ESG. UMTS900 Overview & Deployment Guidelines. November W Rev A. Engineering Services Group

ESG Engineering Services Group UMTS900 Overview & Deployment Guidelines November 2006 80-W1044-1 Rev A

Export of this technology may be controlled by the United States Government. Diversion contrary to U.S. law prohibited. QUALCOMM Incorporated 5775 Morehouse Drive San Diego, CA 92121-1714 USA Copyright 2006 QUALCOMM Incorporated. All rights reserved. UMTS900 Overview & Deployment Guidelines 80-W1044-1 Rev A

Table of Contents 1. Introduction... 1-1 1.1 Purpose...1-1 1.2 Scope...1-1 1.3 Document Organization...1-1 1.4 Conventions...1-1 1.5 Revision History...1-1 1.6 Acronyms...1-2 1.7 References...1-2 2. UMTS900 Overview... 2-1 2.1 UMTS900 in 3GPP Specifications...2-1 2.1.1 Receiver and Transmitter Performance...2-1 2.1.2 Signaling Requirements...2-1 2.2 UMTS900 Benefits...2-2 2.2.1 Coverage, Capacity, In-building Penetration Considerations.2-3 2.2.1.1 Propagation and Link Budget Differences...2-3 2.2.1.1.1 Cell Site Antenna Gain...2-3 2.2.1.1.2 Node B Cable Loss...2-4 2.2.1.1.3 Antenna Height Gain...2-5 2.2.1.1.4 Building Penetration Loss at 900 MHz versus 2100 MHz...2-5 2.2.1.2 Coverage Comparison...2-9 2.2.1.3 Capacity Comparison...2-14 2.2.2 Link Budget Comparison (UMTS900-UMTS2100)...2-14 3. UMTS900 Deployment Guidelines... 3-1 3.1 Deployment Scenarios...3-1 3.1.1 Rural, Urban Deployments & Objectives...3-2 3.1.1.1 Rural Deployments...3-3 80-W1044-1 Rev A iii

3.1.1.2 Urban Deployments...3-3 3.1.2 Coordinated vs. Uncoordinated Deployments...3-5 3.1.2.1 Coordinated Operation...3-6 3.1.2.2 Uncoordinated Operation...3-8 3.2 Design Guidelines for Multi-band and Multi-technology Sites...3-9 3.2.1 Site Configurations for Co-located GSM and UMTS...3-10 3.2.1.1 Configuration 1...3-11 3.2.1.2 Configuration 2...3-11 3.2.1.3 Configuration 3...3-11 3.2.1.4 Configuration 4...3-11 3.2.2 GSM-UMTS Antenna Sharing Deployment (Config. 2 & 4)..3-12 3.2.3 Real Estate Sharing Deployment (Config. 1 & 3)...3-12 iv 80-W1044-1 Rev A

List of Figures Figure 2-1 Building Attenuation at Ground Level [5]...2-6 Figure 2-2 UMTS900 and UMTS2100 propagation using the COST 231-Hata Model...2-11 Figure 2-3 UMTS900 and UMTS2100 propagation using the COST 231-WI suburban model....2-13 Figure 2-4 UMTS900 and UMTS2100 propagation using the COST 231-WI Metro model....2-13 Figure 3-1 UMTS2100, UMTS900, GSM900 Deployments...3-1 Figure 3-2 Isolation distance between UMTS900 and GSM900...3-2 Figure 3-3 UMTS900 deployments in rural areas...3-3 Figure 3-4 UMTS900 deployments in urban areas...3-4 Figure 3-5 Interference Scenarios...3-6 Figure 3-6 Coordinated operation...3-7 Figure 3-7 Sandwich type deployment...3-7 Figure 3-8 Uncoordinated operation...3-8 Figure 3-9 Recommended frequency allocation for GSM micro/pico cells...3-9 Figure 3-10 Possible Deployment Configurations for Co-Located GSM & UMTS Systems...3-10 Figure 3-11 Horizontal Spacing...3-13 Figure 3-12 Vertical Spacing...3-13 80-W1044-1 Rev A v

This page intentionally left blank. vi 80-W1044-1 Rev A

List of Tables Table 1-1 Revision history...1-2 Table 1-2 Acronyms...1-2 Table 2-1 65 degree Horizontal Beamwidth Antenna Characteristics (single vendor)...2-4 Table 2-2 Sample Cable Losses for 50Ω coaxial cable...2-4 Table 2-3 Building Type Definitions [5]...2-7 Table 2-4 BPL as a decreasing function of frequency [1]...2-8 Table 2-5 BPL as an increasing function of frequency [2]...2-8 Table 2-6 Uplink Budget for Scenario 1, Urban...2-16 Table 2-7 Uplink Budget for Scenario 1, Medium...2-16 Table 2-8 Uplink Budget for Scenario 1, Residential...2-17 Table 2-9 Uplink Budget for Scenario 2, Urban...2-17 Table 2-10 Uplink Budget for Scenario 2, Medium...2-17 Table 2-11 Uplink Budget for Scenario 2, Residential...2-18 Table 2-12 Summary of Uplink Budget Maximum Allowable Path Loss Differences...2-18 Table 3-1 Triple Band Antenna Key Parameters...3-12 Table 3-2 Antenna Parameters...3-13 Table 3-3 Antenna Isolation Measurements Results...3-14 Table 3-4 900 MHz Single Band Antenna Key Parameters...3-14 80-W1044-1 Rev A vii

This page intentionally left blank. viii 80-W1044-1 Rev A

1. Introduction 1.1 Purpose The purpose of this document is to give UMTS900 overview along with coverage, capacity, building penetration loss, and link budget comparison with UMTS2100 and to describe UMTS900 deployment scenarios and guidelines, co-existence issues with GSM900, and guard band requirements. 1.2 Scope This document is intended for wireless operators and UTRAN equipment vendors. 1.3 Document organization This document is organized into 3 chapters:! Chapter 1. gives the purpose and scope of the paper along with conventions, revision history, acronyms, and references.! Chapter 2. provides overview of UMTS900 and comparison of coverage, capacity, inbuilding penetration, and link budget in UMTS900 and UMTS2100.! Chapter 3. explains deployment scenarios, frequency allocation considerations, adjacent channel interference issues with GSM900, guard band requirements, and site/antenna sharing considerations. 1.4 Conventions Function declarations, function names, type declarations, and code samples appear in a different font, e.g., #include. Code variables appear in angle brackets, e.g., <number>. Commands and command variables appear in a different font, e.g., copy a:*.* b:. Shading indicates content that has been added or changed in this revision of the document. 1.5 Revision history Table 1-1 shows the revision history for this document. 80-W1044-1 Rev A 1-1

Version Date Description A Nov. 2006 Initial release Table 1-1 Revision history 1.6 Acronyms Table 1-2 lists acronyms used in this document. Term ACIR ACLR ACS BPL BS BTS CM DL MCL RF UE UL VSWR Definition Adjacent Channel Interference Rejection Adjacent Channel Leakage Ratio Adjacent Channel Selectivity Building Penetration Loss Base Station Base Transceiver Station Compressed Mode Downlink Minimum Coupling Loss Radio Frequency User Equipment Uplink Voltage Standing-Wave Ratio Table 1-2 Acronyms 1.7 References [1] A.F. de Toledo, A.M.D. Turkmani and J.D. Parsons, Estimating Coverage of Radio Transmission Into and Within Buildings at 900, 1800, and 2300 MHz, IEEE Personal Communications, April 1998, pp. 40-47. [2] S. Aguirre, L.H. Loew, Yeh Lo, Radio propagation into buildings at 912, 1920, and 5990 MHz using microcells, Universal Personal Communications, 1994 Third Annual International Conference on Record, 27 Sept. 1 Oct. 1994, pp. 129-134 [3] J. D. Parsons, The Mobile Radio Propagation Channel, John Wiley & Sons, 1992, p. 194. [4] W. C. Y. Lee, Mobile Communications Engineering, McGraw-Hill Book Co., 1982, Chapter 4. 1-2 80-W1044-1 Rev A

[5] Kazimierz Siwiak, Radiowave Propagation and Antennas for Personal Communications, 2 nd ed., Artech House, 1998. [6] A.M.D. Turkmani and A.F. de Toledo, Modeling radio transmissions into and within multistorey buildings at 900, 1800 and 2300 MHz, IEE Proc.-I, Vol. 140, No. 6, Dec. 1993, pp. 462-470. [7] COST ACTION 231, Digital Mobile Radio Towards Future Generation Systems: COST 231 Final Report, Directorate-General, Telecommunications, Office of Official Publications of the European Communities, Luxembourg, ISBN 92-828-5416-7, Chapter 4 Propagation Prediction Models, 1999. [8] 3GPP TS. 25.104: Base Station (BS) radio transmission and reception (FDD) (Release 7). [9] 3GPP TR 25.942, 3rd Generation Partnership Project; Technical Specification Group Radio Access Networks; Radio Frequency (RF) system scenarios. (Release 6). [10] 3GPP TAG RAN WG4 Tdoc 631/99: Antenna-to-Antenna Isolation Measurements. [11] ETSI/STC SMG2 Tdoc 48/93: Practical Measurement of Antenna Coupling Loss. [12] 3GPP TR 25.816: UMTS 900 MHz Work Item Technical Report (Release 7). [13] ECC Report 82: Compatibility Study for UMTS Operating Within the GSM 900 and GSM 1800 Frequency Bands. [14] 3GPP TS. 25.101: User Equipment (UE) radio transmission and reception (FDD) (Release 7). 80-W1044-1 Rev A 1-3

This page intentionally left blank. 1-4 80-W1044-1 Rev A

2. UMTS900 Overview 2.1 UMTS900 in 3GPP Specifications In December '05, 3GPP completed study of radio frequency and inter-operability requirements for UTRA-FDD operation in the 900 MHz band and published results in TR25.816. The 900 MHz band, denoted as Band Class VIII, is defined as the paired bands from 880 to 915 MHz in the uplink direction, and from 925 to 960 MHz in the downlink direction. UMTS900 receiver and transmitter performance are specified in TS25.101 and TS25.104 for the UE and the base station, respectively. RRC protocol requirements are specified in TS25.331. 2.1.1 Receiver and Transmitter Performance Because of the small duplex distance (45 MHz) and duplex gap (10 MHz), with the same duplexer technology and size, the 900 MHz duplexer insertion loss in a UE is expected to be bigger than that experienced, e.g., at 2100 MHz. Therefore 3GPP has relaxed sensitivity requirements for UMTS900 UE by 3 db relative to those of UMTS2100, resulting in a reference DPCH sensitivity equal to -114 dbm. On the other hand, UMTS900 base station sensitivity requirements are unchanged because there are no size constraints preventing design of duplexes with the same insertion loss as that experienced at 2100 MHz. Receiver in-band and out-of-band blocking requirements are defined for UMTS900 similarly as for other band classes. Narrow band blocking performance requirements are identical to those for UMTS1800: the UE can suffer no more than 10 db sensitivity degradation in the presence of a GMSK blocking signal received at -56 dbm with frequency offset equal to 2.8 MHz. Finally, receiver demodulation performance requirements are identical to those of the other bands, other than that the UE velocity is scaled proportionally to the carrier frequency. For example, demodulation performance in a fading channel are specified for UMTS2100 for Case I and Case II at 3 Km/h, while for UMTS900 same performance requirements apply at 7 Km/h. Transmitter performance requirements in terms of emission mask, spurious emissions etc. are identical to those applicable to the other bands. 2.1.2 Signaling Requirements RRC protocol extensions to support UMTS900 have been added to Release 6 of the TS25.331 specification. Note however that band class support is a release independent feature, and hence a UMTS900-capable UE can still be implemented following R99 or R5 specification provided it also supports the relevant UMTS900 signaling extensions specified in R6. Such extensions affect mainly the syntax of the base station frequency band indicator and UE measurement capability information elements (IE) in the applicable RRC signaling messages. 80-W1044-1 Rev A 2-1

The frequency band indicator in SIB5, used by the base station to advertise the band classes it supports, has a new enumerate corresponding to Band Class VIII. The UE on the other hand advertises its capability of operation in Band Class VIII by means of the Measurement Capability Extension IE, which is part of the UE Radio Access Capability IE and carried by the RRC Connection Setup Complete message. The RNC at the time of RRC connection establishment stores such information in the context maintained for each connected UE and uses it to control handovers from/to the UMTS900 band. 2.2 UMTS900 Benefits The benefits of UMTS900 are all a result of the lower carrier frequency. UMTS at 900 MHz will propagate further than its equivalent at 2100 MHz. This can be seen directly in the propagation models commonly used for cellular network planning and discussed in detail in a later section. Better propagation characteristics are only part of the picture. Electronic devices generally have more gain and/or lower noise figures in the 900 MHz band than at higher frequencies. In other words, besides propagating better, manufacturers can make more transmit power at less cost and generally better receivers at 900 MHz. Standards will set the maximum UE transmit power as appropriate to the UE class, as well as the maximum noise figure, but the cost to produce should diminish. This phenomenon of electronics is aided by the fact that GSM900 has been successfully bringing down the cost and increasing the availability of components in the 900 MHz band for some years now. It should be pointed out that not all aspects of 900 MHz over 2100 MHz are positive. For example, antenna apertures are a function of wavelength. This means that to have the same antenna gain at 900 MHz as 2100 MHz would require an antenna structure that is approximately twice the size. This phenomenon applies, of course, to the antennas at both ends of the link. But this detriment can be overcome with better propagation characteristics and more efficient power amplifiers. Better propagation also means the interference, both UL and DL, will propagate better. The increase of interference due to better propagation conditions must be considered in the context that UMTS900 systems will likely be built on cell site separations that were originally scaled for GSM900. Since they are both operating in the 900 MHz band, one might assume that increased levels of interference should not be a consideration. This might be true except for the fact interference is mitigated and managed in GSM through the frequency plan. Mature GSM networks have very dense cell spacing, particularly in areas with high capacity requirements such as urban areas. These areas of close cell site spacing will perhaps require lower antenna heights, increased down-tilt and optimization efforts that were not part of the GSM900 network. For that reason, it may be advantageous to use separate antennas for the UMTS900 signals than are used for the GSM900 signals. The ability to independently make antenna adjustments for the UMTS900 and the GSM900 systems may be key to successful optimization of each network. In Section 2.2.1.1, we explore each of the aspects of UMTS900 radio propagation and its influence on RAN planning. Section 2.2.2 presents representative link budgets which compare UMTS900 and UMTS2100 uplink voice service. 2-2 80-W1044-1 Rev A

2.2.1 Coverage, Capacity, In-building Penetration Considerations Differences in achieved coverage and capacity between UMTS at 900 MHz and 2100 MHz are due to two factors: link budget differences and propagation differences. Specific link budgets are presented in Section 2.2.2 below. In preparation for that discussion, changes in propagation as well as other associated physical phenomena are presented in this section. 2.2.1.1 Propagation and Link Budget Differences 2.2.1.1.1 Cell Site Antenna Gain Cell site antennas can be either single band or multi-band. The single band antennas will be optimized, both in terms of efficiency and pattern, for the given band; 900, 1800, or 2100 MHz. Single band antennas would be deployed in situations where individual antennas are used for each band. As discussed briefly above, using separate antennas for each frequency band and for each service type (GSM or WCDMA) has advantages for optimization and coverage. In other words, it permits these services to be optimized independently. The side effect, of course, is that more antennas need to be mounted. This may conflict with zoning requirements. It will also add more cable runs. These are cost factors that apply equally to both 900 and 2100 MHz systems. With regard to dual band and triple band antennas, a quick survey of antenna manufacturers found antenna gains and sizes that could be divided into two classes: 1. Some dual (triple) band antennas have the same gain at 900 MHz, 1800 and 2100 MHz. This is accomplished by changing the horizontal beamwidth of the antenna at each of the bands. For example, one antenna of this type starts with 68 degree horizontal beamwidth at 900 MHz and gradually reduces to a 62 degree horizontal beamwidth at 2100 MHz. This antenna is still advertised as a 65 degree beamwidth antenna, by the way. 2. There are also antennas that keep the horizontal beamwidth constant and let the gain change. A typical example of this type of antenna would be triple band antennas that have approximately 2.7 to 2.3 db of difference in gain between 900 MHz and 2100 MHz. Table 2-1 displays a range of typical antenna gains and sizes for both 900 MHz and 2100 MHz. 80-W1044-1 Rev A 2-3

Table 2-1 65 degree Horizontal Beamwidth Antenna Characteristics (single vendor) It can be seen that gain, vertical beamwidth, and physical antenna dimensions will differ. A range of gains are possible in each frequency band. It might be noted that the 900 MHz antenna with 16.8 dbi of gain is more than twice the size of an approximately equivalent antenna, with 16.5 dbi, in the 2100 MHz band. The choice of antenna will impact the link budget which, in turn, will impact deployment strategy decisions. 2.2.1.1.2 Node B Cable Loss Coaxial cable loss increases at higher frequencies. This means that transmit antennas with the same length cable run will exhibit higher cable loss at 2100 MHz than at 900 MHz. Cable loss is a direct contributor to differences in the link budget between 2100 MHz and 900 MHz systems. Table 2-2 below provides some typical cable loss values at each frequency band assuming a 30 m length of cable. Table 2-2 Sample Cable Losses for 50Ω coaxial cable If it is important to have similar cable losses between the two bands, it may be possible to choose a greater diameter (lower loss) cable for the 2100 MHz system. The specific impact on each link budget depends on the details of the deployment strategy. Tower mounted amplifiers might be considered for the UMTS2100 system in some cases, but would only ease the cable loss requirement on the UL. 2-4 80-W1044-1 Rev A

2.2.1.1.3 Antenna Height Gain In some deployment scenarios, it might be considered to mount the 900 MHz antennas at a height that is less than the 2100 MHz antennas. The purpose would be to manipulate the coverage of the 900 MHz signals to match the existing 2100 MHz system. Please recall that Lee defined an antenna height factor as [4]: G 20log h = 1 h 2 where h 1 and h 2 are the relative antenna heights. As an example, it can be seen that antennas that are mounted half as high would exhibit a negative gain of 6 db relative to the signals from the higher antenna. Note that this gain (or loss) is independent of frequency. The concept of mounting the UMTS900 antennas lower may be appropriate for tower structures and when reusing cell sites that exist on a spacing defined for 2100 MHz propagation. In this case the spacing of the cells is generally closer than that required at 900 MHz and as a result other cell interference will increase on the 900 MHz network. This will result in a capacity reduction, high levels of soft handover, large neighbor lists and a general increase in the optimization effort required. By mounting the UMTS900 radiation centers lower than the co-located UMTS2100 radiation centers, one can reduce the amount of overshoot and begin to match the coverage of each of the systems. The vertical separation would have the added benefit of increasing isolation between the UMTS2100 and UMTS900 transmitters and receivers. Note that this concept of mounting UMTS900 antenna below UMTS2100 antenna would be applicable in the case of installation on free standing tower rather than roof-top mounting. For a roof-top mounting, the antenna height is limited to the structure height, thus limiting the options to control the propagation of the UMTS900 signal. 2.2.1.1.4 Building Penetration Loss at 900 MHz versus 2100 MHz Radio waves propagate into buildings from external transmitters with some loss of energy. This loss is commonly referred to as the Building Penetration Loss (BPL). The building penetration loss is frequency dependent but also depends on the type of building, the building construction, and the floor level of the building. There are two physical phenomena to be considered with respect to building penetration loss. The first is the BPL itself. This loss is a function of frequency and, for any particular building; it is a function of several factors associated with the size, the construction type and location of the building relative to the transmitter. The second is the fact that building penetration loss has been shown to generally decrease with increasing floor height. Since penetration loss decreases by about 1.4 to 1.5 db per floor with increasing floor level [6], some argue that a conservative design would specify coverage on the ground floor. While coverage to the ground floor is imperative, it needs to be stressed that to have many competing signals to arrive at the higher floors will result in degraded service for CDMA systems on those floors. In general there is a steady decrease in building attenuation as frequency increases. This can be seen in Figure 2-1 for three building types. The building types are as defined in Table 2-3. Figure 2-1 shows that urban buildings produce a stronger rate of decrease with 80-W1044-1 Rev A 2-5

frequency than suburban area buildings. More importantly, the change in BPL with frequency begins to flatten out for all three morphologies in the range of frequencies of interest to this document; the range of 900 MHz to 2100 MHz. Building losses shown in Figure 2-1 do not apply to a specific building, as it depicts the median value over a large group of similar structures [5]. Figure 2-1 Building Attenuation at Ground Level [5] 2-6 80-W1044-1 Rev A

Building Type Urban Medium Residential Description Typically large downtown office and commercial buildings, including enclosed shopping malls. Medium-size office buildings, factories, and small apartment buildings. One- and two-level residential buildings, small commercial and office buildings. Table 2-3 Building Type Definitions [5] Besides BPL, other associated aspects of propagation should also be considered such as the statistical distribution of the signals after they enter the building. General conclusions for the propagation of signals into buildings might be summarized as follows [1]:! Small-scale signal variation is Rayleigh distributed.! Large scale signal variation is log-normally distributed with a standard deviation related to the condition of transmission and the area of the floor.! For non-line-of-sight transmissions, the standard deviation is approximately 4 db.! For partial to complete line-of-sight conditions, the standard deviation increases to 6-9 db. The first two bullets imply that propagation into buildings can be modeled as a combination of fast and slow fading similar to what is common for modeling outdoor macro-cellular propagation. The last two bullets should be considered carefully. For non-line-of-sight propagation into buildings, the standard deviation is less. At first, this may seem counter intuitive until one considers that for the non-line-of-sight case, radio signals are entering the building via local scattering. This implies that signals are coming into the building from many different directions simultaneously. As a result, signal level inside the building does not experience great extremes in fading as indicated by a reduced standard deviation. The line-of-sight case is opposite; having a large standard deviation due to the presence of strong signals coming primarily from one direction followed by fading in the areas that are shadowed by internal structures. 80-W1044-1 Rev A 2-7

* Penetration loss is measured on the first (ground) floor of the building. Table 2-4 BPL as a decreasing function of frequency [1] The data in Table 2-4 shows a 1 to 2 db improvement between building penetration loss at 2300 MHz compared to building penetration loss at 900 MHz. The notes in the table also provide the trend of decreasing building penetration loss at increasing floor height. The data above should be contrasted to that from another study shown in Table 2-5. Table 2-5 BPL as an increasing function of frequency [2] The data in Table 2-5 indicates that building penetration increases with increasing frequency. This is in contrast to the data in the previous table as well as Figure 2-1. The behavior of building penetration with changing frequency is the subject of active debate among researchers. While great pains have been made to measure building penetration loss in a consistent fashion, different trends remain in the literature. The differences may be a function of building size and type, construction material or relative window size. This debate alone provides strong motivation for an operator to measure some representative buildings, using transmitter sites of similar height and configuration, in the market of interest in order to determine how building penetration loss will change at UMTS900 as compared to UMTS2100. It should be pointed out that the measurements for Table 2-5 exclusively considered microcell transmission from a 5 m high transmitter. Their purpose was to quantify building penetration losses at various frequencies to determine the viability of indoor coverage using street microcells with base antenna heights below the roof level of nearby buildings. The residential data in Table 2-5 represents the compiled data from 7 different residential houses in suburban environment. The high-rise data is compiled from 4 urban buildings of different 2-8 80-W1044-1 Rev A

construction types. The purpose in providing this counter example is to stress that any particular building may show different BPL characteristics based on several factors, including frequency, building type, distance from transmitter and antenna height. The COST 231 final report has an excellent discussion of building penetration, its measurement, and additional discussion on how such a broad range of values have been reported in the literature [7]. In summary, some propagation studies on building penetration loss suggest that penetration loss decreases with increasing frequency. What seems clear is that penetration loss will be specific to a particular building or site and it should be acknowledged that building penetration may in some specific cases increase with increasing frequency. Never the less, we assume the trend of a slightly decreasing BPL with frequency as we believe this trend might hold true when considering a large number of buildings for the case of primary interest here: macro cellular deployments where NLOS propagation conditions are predominant. One trend that does seem to hold regardless of frequency band is that as one ascends in a building, penetration loss is reduced about 1 to 2 db per floor. This will tend to hold until one is above the average antenna height.[3] This phenomena is advantageous for better coverage in the building but it also promotes a large number of cells to be visible to the UE in these higher floors. The large number of pilot signals requires many handovers and generally decreases the E c /N o quality inside the building. Specific optimization, in the form of antenna down-tilts and neighbor list maintenance, is often required for suitable performance for these indoor environments. 2.2.1.2 Coverage Comparison Having assumed in the previous section that building penetration loss is slightly greater at 900 MHz than at 2100 MHz, it remains to discuss propagation differences. Here, we show advantages of 900 MHz over 2100 MHz. The physics of electro-magnetic wave propagation immediately tell us that path loss, at least free-space path loss, is less at 900 MHz than at 2100 MHz. This can be seen by recalling the equation for free-space path loss: 4π d LP = 20log λ The path loss, LP, gets smaller with increasing wavelength λ. Using 920 MHz as the midband for UMTS900 and 2045 MHz as the mid-band for UMTS2100, the equation above shows that UMTS900 has a 6.9 db path loss advantage over UMTS2100 in free space conditions. Mobile telephony rarely takes place in free-space (line-of-sight) conditions. For this reason, much work has gone into defining propagation models that can be used to predict mobile radio propagation over a variety of environments. Two of the more popular models are the COST 231 Hata Model and the COST 231 Walfisch-Ikegami Model [7]. Each of these models has certain restrictions of applicability. The Hata model is appropriate for cell sizes between 1 and 20 km, while the Walfisch-Ikegami model is more appropriate for smaller cells and more urban environments. Each of these models is restricted to a maximum of 2000 MHz. We will stretch these models to 2045 MHz for comparison purposes. 80-W1044-1 Rev A 2-9

Using propagation models such as these for comparing coverage between UMTS900 and UMTS2100 does not consider the effect of terrain on propagation. Diffraction over hills and around buildings will be much different at 900MHz than at 2100 MHz. The Walfisch- Ikegami model does attempt to account diffraction over roof-tops but terrain is not considered. Yet the comparisons below should provide the reader with some expectation for the differences in propagation between UMTS900 and UMTS2100 in a mobile radio environment. The original Hata model had the following restrictions: frequency: 150... 1000 MHz h Base : 30... 200 m h Mobile 1... 10 m d: 1... 20 km where hbase is the height of the base station antenna in meters, hmobile is the height of the mobile station in meters and d is the distance from the cell in kilometers. COST 231 extended Hata s original formulation to cover the 1500 to 2000 MHz frequencies. The remaining restrictions remained unchanged. We assume the following typical values for the Hata model parameters and compare the difference in propagation loss at the two frequency bands. frequency: 920 and 2045 MHz h Base : 30 m h Mobile 1.5 m d: 1... 20 km We find that for medium sized cities and suburban centers with medium tree density that UMTS900 will have an 11.4 db propagation advantage. This is illustrated in Figure 2-2 below. 2-10 80-W1044-1 Rev A

Figure 2-2 UMTS900 and UMTS2100 propagation using the COST 231-Hata Model The COST 231 extension to the Hata model includes a 3 db factor for metropolitan centers. That is, the difference between UMTS900 and UMTS2100 propagation becomes 14.4 db in metropolitan centers under the above assumptions. Interestingly, if the h Mobile is changed from 1.5 m, as used above, to 10 m, then the advantage of UMTS900 is reduced from 11.4 db to 8.2 db. This implies that a mobile higher in a building might not see such an advantage in 900 MHz as they did at ground level. The Walfisch-Ikegami model is also defined in the COST 231 final report [7]. This model permits for improved path-loss estimation as it considers more environmental parameters in an attempt to better describe the urban environment. Besides the parameters defined above for the Hata model, the COST 231 Walfisch-Ikegami (COST-WI) model also includes: h Roof : average height of the buildings w : width of the roads b: building separation distance φ: road orientation with respect to the radio path where φ = 90 defines a radio path that is perpendicular to the road. The COST-WI model is restricted to: frequency: 800... 2000 MHz h Base : 4... 50 m h Mobile 1... 3 m d: 0.02... 5 km 80-W1044-1 Rev A 2-11

Immediately we see that the Walfisch-Ikegami model will be more useful for comparing smaller cell sizes, as small as 20 m, than the Hata model. Again, we must extend the useful frequency range a bit beyond the published maximum if we use 2045 MHz. Finally, there are many choices to be made in parameters. It is possible to have h Base < h Roof but we are cautioned that the model has poor performance in this region. For comparison purposes, we assume the following values typical of an urban setting: frequency: 920 and 2045 MHz h Base : 30 m h Mobile 1.5 m d: 0.02... 5 km h Roof : 21 m w : 17.5 m b: 35 m φ: 0 < φ < 90 The above parameters might be representative of a UE at street level in an area of 6 to 7 story buildings. We find that under the above assumptions, using the COST 231 Walfisch-Ikegami model, that UMTS900 has propagation advantage of 11.8 db in a medium sized city and suburban centers with medium tree density and a 15 db advantage in metropolitan centers. (The quotes are from the COST 231 Final Report [7]) These differences are illustrated in Figure 2-3 and Figure 2-4 below. 2-12 80-W1044-1 Rev A

Figure 2-3 UMTS900 and UMTS2100 propagation using the COST 231-WI suburban model Figure 2-4 UMTS900 and UMTS2100 propagation using the COST 231-WI Metro model 80-W1044-1 Rev A 2-13

The above differences hold for all angles of φ greater than approximately 15. For angles of φ smaller than about 15 and for short distances from the cell, the difference in propagation loss between UMTS900 and UMTS2100 become smaller. This seems intuitively sound since small angles of φ imply propagation more along the street canyons. This, in combination with being closer to the cell implies a more line-of-sight propagation environment so that free space propagation difference of 6.9 db would be approached. In summary, we have compared UMTS900 and UMTS2100 propagation using two widely known propagation models as well as the free-space propagation equation. We find that, at a minimum, UMTS900 has a 6.9 db propagation advantage under line-of-sight propagation conditions. For larger cells, the COST 231 Hata model indicates that UMTS900 would have an advantage of 11.4 to 14.4 db in suburbia to metropolitan centers, respectively. The COST-WI model predicts that UMTS900 would have a propagation advantage of 11.8 to 15 db, again, in suburbia to metropolitan centers, respectively. While these models do not consider terrain effects specifically, they do provide some indication of the relative advantage of 900 MHz propagation over 2100 MHz propagation. 2.2.1.3 Capacity Comparison For equivalent conditions, equivalent network topologies, the cell capacity of UMTS900 and UMTS2100 would be the same. A frequency change, by itself, would not affect cell capacity. Differences in cell capacity will occur with UMTS900 for primarily two reasons. The first concerns interference from GSM900, which may share the spectrum adjacent to the UMTS900 signal(s). The amount of interference and level of capacity degradation will depend on the guard band chosen and the amount of interference that may be present from GSM900 as a function of network topology. If the GSM900 network is deployed in a fashion that is un-coordinated with the UMTS900 network, then the detrimental impact to network capacity (both networks) will be more extreme. The second source of network capacity degradation for UMTS900 would come from a reduction in per cell capacity as a result of cell spacing that is too close. This might occur if a UMTS900 network is deployed using the same network topology (cell site spacing and antenna heights) as deployed for UMTS2100. As indicated in Section 2.2.1.2 above, UMTS900 could enjoy as much as an 11 to 15 db propagation advantage over UMTS2100. If the UMTS900 cells are spaced according to a UMTS2100 network plan, then the amount of other cell interference as well as the amount of soft-handover will increase in the UMTS900 network relative to the UMTS2100 network. In this latter issue, we must be careful to distinguish between per cell capacity and overall network capacity. If the UMTS900 network is overbuilt along the UMTS2100 network cell site locations, it may be that the overall network capacity is still large. Each individual UMTS900 cell may carry less traffic but if there is a great number of them, then the overall network may have acceptable capacity. Of course, this capacity comes at an increased capital expenditure and a greater amount of network optimization effort. 2.2.2 Link Budget Comparison (UMTS900-UMTS2100) Many components of the Link Budgets for the two frequency bands are the same:! Data Rate (e.g., 12.2 kbs for AMR speech)! Mobile Transmit Power (Same classes defined for both frequencies) 2-14 80-W1044-1 Rev A

! Node B PA Power! Shadowing Margin! Soft Handover Gain! Required E b /N t Link budgets for two representative scenarios are presented below. The primary link budget items are equivalent in the two scenarios except as follows. In scenario 1 (Table 2-6, Table 2-7, and Table 2-8), the antenna gains are assumed to be the same at 16.5 dbi. Note that this implies an approximately 1 m long antenna at 2100 MHz and an approximately 2.3 m long antenna at 900 MHz. We also assume the same size coaxial cable and equivalent cable run lengths so that there is an estimated 0.6 db differential in the cable losses. In scenario 2 (Table 2-9, Table 2-10, and Table 2-11), we assume the same size antenna with a consistent horizontal beamwidth. This represents the case in which one deploys a dual or triple band antenna structure. In this case, the differential in antenna gain is assumed to be 2.5 db. The antenna gains used are representative of 1.3 m long panel antenna. These scenarios are each calculated for three different morphologies: urban, medium, residential. The urban, medium, and residential morphologies are as defined in Table 2-3. The main difference in these morphologies concerns the assumed building penetration loss. The residential morphology is also applicable to rural deployments since we will assume the same building penetration loss for each of these. 80-W1044-1 Rev A 2-15

Table 2-6 Uplink Budget for Scenario 1, Urban Table 2-7 Uplink Budget for Scenario 1, Medium Please note that since line items a-j are the same as in Table 2-6, these were not repeated in Table 2-7. 2-16 80-W1044-1 Rev A

Table 2-8 Uplink Budget for Scenario 1, Residential Please note that since line items a-j are the same as in Table 2-6, these were not repeated in Table 2-8. Table 2-9 Uplink Budget for Scenario 2, Urban Table 2-10 Uplink Budget for Scenario 2, Medium 80-W1044-1 Rev A 2-17

Please note that since line items a-j are the same as in Table 2-9, these were not repeated in Table 2-10. Table 2-11 Uplink Budget for Scenario 2, Residential Please note that since line items a-j are the same as in Table 2-9, these were not repeated in Table 2-11. To summarize the link budget differences between 900 MHz and 2100 MHz, please consider Table 2-12 below: MAPL Scenario 1 Scenario 2 900 MHz 2100 MHz Delta MAPL [db] Urban 127.1 128.5 1.4 Medium 131.1 131.5 0.4 Residential 135.1 135.5 0.4 Urban 124.6 128.5 3.9 Medium 128.6 131.5 2.9 Residential 132.6 135.5 2.9 Table 2-12 Summary of Uplink Budget Maximum Allowable Path Loss Differences Considering Table 2-12, it can be seen that 2100 MHz always has a link budget advantage. This difference is greater in scenario 2, where we constrain the antennas to be the same physical size. It might appear, from only the link budget, that 2100 MHz has a coverage advantage over 900 MHz. The reader is asked to recall the results of Section 2.2.1 where it is shown that, at a minimum, UMTS900 has a 6.9 db propagation advantage under line-of-sight propagation conditions. The propagation advantage was greater for larger cells. The COST 231 Hata model predicts that UMTS900 would have an advantage of between 11.4 to 14.4 db in suburban to metropolitan centers, respectively. The COST-WI model predicts that UMTS900 would have a propagation advantage of 11.8 to 15 db, again, in suburban to metropolitan centers, respectively. Putting these values together with the link budget comparisons above, it is clear that UMTS900 should out perform UMTS2100 coverage in all cases and morphologies. 2-18 80-W1044-1 Rev A

3. UMTS900 Deployment Guidelines 3.1 Deployment Scenarios Better propagation and in-building coverage make UMTS900 deployments attractive in rural, suburban, and urban morphologies. Initial deployments are expected to be in rural areas mainly to increase the UMTS footprint with less cost. Urban deployments will follow later to provide better in-building coverage and to fill in coverage holes in the UMTS2100 network with optimized inter-frequency (inter-band) handovers and reselections. UMTS900 can also be used to share the load of UMTS2100 network. UMTS900 network can be designed in one-to-one overlay with GSM900 network where UMTS900 and GSM900 cells co-site. The sites can share the same antenna or have different antennas with different height, azimuth, etc. This is called the coordinated operation and expected to exist mostly in rural deployments. In urban morphologies, it is rather difficult to have fully coordinated deployment as GSM900 cell density can be very high and GSM900 micro/pico cell deployments are common. In most networks, GSM900 provides blanket coverage, i.e., countrywide, and UMTS2100 exists in islands of varying size covering mostly urban areas. UMTS900 networks can coexist with UMTS2100 in the coverage islands and/or can be deployed as an augmentation to UMTS2100 as shown in Figure 3-1. Deploying UMTS900 on top of UMTS2100 requires a separate network plan as the propagation and the required cell count are different. Figure 3-1 UMTS2100, UMTS900, GSM900 Deployments The carriers used for UMTS900 can be re-used by GSM900 in the areas where UMTS900 is not offered. An isolation distance is required between these GSM900 and UMTS900 sites to 80-W1044-1 Rev A 3-1

avoid the interference as shown in Figure 3-2. Carrier assignments in Figure 3.2, f1-f4, are for representation purposes. GSM900 (f1,f2,f3,f4) Isolation distance UMTS900 (f3,f4) UMTS900 UMTS2100 (f3,f4) GSM900 (f1,f2) UMTS2100 Figure 3-2 Isolation distance between UMTS900 and GSM900 UMTS900 deployments will require User Equipment (UE) and UTRAN equipment to support the additional mobility between UMTS900 and UMTS2100 and mobility between UMTS900 and other GSM bands. UMTS900 equipment shall support the equivalent functionalities to UMTS2100 equipment. Main challenge in deploying UMTS in band VIII (UMTS900) is the interference with GSM900 due to co-existence in the same band. Interference can be overcome by allowing sufficient guard band between carriers, optimizing frequency planning, and deploying special filters. Five co-existence scenarios have been identified and studied in 3GPP [12]:! Scenario 1: UMTS(macro)-GSM(macro) in Urban area with cell range of 500 m in uncoordinated operation! Scenario 2: UMTS(macro)-GSM(macro) in Rural area with cell range of 5000 m in uncoordinated operation! Scenario 3: UMTS(macro)-GSM(macro) in Rural area with cell range of 5000 m in coordinated operation! Scenario 4: UMTS(macro)-GSM(micro) in Urban area in uncoordinated operation! Scenario 5: UMTS(macro)-GSM(pico) in Urban area in uncoordinated operation 3.1.1 Rural, Urban Deployments & Objectives Deployment objectives of UMTS900 vary for rural and urban morphologies. The main objective in rural deployments is to augment coverage and improve 3G footprint. In urban 3-2 80-W1044-1 Rev A

deployments, objectives are multiple: improve in-building coverage, fill in UMTS2100 coverage holes, and load sharing. 3.1.1.1 Rural Deployments UMTS2100 networks mostly cover the city centers and some towns. UMTS900 can be used to extend the footprint of UMTS further into rural areas as shown in Figure 3-3. Figure 3-3 UMTS900 deployments in rural areas UMTS900 has significant coverage advantage over UMTS2100 in rural deployments. For the same service and data rate, the cell range is approximately doubled with UMTS900. This makes UMTS900 ideal for rural deployments. Section 2.2 shows the coverage comparison for different configurations. Several operators may deploy only UMTS900 in rural areas due to cost efficiency and lack of demand for both UMTS900 and UMTS2100. UMTS900 co-exist with GSM900 and the cell sites of both technologies can share the same site with separate or common antenna. This will likely be the common configuration due to cost efficiency and similarity in network plan of UMTS900 and GSM900. Despite the same propagation characteristics in 900 MHz band, GSM and UMTS have link budget difference of around 6 db when voice service is considered. Coordinated deployment of UMTS900 for voice may result in overdesign, but not for high capacity demand services (video telephony, PS data) when UMTS900 service coverage is compared to GSM voice coverage. This coordinated mode operation helps in reducing adjacent channel interference between UMTS and GSM and hence may require less spectrum clearance. The mobility features between UMTS900 and GSM900 shall be supported by UE and UTRAN. GMS900 can be used for load sharing when UMTS900 becomes overloaded. 3.1.1.2 Urban Deployments Goals in deploying UMTS900 in urban morphologies vary: 80-W1044-1 Rev A 3-3

! Improve indoor coverage! Filling in coverage holes! 3G load sharing UMTS900 can be deployed as a blanket to UMTS2100 or can be deployed only in spots as shown in Figure 3-4. Figure 3-4 UMTS900 deployments in urban areas As discussed in Section 2.2, better propagation in 900 MHz band improves the street level and in-building coverage. However, this propagation advantage can become a disadvantage in urban morphologies where the cell density is high. High cell overlap in these scenarios increases the interference and handover percentage which are both detrimental to capacity. UMTS900 network plan shall address this aspect. Following the same discussion, one-to one deployment of UMTS900 with GSM900 and UMTS2100 is not recommended. While high interference is not an issue in GSM900 and UMTS2100 due to frequency planning and higher propagation loss, respectively, it can definitely be a concern in UMTS900 if same cell density is used. The above statement assumes UMTS2100 network coverage is adequate. In some networks, it is observed that UMTS2100 cell density is not enough since the design is based on GSM900. One-to-one overlay of UMTS900 with UMTS2100 can be acceptable in these networks. UMTS900 and GSM900 networks can share some site locations, but it is unlikely to have a fully coordinated network in urban deployments. This is due to high density of macro cells and common use of micro/pico cells in GSM900. This brings the complications of uncoordinated operation: adjacent channel interference and guard band requirements. UMTS900 will likely co-exist with UMTS2100, GSM900, and other GSM bands. Several mobility transitions, inter-band and inter-rat, are possible. Based on the design objectives, only subset of them can be configured. For example, if UMTS2100 is considered as the main UMTS layer, mobility from GSM900 to UMTS2100 is configured, and the mobility 3-4 80-W1044-1 Rev A

from GSM900 to UMTS900 can be omitted. Mobility between UMTS900 and UMTS2100 are configured for coverage continuity and load sharing. Typically one UMTS layer prioritized over the other based on the design objectives such as provide better in building coverage vs. limit handovers, availability of dual band UEs. UMTS2100 is expected to be the main band for a while just due to availability multi-band UEs. Seeding the market with UEs supporting UMTS900 is crucial to benefit from UMTS900. Most UE vendors will soon have products supporting UMTS900, but it is expected to take a few years until all units support UMTS900. 3.1.2 Coordinated vs. Uncoordinated Deployments Coordinated operation requires one-to-one overlay of UMTS900 and GSM900 sites. In uncoordinated operation, all sites or subset of them don t co-locate and their locations are independent of other system s sites. The uplink (UL) and downlink (DL) of GSM900 and UMTS900 are adjacent to each other. So, the main interference to consider is between the base station of one system to the mobile station of the other system (see Figure 3-5), in a near-far scenario. The near-far scenario is a characteristic of uncoordinated system when a mobile station of a system transmits at full power near the base station of the other system, or a base station transmits at full power near a mobile station of the other system. UMTS Node B/GSM BTS impacts GSM MS/UMTS UE on DL and UMTS UE/GSM MS impacts GSM BTS/UMTS Node B on the UL. Interference is not significant in the coordinated case since the wanted signal is also strong when the offending site is strong. On the other hand, in uncoordinated operation, UMTS UE/GSM MS can transmit maximum power near GSM BTS/UMTS Node B. The same applies for DL. 80-W1044-1 Rev A 3-5

Figure 3-5 Interference Scenarios 3.1.2.1 Coordinated Operation Site sharing with GSM900 will be cost efficient to operators and will be favorable as long as network plan for UMTS900 allows co-location. It comes with no big disadvantage in rural morphologies as GSM900 and UMTS900 networks are expected to have similar footprint without causing high interference in UMTS. The same may not hold for urban deployments, the interference issue due to high cell density can be overcome with frequency planning in GSM, but the problem remains in UMTS. Adjacent channel interference is less significant in coordinated operation since UMTS900 and GSM900 sites are co-located as shown in Figure 3-6. An MS/UE close to interferer site is also close to the target site. For example, a UMTS UE close to GSM BTS is also close to UMTS Node B and power controlled, so the interference from UMTS UE to GSM BTS will not be significant. Similarly on DL, since UMTS UE is close to both UMTS Node B and GSM BTS, the interference coming from GSM BTS will not be dominant. 3-6 80-W1044-1 Rev A

UMTS GSM Inter-site distance 3*R Cell radius R Cell range 2*R Figure 3-6 Coordinated operation Fully coordinated operation requires isolation with other operators carriers as well. A non co-located cell of another operator on an adjacent channel can be a significant interference source. Isolation between operators can be provided by deploying UMTS900 carriers in between GSM900 carriers of the same operator, also called sandwich type deployment, as shown in Figure 3-7. GSM900 UMTS900 GSM900 Figure 3-7 Sandwich type deployment Co-locating UMTS900 and GSM900 cells can be in several different configurations. Section 3.2 looks into this in detail. 2.6 MHz carrier to carrier spacing is recommended between a UMTS900 carrier and a GSM900 carrier in coordinated operation [13]. Lab test results and simulations show that carrier separation can be as low as 2.2 MHz in the coordinated operation. In case of carrier spacing less than 2.6 MHz, the GSM carriers overlapping with 2.6 MHz offset from the UMTS carrier are recommended to have no BCCH to help improve narrow band blocking performance of UMTS UE and Node B. 80-W1044-1 Rev A 3-7

3.1.2.2 Uncoordinated Operation Uncoordinated operation is expected to be common in urban areas. The locations of UMTS900 and GSM900 sites can be off as extreme as being at the cell edge of each other as shown in Figure 3-8. UMTS GSM Inter-site distance 3*R Cell radius R Cell range 2*R Figure 3-8 Uncoordinated operation The interference gets worst when GSM BTS is located at the cell edge of UMTS Node B and UMTS UE/GSM MS operates next to GSM BTS/UMTS Node B. Victim system and link can be summarized in the following scenarios:! UMTS900 UL victim: GSM MS operates near UMTS Node B! GSM900 UL victim: UMTS UE operates near GSM BTS! UMTS900 DL victim: UMTS UE operates near GSM BTS! GSM900 DL victim: GSM MS operates near UMTS Node B In all these cases, the wanted signal strength is close to sensitivity level and the interference source is significant. Adjacent Channel Leakage Ratio (ACLR) and Adjacent Channel Selectivity (ACS) are the two contributing factors to adjacent channel interference. ACLR indicates transmitter emissions and is the ratio of the RRC filtered mean power centered on the assigned channel frequency to the RRC filtered mean power centered on an adjacent channel frequency. ACS reflects the receiver selectivity and is the ratio of the receive filter attenuation on the assigned channel frequency to the receive filter attenuation on the adjacent channel. The dominant factor varies per scenario:! GSM MS to UMTS Node B: Both ACLR and ACS are dominant! UMTS UE to GSM BTS: UMTS UE ACLR dominant! GSM BTS to UMTS UE: UMTS UE ACS dominant! UMTS Node B to GSM MS: UMTS Node B ACLR dominant 3-8 80-W1044-1 Rev A

Guard band requirement is more stringent for the uncoordinated operation as the adjacent channel interference can get significant for certain scenarios. 3GPP recommends a carrier separation of 2.8 MHz for uncoordinated operation [12]. 3GPP simulations which use the network layout in Figure 3-8 and the ACLR and ACS specifications show that the impact of GSM900 interference on UMTS900 capacity is less than 5% and the impact of UMTS900 interference on GSM900 is negligible with 2.8 MHz carrier to carrier separation [12], [8], [14]. Adjacent channel interference can get more severe in GSM micro/pico cell and UMTS UE operation. Typical minimum coupling loss (MCL) between a micro cell and a UE is much less than the MCL between a macro cell and a UE (by 30 to 40 db). While 2.8 MHz provides sufficient isolation in macro cell environment, more frequency separation is required in GSM micro/pico cell and UMTS UE operation. A recommended scheme is to separate GSM micro cell sub-band and UMTS macro cell sub-band with GSM macro cell sub-band, as shown in Figure 3-9. Figure 3-9 Recommended frequency allocation for GSM micro/pico cells Transmitter emission mask and receiver blocking requirements specified in the standards along with 2.8 MHz carrier separation avoid interference between UMTS900 and GSM900 in most uncoordinated operation scenarios. For the few cases where required isolation can not be obtained, additional transmitter and/or receiver filtering can be used at the particular cell sites. 3.2 Design Guidelines for Multi-band and Multitechnology Sites Many operators who are in the process of expanding or rolling out a new network find it difficult to obtain new sites for radio base stations. In many countries, operators rolling out WCDMA networks expect to deploy up to 80% of the new system alongside the existing GSM equipment. Co-location solutions enable operators to reuse existing equipment and to reduce CAPEX. This type of deployment could be considered as the best case scenario when 80-W1044-1 Rev A 3-9

the mutual interference between mobiles and base stations of two technologies are minimized. To fulfill this low interference requirements special care shall be applied in order to prevent the receivers of the BSs being desensitized by emissions from a BS transmitter. Power of any spurious emission in BTS Rx band (880 915 MHz) shall not exceed the -96 dbm/100 KHz [8]. This requirement assumes 30 db isolation between the transmitter and the receiver of the same class base stations. Additional isolation will be required for colocated BTS of different classes (e.g., Macro GSM BTS Micro UMTS Node B). This section considers co-location of macro GSM BTS with macro UMTS Node B only. 3.2.1 Site Configurations for Co-located GSM and UMTS Co-location of two sites of different technologies could be implemented by either antenna sharing or real estate sharing. In first case the same antenna is employed by the base stations of both technologies. In second case the space on the rooftop or tower is shared between GSM and new installed UMTS antennas. Possible configurations to deploy UMTS900 system on the existing GSM900 site are shown in Figure 3-10. Figure 3-10 Possible Deployment Configurations for Co-Located GSM & UMTS Systems 3-10 80-W1044-1 Rev A

3.2.1.1 Configuration 1 Configuration 1 is simple sharing of real estate. Using established 2G sites and associated civil structures to support mechanically and electrically separate antenna/feeder/node B assemblies for new UMTS system. The benefits of such approach are electrically isolated feeders and antennas which ensure the highest level of RF isolation (unless antennas of different technologies are facing each other) between the co-located systems. Also, this approach permits independent optimization of the co-located networks careful selection of the feeder diameter for each frequency range can aid in network optimization, without creating any disturbance to the co-located network. Another benefit will be the absence of diplexers to combine signals and then to split them if one common feeder would be used, thus reduced loss and risk of intermodulation. The cons of real estate sharing are the deployment cost and required real estate space. 3.2.1.2 Configuration 2 Dual and triple band functionality from a single antenna housing, while using mechanically and electrically separate feeders, Base Transceiver Stations (BTS) and Node Bs. The pros of this configuration are conserved space on antenna tower or rooftop which could be the critical factor when new UMTS system is deployed. The con is that this solution might affect existing GSM cell coverage, or limit the UMTS optimization, since the antenna direction is the same for each system. To overcome this optimization issue, when shared antennas are used, they should be of a type that allows independent control of electrical downtilt for each band. 3.2.1.3 Configuration 3 Mechanically and electrically separate antennas, supported by a single shared feeder system and separate Node B/BTS. Operator might benefit from feeder-sharing when the distance between the base station and the antenna is large and cost of deployment is more important than radio performance. This configuration is also useful when space limitations prevent operators from installing more feeders at a site. The negative side of such approach is necessity of two diplexers for each feeder in the sector: One close to the base station, and another close to the antenna. The use of diplexers results in a slight loss of output power from the base stations. An insertion loss between 0.2 and 0.7 db per diplexer should be compensated by either higher antenna gain or additional PA power. Some reduction in isolation between Node B and GSM BTS is also possible. 3.2.1.4 Configuration 4 Configuration 4 includes antennas supporting broadband functionality, which are able to accommodate a number of cellular services within this bandwidth, implying common feeder and hybrid BTS/Node B. The benefits of such configuration are reduced cost of deployment, lesser real estate space requirements. The cons are limited flexibility in operation and maintenance of hybrid site, inability to independently control the coverage of GSM and UMTS system, thus possibly leading to the need to re-plan the GSM system. Deployment configuration 1 assumes separate feeder for each antenna, whereas configuration 3 employs feeder sharing. When RF performance is considered, separate feeders are always preferred. Feeder sharing solutions have greater loss and increased risk of intermodulation, and require higher voltage standing-wave ratio (VSWR) limits. 80-W1044-1 Rev A 3-11

3.2.2 GSM-UMTS Antenna Sharing Deployment (Config. 2 & 4) Depending on the requirements, there are several ways of co-locating GSM and UMTS antenna systems. The simplest method is to share antennas, replacing existing antennas with cross-polarized multi-band antennas. This solution might affect existing GSM cell coverage, since the antenna direction is the same for each system. In the same time having two or three antennas in a single housing is an ideal solution for operators who need to minimize the visual impact of the site or have limited real estate recourses for antenna system expansion. There are currently numbers of high quality multi-band base station antennas on the market, suitable for antenna sharing deployment. As an example, the specification for such antenna is shown in Table 3-1. Antenna is capable of supporting 2G (GSM900, GSM1800) and 3G (UMTS900, UMTS2100). Frequency range 806-960 1710-1880 1920-2170 (MHz) 806-866 824-894 880-960 1710-1880 1920-2170 Gain (dbi) 2x16 2x16 2x16.3 2x17.5 2x18 Polarization (deg) 45, -45 45, -45 45, -45 45, -45 45, -45 Half-power beam width (deg) Horizontal: 69 Vertical: 7 Horizontal: 68 Vertical: 10.7 Horizontal: 67 Vertical: 9.8 Horizontal: 65 Vertical: 5.1 Horizontal: 65 Vertical: 4.8 Electrical Tilt Continuously adjustable (deg) Isolation Intrasystem (db) 0-10 0-10 0-10 0-6 0-6 >30 >30 >30 >30 >30 Table 3-1 Triple Band Antenna Key Parameters Antenna has 1.7 db less gain in 900 MHz band if compared with 2100 MHz, however better propagation at 900 MHz should offset the gain difference. Antenna isolation between two 900 MHz ports is at least 30 db which should be sufficient for co-located GSM900 and UMTS900 systems. Notice that the vertical beamwidth of antenna in 900 MHz band is significantly larger than that at 2100 MHz, which in turn might affect the UMTS 900 optimization. Coverage of each GSM / UMTS antenna could be adjusted individually by tuning electrical down tilt if needed. Feeder-sharing and antenna-sharing can be used in combination. In this case, the diplexer closest to the antenna is either an external device or built into the antenna. The combination can also be applied when co-locating GSM/UMTS900, GSM1800, and UMTS2100. However, to combine signals from three systems, triplexers are used instead of diplexers. The alternative solution in antenna sharing would be re-using of existing GSM antennas. This approach will reduce cost of UMTS900 deployment The negative part of re-using existing antennas would be larger beamwidth and reduced gain of old antennas, which might impact coverage of new UMTS systems. 3.2.3 Real Estate Sharing Deployment (Config. 1 & 3) For cost saving operations and/or real estate limitations, the UMTS900 Node B antennas could be co-located with existing GSM BTS antennas by sharing the same structure 3-12 80-W1044-1 Rev A

(monopole, tower) or the same rooftop. The requirements for co-location of base stations assume 30 db minimum coupling loss between base stations of the same class (Macro UMTS900 Macro GSM900 in this case) [9]. These requirements shall be applied for the protection of other BS receivers when GSM900 BTS is co-located with a UMTS900 Node B. The antenna to antenna isolation measurement results were presented by several reports [10], [11]. 30 db of isolation is typical to many existing installations, as reported by several operators. QUALCOMM has conducted series of antenna isolation measurements for the cases when two antennas are located side-by-side, (horizontal spacing) or end-to-end (vertical separation). Table 3-2 shows antenna type utilized for isolation measurements and key parameters. Horizontal beamwidth Max Gain Polarization Frequency band 65º 18 dbi +/- 45 o 870-960 MHz Table 3-2 Antenna Parameters Figure 3-11 Horizontal Spacing and Figure 3-12 Vertical Spacing illustrate geometry of the measurements. Figure 3-11 Horizontal Spacing Figure 3-12 Vertical Spacing Antenna isolation measurement results are presented in Table 3-3. Antenna spacing is shown in inches. Measurement data presented in table concludes that minimum 30 db of electrical 80-W1044-1 Rev A 3-13