Initially defined in the core band of 2100MHz, UMTS is now blossoming in the low bands–most notably in Europe's GSM 900MHz band. RFS explores the opportunities and the RF challenges.
The past five years have seen extensive deployment of third-generation (3G) ‘universal mobile telecommunications system' (UMTS) wireless networks the world over. Invariably, this has been in UMTS's ‘native home'–the so-called high bands. To-date these deployments have largely been in two of the nine bands defined by the Third Generation Project Partnership (3GPP) for UMTS: the 2100MHz band in much of the world, and the 1900 MHz ‘Personal Communications Services' (PCS) band in North America. Most recently, UMTS deployments have commenced in North America in the 2100/1700MHz Advanced Wireless Services (AWS) band.
But the times are changing for UMTS. There is a distinct push to see UMTS rolled out in the low bands defined by 3GPP for UMTS (900MHz and, to a lesser extent, 850MHz). Most particularly this push is coming from Europe, where there is strong interest in its deployment at 900MHz, the traditional band for 2G ‘global system for mobile communication' (GSM) networks. Since late 2006, successful European trials of UMTS900 have been completed, including those of Finnish telco Elisa, France's SFR and Orange, and O2 subsidiary, Manx Telecom. Meanwhile, European spectrum regulators–including the committee responsible for pan-European spectrum harmonisation, the Electronic Communications Committee (ECC), France's ARCEP and Switzerland's ComCom–are moving apace to determine clear guidelines for the reuse (or ‘refarming') of the GSM900 spectrum for UMTS.
"It is clear that 2007 will be an important year for UMTS900 across Europe, most notably in Finland, France and Spain," says Rémi Deniel, Wireless Infrastructure Solutions (WIS) Area Product Manager with RF technology group, Radio Frequency Systems (RFS). "We are actively working with operators and OEMs to pursue the appropriate RF solutions to streamline this." Deniel cites two key short-term drivers behind the push for UMTS in the low band: the improved indoor penetration (typically 30%) that can be achieved at 900MHz compared with 2100MHz; and the greater coverage areas that can be achieved at 900MHz. The latter promises a reduction in total UMTS cell count of around 60 per cent, which presents great advantages in rural deployments.
Driving down opex
The long-term benefit of making the ‘2100-to-900' UMTS shift is, according to RFS's Global Product Manager WIS, David Kiesling, even more significant. "In most parts of the world, 3G data revenue is ‘propping up' falling voice revenues, so there is increasing pressure to cut opex," he says. "The refarming of GSM900 spectrum to UMTS is in direct response to this need. This is because the marginal cost of carrying one minute of traffic on UMTS is less than the equivalent traffic on GSM. UMTS900 is an economic practicality." In the absolute long term, this refarming will support the eventual shutdown of the GSM networks in favour of UMTS.
In broad principle, European 900MHz refarming is straightforward, as it takes advantage of the facts that many operators already own 900MHz spectrum, and have well-established GSM900 base station sites. It is simply a matter of co-location. Yet, in reality, the 900MHz refarming will present operators with something of a ‘balancing act' challenge. Europe's existing GSM900 networks are core revenue earners. "Operators are advising us that the main goal is to realise UMTS900, but to retain–as far as possible–exactly the same GSM900 performance," Deniel says. "They don't want to impact the network [GSM900] that is such an important revenue earner. Much of the real deployment and RF challenges stem from this challenge."
Making UMTS900 happen while ensuring minimal impact on the GSM900 network will present challenges on three key fronts: spectrum management of the 900MHz band; minimising base station site impact and acquisition/rental costs (where required); and overall network management of both the established GSM900 and the new UMTS900 networks.
Spectrum management
Interleaving the paired 5MHz channels of UMTS–a wideband technology–within the narrowband 200kHz channels of GSM presents interesting spectrum and interference management challenges. In short, it will require a judicious balance of guard band allocation, spectrum allocation and co-location filtering. The ECC has addressed much of this in its ECC Report 82 (May 2006), nominating suggested guard band allocations and spectral arrangements, such as the so-called ‘sandwich' configuration, to minimise GSM900/UMTS900 co-location interference.
RFS Chief Technical Officer for WIS, André Doll, has doubts about the real-world practicality of such strategies. "The UMTS channel requires 5MHz, so in many cases the ‘sandwich' approach won't be practical. More likely it [the UMTS900 allocation] will be on the side, just for practical spectral reasons," he says. Achieving adequate guard banding, most particularly in urban areas, where ‘coordinated' (or one-for-one) GSM900/UMTS900 site co-location is unlikely, will be a challenge. In such urban situations, the lower power GSM900 pico and micro cells will be most susceptible to co-location interference.
To this end, RFS is currently working with OEMs to develop appropriate filter solutions, specifically addressing the unique waveform of the UMTS operator. "In UMTS, multi-carrier power amplifiers (MCPAs) are used, whereas in GSM, single carrier power amplifiers (PAs) are used. MCPAs typically produce intermodulation products, or what we call ‘shoulders'," Doll says. "These ‘shoulders' are parasitic. We are currently developing filters that deal with such shoulders in GSM900/UMTS900 co-location scenarios. Such filtering is effectively impossible in a ‘sandwich' configuration. Even with the UMTS900 allocation on the side, the filter specifications are challenging–15 to 25dB of additional rejection is typically required on the sides."
Footprints and side-lobes
While there is interest in sharing site RF infrastructure between the two 900MHz services, the use of a shared antenna is almost universally rejected. "The ‘breathing cell' nature of the UMTS network contrasts sharply with the static cell nature of the GSM network," Kiesling explains. "There will be a great need for dynamic cell footprint and antenna tilt management in UMTS900, as we've seen in other CDMA technologies.
"With a GSM network, the footprint is static. Once you get it working, it doesn't change. As a result, everyone we're speaking to regarding UMTS900 overlays is requesting an antenna that is separate in terms of RF performance."
While the antenna should be ‘separate', there is a crucial need–from a site rental cost and planning permit perspective–to minimise the total number of antenna housings that are present on any site. To this end, RFS is developing a range of unique ‘side-by-side' quad-pole 900MHz solutions, plus dual-band UMTS2100/UMTS900 antennas, all in compact form-factors. Minimising the antenna's visual impact is crucial here.
The nature of the antenna beam, most notably the side and rear lobes, also becomes a critical element in the UMTS network. "CDMA technologies are susceptible to ‘pilot pollution'–a phenomenon unknown in GSM," Kiesling says. "'Pilot pollution' is exacerbated by side- and rear-lobe interference, especially in built-up urban areas. It degrades the frame error rate (FER) and increases dropped calls. As a result, side- and rear-lobe control is essential." RFS's premium performance base station antenna range, the Optimizer family, addresses this directly by providing upper side-lobe suppression better than 20dB across the entire tilt range, plus premium front-to-back ratios. Importantly, all lobes–front, rear and side–are tilted together during electrical tilt activities. This avoids unpredictable side- and rear-lobe interference problems.
To address UMTS's ‘breathing cell' network optimisation requirements, dynamic antenna tilt control will be an essential. An RF control and monitoring solution set makes it possible for operators to accurately control and monitor tower top components, either local to the tower, or remotely at the operations and maintenance centre (OMC). Importantly, it must be compliant with the latest revision (version 2.0) of the open communications standard of the Antenna Interface Standards Group (AISG). "The AISG v2.0 is directly linked to the 3GPP standard, making such control and monitoring solutions UMTS900 essentials," says Kiesling.
Feeder sharing
To what extent feeder sharing might occur in GSM900/UMTS900 co-location comes down to the operator's particular goals and strategies. Rémi Deniel cites one major pan-European operator that plans to run separate feeders on most sites. "This particular operator has found that on many sites it can be quite difficult to place a new UMTS node B close to the GSM base transmitter station (BTS)," he says. "The cabling, or jumper, distance between the two might be as high as five meters. The operator believes that the dB losses in such jumpers, combined with that of a diplexer, would cut too deeply into the coverage advantages offered by UMTS at 900MHz."
Doll contrasts this with many other European operators, who are driven hard by opex reduction goals, and seek the widest range of RF infrastructure sharing solutions, including feeder cable sharing. "Reduced RF component count reduces site tower loading, and in turn, cuts long-term site rental costs," he says. "Feeder sharing will be part of this, so advanced diplexer solutions will be essential. From a capex perspective, the ‘rule-of-thumb' break-even between ‘separate feeder' solutions versus ‘diplexer-based' solutions is typically at feeder lengths of 20 to 25 metres."
Addressing the loss incurred in the diplexer presents an important opex challenge. Using a conventional 3dB diplexer will reduce the effective power output at the antenna, thus diminishing both the GSM900 and UMTS900 RF footprints. "You can compensate for this by increasing the power output of the PAs, but this means you have to increase support infrastructure such as air conditioning. This, in turn, impacts your opex," Doll says.
"Much of how this is addressed diplexer-wise will depend on the nature of the guard bands applied. Operators, OEMs and antenna manufacturers are working together to develop a new range of solutions addressing this, including low-loss diplexers. A further area of interest, and perhaps the most advanced solution, is the remote radio head (RRH). This simply allows the active components to be located at the tower top, with node B to antenna connection achieved via a simple fibre optic link."
Looking forward
Kiesling believes that–in the broadest sense–the low band is the true global home for UMTS. "This transition of UMTS to low band is definitely not limited to Europe. We are talking to operators across South America who are exploring it. We are working with operators in Australia who are trialing and even deploying it. In addition, the 3GPP is currently considering a 450 to 650MHz band UMTS standard for sparsely populated areas. UMTS low-band deployments will be a global phenomenon."