Out of the wireless access bottleneck trap: technologies, economics, regulation and standardization perspectives
- First Online:
- Cite this article as:
- Reichl, W., Reichl, P. & Reichel, P. Elektrotech. Inftech. (2012) 129: 400. doi:10.1007/s00502-012-0056-6
- 105 Views
The current evolution of mobile broadband is expected to significantly increase the demand for wireless access resources in the near future. In this article, three ways to escape this wireless access bottleneck trap are examined. The first option is technologically driven and corresponds to enhancing the spectral efficiency of access technology, thus making better use of available wireless resources. However, there are theoretical limits for this approach, and further improvements might be complex and costly. The second option is based on allocating additional spectrum for mobile applications. Here, the “digital dividend” provides a unique opportunity, but the corresponding process is lengthy due to the required international harmonization. The third option focuses on densification of access points, which usually requires considerable investments. However, user provided small cell networks could provide an ingenious way out of the trap. Finally, the three options are compared, and drivers and barriers for the different approaches are analyzed.
Keywordsmobile broadbanddigital dividendwireless access resources3G4GLTEGSMUMTSFONsmall cell networksfemto cells, WiFi aggregation
Mehr Ressourcen für mobiles Breitband: Technologien – Märkte – Regulation – Standardisierung
Die derzeitigen Entwicklungen auf dem Gebiet des mobilen Breitbands lassen eine weitere deutliche Steigerung der Nachfrage nach Ressourcen in drahtlosen Zugangsnetzen für die nahe Zukunft erwarten. Der vorliegende Artikel analysiert drei unterschiedliche Optionen zur Beseitigung dieser Ressourcenknappheit. Der erste Ansatz ist stark technologiegetrieben und läuft auf die weitere Steigerung der spektralen Effizienz hinaus, um eine bessere Nutzung vorhandener Ressourcen im Zugangsnetz zu ermöglichen. Hierfür gibt es jedoch theoretische Grenzen, und Verbesserungen hier können sich durchaus als komplex und kostspielig herausstellen. Alternativ dazu ist die Zuteilung von zusätzlichem Spektrum vorstellbar. In diesem Zusammenhang stellt die so genannte „Digitale Dividende“ eine einzigartige Gelegenheit dar, wobei allerdings die damit zusammenhängenden Prozesse langwierig sind und eine Harmonisierung auf internationalem Gebiet erfordern. Drittens schließlich könnte man die räumliche Dichte der Zugangsknoten erhöhen – üblicherweise ebenfalls mit erheblichen Kosten verbunden. Hier allerdings könnte die Einbindung von endnutzerbetriebenen so genannten „Small Cell“-Netzen einen unerwartet einfachen Weg aus der Problematik bereitstellen. Die drei Optionen werden abschließend miteinander verglichen, um die jeweiligen Technologietreiber bzw. -barrieren herauszustellen.
Schlüsselwörtermobiles Breitbanddigitale Dividendedrahtlose Zugangsnetze3G4GLTEGSMUMTSFONFemto-Zellen: WLAN-Aggregierung
The Global mobile Suppliers Association (GSA)1 reports over 5.2 billion cellular mobile subscriptions (GSM, WCDMA-HSPA and LTE technology) globally at the end of 2011. Although most of these user equipment (UE) units are used for voice telephony, the ratio of mobile data connections is quickly rising.2 Historically, while mobile data connections have been possible already since the introduction of GSM in the early 1990s, mobile data has remained a niche market for some time. Technological evolution, the emergence of smartphones and tablet PCs and the increasing competition between mobile operators have now made the provision of mobile data cheap enough to challenge fixed broadband (Feiel 2010). Current statistics demonstrate the importance of mobile broadband as an emerging market which has reached significant penetration rates in many countries.3
More efficient use of existing frequencies: Spectral efficiency is commonly defined as the information rate that can be transmitted over a given bandwidth in a specific communications system. For historical reasons, spectral efficiency used to be considered independent of the channel condition (i.e. Signal-to-Noise Ratio, SNR), while it seems to be more appropriate to refer to the ideal case (i.e. best channel condition). In order to compare the capacity of cellular network technologies, the notion of “system spectral efficiency” is used, which is measured in bit per second per Hertz per area and describes the quantity of users that can be simultaneously supported by a limited radio frequency bandwidth in a defined geographic area. Advances in micro-electronics allow more sophisticated channel coding techniques and the use of multiple antennas enable higher system spectral efficiency which might get close to the theoretical limit which is described for the case of point-to-point spectral efficiency by the Shannon-Hartley theorem and may be surpassed only by multi-user coding.6
Additional spectrum: Beyond the improvements of efficient usage of the existing spectrum, a second source for increasing wireless access capacities may arise from time to time in the form of additional spectrum. One particularly important example concerns the current changeover from analog to digital television which has freed up spectrum resources known as the “Digital Dividend”. As a consequence, frequency spectrum previously used for television distribution has been reallocated to mobile communications. In Germany, for instance, the spectrum band between 790 and 862 MHz has been auctioned in April and May 2010, and successful bidders have been required by the Deutsche Bundesnetzagentur to use this spectrum primarily for the coverage of so called “white spots”. Other National Regulation Authorities (NRAs) will follow, and it is envisaged that also further parts of the spectrum currently used for terrestrial television will be made available for mobile communications. The “Digital Dividend” is very attractive for cellular communications because of the favorable propagation characteristics. Other frequency bands being made available for cellular communications are the 2 GHz and 3.5 GHz band (note that in many countries 1.8 GHz and 2.6 GHz are already allocated for LTE). However, spectrum is scarce, and any reallocation, especially due to the need for highly complex international harmonization, requires long lead times and transition periods before becoming effective.
Access points densification: The third option to increase capacity is to decrease the size of the cells. This is already required in urban areas and for inhouse coverage. If cells become smaller and smaller, they might in the end be restricted to one apartment—this is known as the idea of femtocells using already existing fixed Internet infrastructure for connecting the base stations, i.e. for backhauling. Although this idea is quite compelling, it might cause higher investments, could mean a disruption of the business models of mobile operators and increase the requirements for efficient and robust Radio Resource Management (RRM) techniques which are able to cope with huge numbers of small cells. As an alternative to femtocell base stations controlled by the operators and using licensed spectrum, small cell networks could also be based on aggregated WLAN Access Points (APs) in the unlicensed ISM (Industrial, Scientific and Medical) band.
The evolution of technologies, which allow more efficient uses of spectrum is discussed in Sect. 2. The quest for further spectrum is analysed in Sect. 3 and access point densification is discussed in Sect. 4. Section 5 summarizes our findings and examines the drivers and barriers for each of the three ways to get more wireless access resources.
2 Technologies for increasing spectrum efficiency
GSM technology uses a 200 kHz carrier spacing divided into 8 time slots (FDD/TDD). GPRS allows data transmission up to 114 kbit/s, while EDGE implements advanced modulation technologies and will be superseded by evolved EDGE allowing data rates of more than 1 Mbit/s. GSM networks deployed today usually support GPRS and EDGE technology; the GSM air interface uses primarily the 900 and 1800 MHz (in the United States also the 1900 MHz) band.
Wideband CDMA: In addition to the GSM interface an air interface based on wideband CDMA has been developed, commonly known as 3G mobile communications. The basic data rate for UMTS is 384 kbit/s downstream and 64 kbit/s upstream, meanwhile, also within UMTS developments higher data rates have been achieved. HSPA, which is a combination of HSDPA and HSUPA, allows for 14.4 Mbit/s in the downlink and 5.76 Mbit/s in the uplink. HSDPA technology has been introduced by almost all wideband CDMA operators. The successor technology HSPA+ will lead to a further increase of data rates. According to GSA, as of March 2012 a total of 451 commercial HSPA networks has been deployed, while 187 commercial HSPA+ systems exist in 96 countries.
LTE: The next technological step, called Long-Term Evolution (LTE), uses an air interface based on OFDMA in the downlink and SC-FDMA in the uplink. LTE is optimized for packet transport and is supposed to allow theoretically for data rates up to 326 Mbit/s in the downlink. LTE is being marketed in the United States as mobile telephony of the fourth generation; for instance Verizon has announced to cover two thirds of the US population with LTE within the next 18 months, while AT&T will finalize its rollout at the end of 2013.
LTE-Advanced as designed by the ITU-T is supposed to bring data rates of up to 1 Gbit/s in the downlink and will be the first release which is compliant to the requirements of the IMT-A definitions for future 4G technologies.
LTE with theoretical data rates of 326 Mbit/s and LTE advanced with data rates up to 1 Gbit/s will definitely constitute competition to fixed broadband networks. The worldwide mass market of more than 5 billion users will also allow economies of scale, which will lead to falling equipment prices, thus mobile infrastructure will also be able to compete with fixed infrastructure from an economic point of view.
Note, however, that theoretical peak bit rates are often not sufficient for enabling a satisfying mobile broadband experience. According to the netindex, published by Speedtest.net the average actual download speed in fixed broadband networks in spring 2012 is nearly 10 Mbit/s.9 Traffic theory shows that not all subscribers need this data rate at the same time. Considering about 5 GByte monthly download volume, an average of 100 kbit/s per user needs to be considered the basis for network planning. Rysavy Research (2010) shows a comparison between theoretical data rates of mobile access technologies and typical user rates achieved in real networks. Based on that, only HSPA+ and LTE are currently capable of competing with the actual download speed experienced in fixed networks. Apart from looking at the achievable download rate, also the capacity of the cell has to be designed correctly. In practice, a network is first and foremost planned on the basis of coverage, while densification is additionally applied in areas of high demand (i.e. mostly urban areas). The use of mobile broadband will probably show a different pattern than fixed broadband since it allows for more flexibility. To understand the precise demand characteristics of mobile broadband needs further research, while it is universally accepted that future demand will steadily increase.10 This is for instance due to the technological evolution of tablet computers which are expected to become the dominant source of bandwidth needs, roughly doubling their capacities every year. The available infrastructure should allow to cover this demand, new sites are costly and network planning should be sufficiently flexible to increase capacity without adding sites.
Although mobile access technology can technically provide data rates comparable to fixed broadband networks, to achieve high capacity significant investment is required. The willingness of customers to pay for higher data rates is limited. Economic reasons therefore represent a barrier to roll-out high speed access networks. This holds true as well for the deployment of fibre in fixed broadband access networks as for high capacity mobile access networks.
3 Augmenting spectrum for mobile communications
The original spectrum used for GSM is the 900 MHz band, and building their GSM network grid on this band has enabled the operators to achieve extensive coverage. Rising demand has required enhancements, and in this context the 1800 MHz band is of specific importance, despite the fact that the physical propagation properties in this specific band are less favorable and require new approaches in network planning.11 UMTS has been deployed primarily in the 2.1 GHz band, while coverage obligations for UMTS are generally lower as for GSM.
Note that in general frequency bands below 1 GHz have better propagation characteristics and do not require new sites. One of the most far-reaching opportunities for allocating such spectrum below 1 GHz to mobile communications is provided by the so-called Digital Dividend. The term “Digital Dividend” refers to spectrum which is released in the process of transition from analog to digital transmission of television programs. Here, compression technologies allow transmitting eight digital channels by using the same amount of spectrum as formerly one analog channel. However broadcasters and mobile industry have argued that they need the Digital Dividend for further growth. Broadcasters require additional capacity e.g. for HD channels and point to the immense value of broadcasting for opinion diversity (SBR Juconomy Consulting 2009). On the other hand, mobile operators have argued that the economic gains are much higher if the spectrum is allocated to mobile communications. At the end regulatory authorities seem to be more convinced by the arguments of the mobile operators.12
In May 2010 the Commission of the European Communities decided to harmonize the technical conditions for the availability and efficient use of the 790-862 MHz band (800 MHz band) for terrestrial systems capable of providing electronic communications services in the European Union.13 Germany has auctioned the spectrum in 2010 under the condition that operators use the 800 MHz band for the coverage of underserved rural areas with mobile broadband; this obligation has already been fulfilled to a large extent.14 Austria is planning an auction of the 800 MHz band to be held probably in 2013 or 2014.
During the World Radio Conference 2012 (WRC-12), delegates from Africa and Middle East have indicated demand for a second Digital Dividend in the 700 MHz band, since in these areas parts of the 800 MHz band are used for other services and systems. This issue has been placed on the WRC-15 agenda together with the need to consider additional spectrum allocations for the mobile service. In the meanwhile, WRC-12 has invited administrations to indicate spectrum requirements for mobile services, broadcasting services and other services. ITU-R will study spectrum requirements, channeling arrangements and the compatibility between the mobile service and other services currently allocated in the frequency band 694–790 MHz.15
Reallocation of spectrum, however, is a lengthy process, and the potential auctioning of the 700 MHz spectrum might start as late as 2017. Considering the estimated growth of mobile broadband, the allocation of a second Digital Dividend for mobile communications will not be sufficient to fulfill the demand, and thus additional spectrum will be only one amongst several other options for augmenting wireless access resources.
Operators of mobile networks deploy various radio technologies in various frequency bands, and—as already mentioned—spectrum below 1 GHz has good propagation characteristics and is well-suited for all technologies. A further trend is the flexibilization of frequency use.16 Accordingly, spectrum previously reserved for GSM technology will be available for UMTS and LTE in the future as well. Altogether, both enlargement of available spectrum and flexibilisation of usage will contribute to a more efficient use of wireless access resources.
4 User provided small cell networks
Increasing the capacity of the macro network infrastructure through placing base stations into the living rooms or offices of the end customers and connecting (backhauling) them to the core network through already existing cheap public Internet access like cable or DSL provides a true alternative to the approaches discussed so far, allowing for focused macro network offload in regions with high total bandwidth requirements. Economically, both the significant reduction of power consumption and the advantages due to economies of scale (i.e. cheap production of large numbers of equipment) are of interest.
3GPP femto cells operate in the licensed spectrum, employing either UMTS (based on WCDMA) or LTE technology (based on OFDMA) which is based on a flat All-IP architecture in the packet core network. This allows the mobile operators providing impressive downlink bitrates (up to 84 Mbps for HSDPA+ and more than 300 Mbps for LTE) as well as enforcing QoS, based on their control over radio resource management (RRM) and AAA.
In contrast to femto cells, WLAN (IEEE 802.11 a/b/g/n) is operating in the unlicensed ISM band, based on OFDM technology with high link bit rates, low latencies and ad-hoc (flat hierarchy) RRM mechanisms. However, WLAN has not been conceived as a truely mobile technology, which is mirrored in the well-known problems of connection establishment to WLAN APs due to interference, congestion and/or coverage issues as well as the lack of SIM-based AAA.
In a first step, the project has performed a measurement campaign investigating the actual status of small cell coverage in urban areas. To this end, an “exhaustive walk” through every street of a 500×500 m2 part of the 2nd district of Vienna has been performed, resulting in a total route length of around 10 km. Along this route, statistics of local distribution and signal strength of in-door Wi-Fi installations have been collected together with signal strengths of the W-CDMA macro network of a leading Austrian mobile operator. The resulting measurement data indicate that already now more than 90 % of the street area could be covered if only half of the CPE equipment of the operator would be converted to support roaming. Furthermore, if the two largest operators collaborate, a 100 % coverage could be easily realized in this way (Fuxjäger et al. 2010). Finally, it is extremely interesting to note that the signal strength of the indoor Wi-Fi access points usually match or even exceed those due to the macro network infrastructure (Fuxjäger et al. 2011).
Another branch of the AWARE project deals with design and implementation of a prototype which comprises a novel measurement system placed at the WLAN access point for determining the intensity of radio resources used by the mobile terminals. On the side of the smartphones, a dedicated connection manager is employed supporting network selection protocols like IEEE 802.21 (Media Independent Handover) or 3GPP ANDSF (Access Network Discovery and Selection Function) which connects to a network server and uses congestion information for deciding on the optimal wireless access. In this way, congestion situations on the WLAN links may easily be detected and remedied through migrating one or more selected users back to the cellular macro network.
From an economic perspective, research in AWARE focuses on the fundamental role change from traditional end users towards so-called “prosumers” which both offer and use wireless access resources in parallel (Reichl et al. 2011). To this end, a model for wireless access resource sharing has been studied which is strongly influenced from the FON network17 which has been created in 2005 and claims to have grown afterwards into the largest WLAN network in the world with allegedly over 4 million access points. Basically, FON creates a user community where all members agree to offer part of their WLAN resources to other members or roaming users in exchange for money and/or free access if members are roaming themselves. Depending on their profile, FON members have originally been classified as either Bills (selling wireless access resources), Linuses (offering part of their resources for free and roaming for free) and Aliens (paying for access, but not contributing own resources).
While there are clear indications that the initial success of FON might not turn out to become a sustainable one, the FON model may nevertheless serve as inspiration for a similar approach in the area of UPSC. Thus, the model of the corresponding AWARE community distinguishes between Service Providing Customers (SPCs) and Service Requesting Customers (SRCs), while in addition also the Mobile Network Operator (MNO) plays a key role. SPCs own wireless access contracted to an MNO, while SRCs are interested in using resources from nearby access points for operating their terminals.
It is an interesting question whether the sketched MNO-driven approach would suffer a similar destiny like FON, or whether the presence of the MNO will make a significant difference. To start with, AWARE has started to analyze the resulting SPC and SRC strategies from a game-theoretic perspective in order to establish corresponding Nash equilibria (Reichl et al. 2011). In the next step, a learning algorithm has been proposed which allows reaching these equilibria through periodically updating a probability vector for the available pure strategies based on utility gains from the previous iteration while an additional parameter controls the speed of the learning process.
Demand for mobile broadband will increase and put more and more pressure especially on resources in the wireless access. Therefore, in the previous chapters we have described and analyzed three different ways to enhance the capacity of cellular networks, while we are now going to discuss the drivers and barriers for each of these approaches.
The technological drivers can be listed in a straightforward manner. The suppliers drive the standardization of new technologies and spend considerable amounts of money in research and development. We observe the trend to implement more and more in software implementation. Hardware will play a minor role in the same time. Today base stations are marketed which support all air interfaces. The huge worldwide market allows the suppliers to benefit from economies of scale. The implementation by network operators will be driven by demand, while the operators will throughout prefer upgrades that do not require new sites.
Further spectrum resources will be made available, but there is a considerable difference in market value for the different spectrum bands. The 800 and 900 MHz band have received considerable attention due to the excellent physical propagation properties, and auction results are expected to be rather high. On the contrary, the demand for frequencies in the 2 and 3 GHz bands is less and auction results are lower. One of the key reasons for that might be that most of the network operators have their grid still based on 900 MHz frequencies and do not have sufficient coverage in the 3G grid. Finally, note that occasions like the spectrum reallocation in the course of the “Digital Dividend” are more or less individual events which happen only occasionally and on a rather long-term time scale which makes it difficult to seriously predict their contribution to a steady growth of available access bandwidth.
In this respect, advances in spectral efficiency are achieved in a much more continuous and predictable manner, however, also here the necessary harmonization procedures slow down the adoption process significantly. Together with the problems of fragmented bandwidth allocation, these are considered to be the main stumbling blocks towards a fast adoption of LTE and LTE-Advanced systems.
User-provided small cell networks offer a relatively straightforward way to achieve a significant increase in wireless access capacity already on a short term perspective. However, the resulting densification due to the huge number of low power radio installments will provide a true challenge to legacy MNO models. Note especially that all of the Wi-Fi access points need backhauling and thus require some form of management, which results in the strict requirement that it must be avoided to attach a complex stack to the APs, as otherwise costs for the packet core might grow too high. Solving the resulting tension between future 3GPP technologies and the evolution of the IEEE 802.11 standards family will become a pivotal research challenge for the years to come.
Drivers and barriers for the three ways to enhance wireless access resources
Increasing spectral efficiency
• Advances in micro electronics
• Global mass market
• Software instead of hardware
• Equipment life cycle
• Supporting existing UE
Provisioning additional spectrum
• Digital Dividend
• Long lead times
• Propagation characteristics
• International harmonization
User provided small cell networks
• Easy to implement
• Cost efficient
• Backhaul infrastructure available
• Business models
Finally, we would also like to underline once more the role of the different time scales related to each of the three options. While the increase of spectral efficiency has been shown to steadily increase with every release of a new access technology specification (which usually happens on a biannual basis), the release of additional spectrum which could be allocated to mobile applications must be considered a relatively rare and irregular even, while the length of the corresponding lead times is hardly aligned with usual innovation cycles in the telecommunications industry. This is not true for the small cell approach as the required cheap backhauling infrastructure is already available, and a commercial roll-out could begin relatively soon. From this perspective, it might be indeed a combination of all three of these approaches (each one realized on its own time scale) which will finally allow to cope with the access bottleneck problem by increasing wireless access resources to such an extent that, even with the current CAPEX and OPEX restrictions, the resulting Quality of Experience for the mobile broadband user may be constantly improved in the foreseeable future.
In the case of Austria, RTR Telekom Monitor 1/2012 page 38 shows that already 60 % of all broadband connections are mobile.
See e.g. http://en.wikipedia.org/wiki/Spectral_efficiency for a survey table of spectral efficiency of current mobile networking technologies/.
The targets for average spectrum efficiency of LTE-Advanced are defined in 3GPP TR 36.913, Table 8.1. For instance the target for downlink spectrum efficiency for antenna configuration 4×4 is 3.7 bps/Hz/cell. The target for peak spectrum efficiency in the downlink is 30 bps/Hz.
Cf. Cisco Visual Networking Index (VNI).
See: European Commission Decision 2010/267/EU.
WRC-12: Resolution COM5/10.
See European Commission Decision 2009/766/EC for the usage of other terrestrial systems in the 900 and 1800 MHz band.
Part of this work has been funded by the Austrian Government and the City of Vienna within the COMET program; further support from Université Européenne de Bretagne and Télécom Bretagne Rennes as well as LAAS-CNRS Toulouse, France and the Department of Communications and Networking, Aalto University Helsinki, Finland, is gratefully acknowledged. The authors would like to cordially thank Paul Fuxjäger for his extremely helpful input, and Ivan Gojmerac, Ronny Fischer and Olivier Marcé together with the entire AWARE team as well as Helmut Malleck and Olav Ruhle for many interesting and fruitful discussions.