and of course they don't feel sick in the subway, even though this is definitely THE place they get exposed to the strongest RF fields (apart from those living close to powerful VHF/UHF stations such as the Eiffel Tower)...
> Does radiating cable = leaky coax? Yes.
(not to be confused with 'leeks')
;-)
> Olivier, I'm zonked with all this. I'm zonked with the RF design of the transportation company's network, hope my brain will not finally turn into a leek ;-) !
 Signature Olivier
John S. - 05 Feb 2004 18:26 GMT >> Firstly, nobody shares infrastructure (radios). Nobody. Ever. You might want to re-think your statement. We here all know that Cingular and T-Mobile share their infastucture in California/Nevada and the New York markets.
Same antenna, tower, radio, coax, switch etc.....
Keep in mind that switches can be partitioned so that billing is split but that is about all that isn't shared!
-- John S. e-mail responses to - john at kiana dot net
Jer - 06 Feb 2004 04:44 GMT >>>Firstly, nobody shares infrastructure (radios). Nobody. Ever. > [quoted text clipped - 3 lines] > > Same antenna, tower, radio, coax, switch etc..... Let me clarify...
Sharing antennas/coax is apparently done in a rare few systems - none in the U.S. that I'm aware of (yet). Sharing radios does not happen because it cannot happen. A radio can only be connected and controlled from one switch. For a BSC radio to be connected to more than one switch, the control channel would have to have a data splitter and it would have to lock out any other control channel that wasn't active. If a control channel goes inactive, the BSC goes offline and the switch generates an alarm for that BSC. Restoring a control channel in most systems requires a moment or two for each data link (there are always two), first one, then the other. The idea of the control channel going in and out of service due to multiple switches attempting to control the same radio is simply ludicrous. A flapping BSC will not be allowed to return to service due to hold-over timing, thus preventing voice radios from being opened to traffic just in time for another flap to occur, removing the BSC from service again, and clearing all voice channels again. Did I get that right Lee? I'm sorry, I didn't introduce Lee... Lee is a long time RF systems engineer for Cingular visiting me tonight. Lee's a good egg, and I'll (he'll) repeat, BSc radios are never shared - they are always connected to only one switch, and that one switch always controls all calls on that BSC radio. All radios in a BSC are located in the same cabinet and there's only one pair of control channels for that cabinet. No exceptions.
Yes, multiple radios can be connected to a single RF radiator (antenna) using combiners, circulators and splitters. Virtually all panel antennas at the top of Cingular's AMPS/TDMA/CDMA/GSM sites are actually multiple radiators hiding under a single plastic cover, and each radiator is connected to it's radio cabinet with a separate pair of coax cables - a pair being one transmit and one receive. A single radio cabinet can contain radios for multiple codecs used in the same frequency band, which is commonly done in 'overlay' systems. 800Mhz systems and 1900Mhz systems will have separate cabinets for each, and each cabinet will be connected to it's own antennas via it's own coax. This can cause a tower to appear to be a co-locate due to more than one set of panel antennas, but if both sets of antennas are conected to the same carrier, it's not a co-locate - it is simply two separate service bands offered by the same carrier at that cell site.
Now, to the sharing of infrastructure... the term "co-locate" means a tower is shared with multiple carriers, but each carrier still has their own antennas on that tower, and each antenna set is connected via coax to indivisdual cell sites (BSCs), and each BSC is housed in a separate enclosure at the tower base. This is often considered 'sharing infrastucture' since each carrier's radios/antennas are sharing the same tower - but each carrier's radios are still housed in a separate enclosures, and equipment in each enclosure (BSC) is only connected to that carrier's switch. So apparently, 'sharing infrastructure' in most cases means only the tower is shared, but each carrier still has all it's own equipment securely enclosed and not shared with any other carrier.
> Keep in mind that switches can be partitioned so that billing is split but that > is about all that isn't shared! If a BSC radio is connected to a switch, than that switch is the only control point for that radio, hence it is not shared. Earlier in this thread, it was noted that some systems (radios) are sharing antennas - i.e. one antenna connected to more than one radio, but each radio is still connected to only one switch, hence only one control point for that radio.
Switches are not partioned, but billing records have been split ever since the first roamer made the first call. A single BSC, connected to it's switch, can easily handle calls for home users as well as roamers - each getting their bills separately - but it's still only one BSC, one switch, and one carrier maintaining the whole enchilada.
I know this is verbose, but...
 Signature jer email reply - I am not a 'ten' ICQ = 35253273 "All that we do is touched with ocean, yet we remain on the shore of what we know." -- Richard Wilbur
Donald Newcomb - 11 Feb 2004 02:53 GMT > Let me clarify... > [quoted text clipped - 4 lines] > switch, the control channel would have to have a data splitter and it > would have to lock out any other control channel that wasn't active. I have to say, this is one of the more interesting discussion's I've read in a while. Very informative. So, as far as the US is concerned, infrastructure sharing, thus far, has involved only sharing towers? Except when you are talking about sharing infrastructure for a multi-technology overlay on a carrier's license (e.g. TDMA + CDMA), in which case, the most that has been shared is the box that covers the antennas.
What happens when a carrier needs to use two GSM ARFCs on a single sector? Does a single radio support multiple ARFCs, or does there need to be two sets of antennas? If a single radio/coax/antenna set can support multiple ARFCs, what would keep each ARFC from being controlled by a different switch and having its own control channel? The picture I have in my head is a single rack with various plug-in modules (not unlike a VXI chasis), one BSS control module, one transmit module and one receive module for each ARFC and one RF transmit amp module. As more ARFCs are needed, more modules are added. Each transmit module is mixed into the single RF transmit amp. Different networks could share everything but the transmit and receive modules. Even the backhaul could be multiplexed and shared.
I know this is contrary to the way things are done in a competitive business environment, but assume the case of the subway system, airport or tower on Disney World, where cooperation is forced by the landlord/government. I'm just curious how far the sharing could go. Oliver's discussion of the Paris subway has been very enlightening. Thanks!
 Signature Donald R. Newcomb DRNewcomb (at) attglobal (dot) net
Jer - 11 Feb 2004 03:49 GMT >>Let me clarify... >> [quoted text clipped - 11 lines] > carrier's license (e.g. TDMA + CDMA), in which case, the most that has been > shared is the box that covers the antennas. To the extent of my knowledge, yes, the above is an accurate characterisation. As far as Cingular (in the U.S.) is concerned, I'm unaware of any antenna structure that is shared with another carrier, although the antennas used within their overlay system are dual dipole arrays, each with independent coaxial feeds. To the unaware eye, it appears sharing is in play, until one learns that both feeds originate from the same (or only) enclosure at the tower base, and connect to separate radio frames internally.
> What happens when a carrier needs to use two GSM ARFCs on a single sector? > Does a single radio support multiple ARFCs, or does there need to be two [quoted text clipped - 7 lines] > Different networks could share everything but the transmit and receive > modules. Even the backhaul could be multiplexed and shared. All good questions! and I don't know if Cingular uses multiple absolutes in a sector or not, nor where if they do. I know Cingular doesn't share the bottom end of their coax, but I am gonna ask my bud for clarification first change I get.
> I know this is contrary to the way things are done in a competitive business > environment, but assume the case of the subway system, airport or tower on > Disney World, where cooperation is forced by the landlord/government. I'm > just curious how far the sharing could go. Oliver's discussion of the Paris > subway has been very enlightening. Thanks! Yeah, it's keeping me up nights - almost. :)) It would appear the term 'competitive' is giving the benefit of doubt, whereas the term 'cut throat' may offer improved focus on the nature of beast, like Oliver's subway seems.
 Signature jer email reply - I am not a 'ten' ICQ = 35253273 "All that we do is touched with ocean, yet we remain on the shore of what we know." -- Richard Wilbur
|
|
|