Dokumentnamn. Document - Ref PTS-ER-2004:32

Transcription

1 1 (18) 1 SUMMARY The purpose of this report is to find out how the signal requirement matches the service requirement in a UMTS network, and to comment on some of the specific issues raised by the Swedish Operators in a request for changed requirements. The key findings from the report are: UMTS coverage is in the majority of the situations limited by the uplink while the coverage requirements for the UMTS License holders in Sweden is set on the downlink. This means that the licence criteria can not be evaluated / determined without making some assumptions on how base stations are configured. The aspects that have most impact on the signal requirement are: - building penetration loss, and it s variation in different types of buildings - the use or non-use of tower mounted low noise amplifiers on the uplink - the interference level in the network The link budget calculations show that current license requirement of 58 dbuv / m measured on the primary common pilot channel is a too low requirement in city centres, while it is too high in rural areas. Suggested range is from 65 dbuv / m in dense city centres to 50 dbuv / m in rural areas. Buildings in city centres are normally big constructions of concrete or brick, while habituated buildings in rural areas are primarily single family houses constructed out of wood. This difference in construction means a difference in penetration loss, which motivates a 9 db difference in signal strength requirement to achieve the same service level in both environments. Differences in uplink interference level in high and low traffic areas can motivate an additional 2 db difference between dense urban and rural areas. Coverage in areas with low traffic and therefore low interference will be limited by the uplink performance, and the use of tower mounted amplifiers can improve the uplink performance by 4-5 db. If tower mounted amplifiers are used in rural areas and not in urban areas, the additional difference in signal requirement to achieve same service level should be approximately 4 db. 50 dbuv / m should be a sufficient coverage requirement in low traffic rural areas if tower mounted low noise amplifiers are used on the base stations and no more than 10% of the total power is allocated to the primary common pilot channel channel.

2 2 (18) TABLE OF CONTENTS 1 SUMMARY TABLE OF CONTENTS INTRODUCTION BACKGROUND PURPOSE RESERVATIONS LINK BUDGETS FOR UMTS PURPOSE OF THE LINK BUDGET FUNDAMENTAL DIFFERENCE BETWEEN LINK BUDGETS FOR TDMA AND CDMA NETWORKS BASIC UPLINK POWER BUDGET FACTORS AFFECTING THE LINK BUDGET CURRENT LICENSE REQUIREMENTS AND HOW IT RELATES TO THE LINK BUDGET ABOVE COMMENTS ON OPERATOR PROPOSAL CHANGED SERVICE THRESHOLD CHANGED PROBABILITY MEASUREMENTS IMPLICATIONS OF CHANGED REQUIREMENTS OPERATOR INVESTMENT REFERENCES... 18

3 3 (18) 2 INTRODUCTION 2.1 Background UMTS license conditions[1] in Sweden include a requirement on the operators to provide a measured field strength on the primary common pilot channel of 58 dbuv / m with 95 % area probability. In a clarification to the signal strength requirement it is stated that it corresponds to service bit rates between 144 and 384 kbit/s in an indoor environment. 2.2 Purpose The purpose of this report is to find out how the signal requirement matches the service requirement, and to comment on some of the specific issues raised by the operators[5]. 2.3 Reservations In order to calculate the required signal strengths for the different services and different environment, a link budget is presented. Link budgets always makes a number of assumptions, and engineers can gladly spend hours debating individual values. UMTS technology is still not very mature, and some early assumptions may still prove wrong in the real life. In this report we have tried to use values that seem average in the literature, so that any deviations in one value that an engineer will raise would hopefully equal out the claims for changes of another value by another engineer. Still, one should regard the final figures with a +/- 2dB uncertainty.

4 4 (18) 3 LINK BUDGETS FOR UMTS 3.1 Purpose of the link budget The purpose of the link budget is to calculate the maximum path loss allowed between the base station and the mobile for a given service. The maximum path loss is then used to choose antennas and configurations for the base stations, and for planning of output power on the different channels of the base stations. The maximum path loss determines the range of each base station and thus the base station density, and number base stations required to cover a given area. Another important purpose of the link budget is to balance the up and downlink. Very high output power of the base station is of little use if the base station can not hear the mobile station. 3.2 Fundamental difference between link budgets for TDMA and CDMA networks In 2G TDMA networks, such as GSM, all radio resources on a given frequency are dedicated to a certain mobile at any given point in time. It is thus fairly straight forward to calculate any imbalances between up and downlink, and to make sure the link is balanced. In CDMA systems on the other hand, base station output power is shared between all connections on the base stations, so the amount of power available for a certain connection will vary with load and positions of the connected mobiles. Since the same frequency is used on all cells, the link budget also needs to include margins for the interference created by other connections, both from the own cell and other cells. This interference is often treated as an additional noise that is added to the thermal noise. Load on the CDMA system may also be asymmetric, as some of the packet data services, such as web browsing, will likely generate more downlink than uplink data. In CDMA networks, the uplink is normally regarded as coverage limited, while the downlink is interference limited as a result of the load on the network. The basis for this is that the base station has typically W (40-46 dbm) output power available, while the mobile unit has W (21-24 dbm). This means that in low traffic situations, the uplink is the limiting link, while in high traffic situations, downlink becomes the limiting link. This gets even more complicated when a mobile is indoors, since penetration loss will also attenuate the interference on the downlink, but not on the uplink. Existing literature often ignores the downlink altogether when discussing the link budget for planning purposes. In this report we will assume uplink being the limiting link for coverage purpose, but also check that a sufficient interference margin is available on the downlink, under the assumption that no more than a certain percentage of the power is allocated to a single user.

5 5 (18) 3.3 Basic uplink power budget The following example link budget is taken from the book WCDMA for UMTS [2], page 158. This link budget is not directly applicable to the requirements in Sweden, but serves as a starting point to explain the general maths of the link budget calculation, and establish a baseline that is commonly available in the literature. Deviations from this link budget will be explained throughout the report. Transmitter (mobile) Max mobile transmit power [w] [w] 0,25 As above in dbm [dbm] 24 a Mobile Antenna Gain 2 b Body Loss 0 c EIRP [dbm] 26 d = a + b - c Receiver (base station) Thermal Noise Density -174 e Base station receiver noise figure 5 f Receiver noise density -169 g = e + f Receive noise power -103,2 h = g + 10*log( ) Interference margin 3 i Receiver interference power -103,2 j = 10*log(10^((h+i)/10-10^(h/10)) Total effective noise + interference -100,2 k = 10*log(10^(h/10)+10^(j/10)) Processing Gain 14,3 l = 10*log(3840/144) Required Eb / N0 1,5 m Receive sensitivtity -112,9 n = m-l+k Base station antenna gain 18 o Cable loss in the base station 2 p Fast fading margin 4 q Max Path Loss 150,9 r = d - n + o - p - q Coverage probability 80% Log Normal fading constant 12 Propagation model exponent 3,52 Log Normal fading margin 4,2 s Soft handover gain 2 t Indoor Loss 15 u Allowed propagation loss for cell range 133,7 v = r - s + t - u Already at this point it is important to point out that link budgets in literature normally includes a margin for the log normal fading, which can be described statistically, to arrive at a maximum path loss that can be used for radio planning purposes. When comparing the above link budget with the license requirements, it is important to understand that the margin for the statistical variation of the measured signal in the outdoor environment is already

6 6 (18) taken care of in the license conditions[1] ( with 95 % area probability ), but that the license condition indirectly also makes room for Building Penetration loss. Also, a coverage probability of 80 % is widely regarded as a very low value. Thus, the path loss to use when discussing the license conditions is to be found somewhere between Max Path Loss and Allowed path loss for cell range. Later we will split the link budget up to find the path losses that can be used in conjunction with the license conditions. It can also be noted that the above link budget assumes a data terminal with 24 dbm output power, which is not readily available on the market. An antenna gain of 2 db is also a very optimistic value for a normal voice primarily handset. 3.4 Factors affecting the link budget Services and service mix A higher bit rate service requires a higher received power than a lower bitrate. This is included in the processing gain, line l above. A higher bit rate service require a lower E b /N 0 than a low bit rate service. This is due to a fixed amount of overhead for control channels, regardless of the service bit rate. (A in detail description can be found in chapter in [2]). The E b /N 0 values assumed normally includes the diversity gain, and thus the required E b /N 0 is normally lower in the uplink than downlink. License conditions in Sweden mention bit rates from kbit/s indoors. In this report we assume - Minimum CS 128 on the uplink for good video quality or PS 144 for uploads. - Minimum PS 384 on the downlink for fast web browsing and downloads As uplink is normally the coverage limiting link, UL 144 kbit/s will be the dimensioning criteria.

7 7 (18) Building Penetration loss Mobile phones and data terminals are used in a variety of environments, but to a very large degree they are being used indoors, the signal thus being attenuated as it has to propagate through the walls or windows of the building where the user is. The link budget needs to include a margin for the penetration loss in case service is planned for indoor users. Building attenuation varies greatly between buildings and is affected by Building material Wall thickness The amount of windows Presence of sun reflective shielding on windows How deep into the building the user is Whether the building itself has Line Of Sight (LOS) with the serving base station Angle of incidence of the incoming signal On which storey of a building the receiver is located It is difficult to find any conclusive scientific research on this subject, results vary greatly between different studies, and no research covers all aspects listed above. The COST 231 final report [3] attempts to summarize all studies within it s framework to 3 equations to be used in 3 different scenarios 1. micro cellular 2. macro cellular with line of sight between base station and building 3. macro cellular with non line of sight between base station and building The third scenario seems most applicable when discussing signal strength on cell boundaries where the license conditions apply. The formula in this case is: L = L outside + W e + W ge + a * d - G fh Where L is the total path loss L Outside is the path loss between the base station and a point at ground level just outside the building W e is the attenuation in the external wall W ge is a compensation factor for non perpendicular radio wave arrival D is the how far away from the wall a receiver is A is a constant, 0.6 db / m is recommended G fh is the floor gain, i.e. the signal will normally be higher on higher floors Recommended values are We : 4-10 db (4 db for wood, 7 db for concrete walls) Wge : 3-5 db at 900 MHz, 5-7 db at 1800 MHz

8 8 (18) It is evident that a single penetration loss value will not apply to all environments. A small building normally has thinner walls than a large building. A small building also has windows in different directions, thus giving a lower penetration loss. In Sweden single family house are mainly constructed out of wood, while multi family and multistory buildings are normally made of concrete. Even if not all single family houses are constructed of wood, one should understand that the measurements conducted by COST members to arrive at a value specific for wood, have been carried out in small buildings (since no large buildings are made of wood). Hence, the wood value is to some extent applicable not only to wooden buildings, but also to small buildings in general. The following distinction is suggested: 1. Single family houses (treated as wood) 2. Radhus (treated as wood) 3. Houses with more than 2 storeys (concrete) with large separation (urban) 4. Houses with more than 2 storeys (concrete) with small separation (city, dense urban) While all this is nice in theory, it must be possible to classify areas also in practice. Gröna Kartan has the following classifications that seems useful Låg Bebyggelse (mainly single family houses, i.e. 1 and 2 above) Hög bebyggelse (3 above) Sluten bebyggelse (4 above) In a smaller building a user is more likely to be close to an external wall than in a large building. The following values are suggested based on the COST 231 formula above, assuming values for the ground floor, i.e. Gfh = 0. Class We Wge D (m) Average Penetration Standard Deviation (db) (db) Loss Låg Bebyggelse db 4 Hög bebyggelse db 6 Sluten bebyggelse db 8 Since the variation of the penetration loss is significant (> 1 db), the variation needs to be included and dealt with statistically. Standard deviation figures for the different building types is not readily available from the COST report, but various other reports suggests values in the range 4-8 db.

9 9 (18) Traffic Load Higher average traffic will result in higher noise rise on both up and downlink. On the uplink it is suggested to use 1 db (corresponding to 20 % cell load) noise rise for low traffic cells (rural areas), and 3 db noise rise (corresponding to 50 % cell load) for cells in more populated environments (dense urban, urban, suburban). Values are taken from [2]. The noise rise on the downlink is not so straightforward to calculate. In the literature it can be found to be modelled as a single value based on the results of some simulation, but in reality it will depend on the receiver location in the cell, and whether the receiver is indoors or not. Typically the noise rise is higher on the downlink than in the uplink, partly due to asymmetric load, partly due to other spreading techniques. The link budgets in this report calculates the path loss margin the downlink has over the uplink, i.e. how much noise rise can be tolerated on the downlink, and draws some conclusions from this Use of TMA Coverage in UMTS networks is largely considered to be uplink limited in low traffic situations. The basis for this is that the base station has typically W (40-43 dbm) output power available, while the mobile unit has W (21 dbm). Even when considering the better Noise Factor of the base station (typically 4 db compared to 7 db in the mobile) and the approximately 4-5 db lower Eb/N0 required in the uplink due to antenna diversity on the base station, downlink still has a db path loss advantage over the uplink in a symmetrical service. In case of asymmetrical load (higher bitrates in the downlink than in the uplink), the db advantage reduces to around 5-10 db (assuming 384 kbits/s downlink and 128 kbit/s uplink). Uplink coverage can be improved by introducing Tower Mounted Low Noise Amplifiers, i.e. an amplifier directly after the antenna. The gain of this is that the feeder losses in the uplink can be ignored (expect for a short jumper cable between the antenna and the amplifier), and that the TMA often has a better Noise Factor (NF) than the base station (1.5 2 db compared to 4-5 db). TMA is widely used by the operators to improve coverage in rural areas Use of Boosters and high power amplifiers As indicated before, coverage in UMTS networks is normally uplink limited. Introduction of TMA can change this, which means higher output power from the base station than the standard W can be required. Boosters and high power amplifiers are of course also a possible ways for the operators to increase the pilot signal, but not always necessarily improving the coverage.

10 10 (18) Service Probability Radio signals fluctuates significantly when a receiver is moving around, even within a fairly small area where the distance and angle to the base station remains fairly constant. Two mechanisms are often used to physically describe what is happening Large objects such as buildings and trees obstructing the signal. These variations are referred to as shadow fading, lognormal fading or slow fading and occurs when a receiver moves tens of metres. All objects in the vicinity of a receiver cause reflections of the radio signal. In the receiver these reflections will add to one another, with different phase as a result of different propagation distances. This gives rise to fast fading or rayleigh fading. When planning a network, margins need to be added for both these phenomena. The margin for fast fading is especially important for slow moving mobiles, since they run a higher probability of being still in a fading dip. An indoor user will of course always be slow moving. The fast fading margin is found in row q in the link budget. The margin for shadow fading, or lognormal fading margin on row s in the link budget above, is already included in the license conditions ( with 95% area probability ), but only for the outdoor scenario. When planning a cellular network, a planning margin for the variation in the penetration loss is normally also added. As shown in section 4.2, a 95 % requirement for the outdoor signal corresponds roughly to 90 % indoor probability, and therefore no additional margin is added Soft handover gain The soft handover feature in CDMA networks means a mobile on the border between two cells will be connected two both, at each point in time using the best connection. When planning a network this gives two benefits: 1. A mobile can always choose the strongest cell 2. Diversity effects since an additional antenna is receiving the signal When planning a network, a statistical margin for fading is added. The mathematics of this normally deals with one cell at a time, which could be OK for a 2G network, where a handover to the stronger cell cannot not occur immediately. With soft handover, the mobile will always be connected to the stronger cell, and a radio planner could calculate a little gain in his calculations since the slow fading between the two cells is not 100 % correlated. However, this gain is already taken care of by the license conditions and measurement method, since that will always deal with the strongest cell. The additional antenna gain from increasing the number of antennas is marginal except in the case when path loss to both stations is exactly the same. For these reasons, no soft handover gain is assumed in the link budget.

11 11 (18) Link budgets for typical environments Classification of environments Since the link budget will depend greatly on the building type and environment, the following four typical environments have been defined: Rural: Single family houses in rural low traffic / interference areas. To be low traffic, area has to be a certain distance away from more densely populated areas, i.e. not being served by a base station in the more densely populated area, or a basestation that is affected from interference from the populated area. Coverage ranges is a couple of kilometres in suburban areas, so a value of 3 km from the boundary of the populated area is suggested. A suggestion on definition applicable in Sweden can be found in 5.2 If not classified as rural, areas should be classified as follows: Suburban. Single family houses in suburban areas. Buildings primarily constructed out of wood. låg bebyggelse according to Gröna Kartan. Urban: scattered multi-storey buildings with external walls of concrete or brick. Hög bebyggelse according to Gröna Kartan. Dense Urban: buildings are built next to one another in blocks, separated by narrow (10-50 meter) streets. These areas are classified as sluten bebyggelse in Gröna Kartan. In the link budget below, numbers have been rearranged slightly compared to the basic link budget above. The main purpose is to exclude the statistics that are already covered by the licence conditions from the calculation, and include a downlink calculation to find the limiting link.

12 12 (18) Dense Urban Suburban Rural Rural TMA Rural TMA Service UL kbit/s Service DL kbit/s Transmitter Max mobile transmit power dbm Mobile Antenna Gain dbi Body loss db EIRP dbm UL Receiver Thermal Noise dbm/hz NF Noise Density dbm/hz -170,0-170,0-170,0-170,0-172,0-172,0 Noise Power dbm -104,2-104,2-104,2-104,2-106,2-106,2 Interference Margin db Receiver interference Power dbm -104,2-104,2-104,2-110,0-112,0-112,0 noise + interference dbm -101,2-101,2-101,2-103,2-105,2-105,2 Processing Gain db 14,3 14,3 14,3 14,3 14,3 17,8 Required Eb/No db 1,5 1,5 1, Receiver Sensitivity dbm -113,9-113,9-113,9-115,4-117,4-120,9 Base station antenna gain dbi Cable loss db Max Path Loss db 148,9 148,9 148,9 150,4 155,4 158,9 Fast fading margin db Max Fading Path Loss db 144,9 144,9 144,9 146,4 151,4 154,9 Average Penetration Loss db Max outdoor UL Path loss db 124,9 128,9 133,9 135,4 140,4 143,9 Downlink Transmitter total available Power dbm Cable Loss db Antenna Gain dbi Transmitter total ERP dbm Max Service Power % 25% 25% 25% 50% 50% 50% Max Service ERP 51,0 51,0 51,0 54,0 54,0 54,0 DL Receiver NF Noise density dbm / Hz -167,0-167,0-167,0-167,0-167,0-167,0 Thermal Noise Power dbm -101,2-101,2-101,2-101,2-101,2-101,2 Processing gain db 10,0 10,0 10,0 10,0 10,0 10,0 Required Eb/No db RX sensitivity dbm -105,2-105,2-105,2-105,2-105,2-105,2 Max DL Path loss db 156,1 156,1 156,1 159,1 159,1 159,1 Fast Fading Margin db Average Penetration Loss db Max outdoor DL Pathloss db 132,1 136,1 141,1 144,1 144,1 144,1 Limiting Link UL UL UL UL UL UL Limiting PathLoss db 124,9 128,9 133,9 135,4 140,4 143,9 Downlink Margin 7,2 7,2 7,2 8,7 3,7 0,2 Pilot Power % 10% 10% 10% 10% 10% 10% Pilot ERP Measured Outdoor Pilot dbm -77,9-81,9-86,9-88,4-93,4-96,9 Measured Outdoor Pilot dbuv / m 65,1 61,1 56,1 54,6 49,6 46,1 The first part deals with the uplink calculation and follows the layout from the example in the beginning of the report, ending with a maximum allowed path loss as far as uplink is concerned. Then follows a calculation of the maximum allowed path loss on the downlink, to make sure we are not being downlink limited. First step is to decide the amount of power available for a single user. The assumption is a maximum of 25 % for a single user in high traffic areas (dense, urban, suburban), and up to 50 % in low traffic (rural areas).

13 13 (18) Then comes a calculation on the base station receiver sensitivity. An E b /N 0 value of 6 db has been assumed for all environments. The maximum outdoor path loss has been calculated from a downlink point of view has been calculated, without any margin taken for downlink interference and then the minimum of UL and Dl path loss values. Next, the downlink over uplink margin has been calculated. This value shows how much interference margin is allowed on the downlink. As can be seen, the link budget is uplink limited in all environments, except for the 384/64 kbit/s with TMA example, which seems to be almost exactly in balance, i.e. there is no margin at all for downlink interference. This example is not relevant for the license conditions, but shown as an example when link is being, or close to being, downlink limited. Before being able to calculate the minimum received pilot power, we need to decide the amount of power set aside form the pilot channel. The pilot power is parameterized in the base station and the operator can set virtually any value. In practice the range would be limited. In the lower end it will be limited by two factors: 1.The pilot signal must be strong enough to be heard over the thermal noise by all mobiles in the cell that are within uplink coverage range. This requirement means very little power is required. 2. As the pilot signal is used for measurements of target cells before a handover can take place, this means the pilot signal must be heard over the interference outside the own cell Total Power Cell A "Pilot Power Cell A" "Total Power Cell B" The figure above illustrates the received total power from two cells A and B and the received pilot power from cell A. When a mobile is moving from cell B towards cell A it must measure the signal strength on cell A s pilot power some time before going into a soft

14 14 (18) handover with cell A. Typically, cell A s pilot must be detected when cell B is 5 db stronger, in order to initiate a soft handover that will occur a little later. The required E b /N 0 on the pilot on the downlink is approximately 10 db. The processing gain on the pilot is 10*log(3840 / 12.2) = 25 db which means the minimum output power for the pilot is approximately = -10 db (= 10 %) compared to the output power of cell A and B. (This calculation is a little simplified. In reality interference is a sum of all signal from all cells) With a known percentage of the power set aside for the pilot power we can now calculate the received pilot signal corresponding to the maximum uplink path loss. It should be stressed that an operator can choose a higher value than 10 % of the total power for the pilot. Literature often suggests value in the range 5-20 %. An increase in pilot power over 10 % should affect the license requirement in direct proportion. It will however be in the interest of the operator not to increase the pilot power unnecessarily. Raising the pilot power will mean less power is available for services, and increase the interference level, thus also increase the required power for the service bearers, but as it is impossible to measure the amount of power set aside for the pilot, it may be very tempting for the operators to boost the pilot power over 10 % in order to achieve coverage according to the license conditions. 3.5 Current License Requirements and how it relates to the link budget above As can be seen from the link budgets above, the existing license condition of 58 dbuv / m is right between the values calculated in urban and suburban areas, if the operator is not using TMA:s. It is approximately 7 db to low for dense city centres and approximately 8 db too strict in a rural environment, assuming operators are using TMA and 10% pilot power.

15 15 (18) 4 COMMENTS ON OPERATOR PROPOSAL 4.1 Changed service threshold The proposed received pilot power of 50 dbuv / m matches the calculated requirement very well in low traffic rural areas, assuming operator are using TMA:s and a pilot power setting of 10 %. If 50 dbuv / m is to be accepted, it is proposed to do so only in a low traffic rural environment, while maintaining the requirement in other areas, or apply the calculated values for the other environments as well. This would mean increasing the signal strength requirement in urban and dense urban areas. A proposal for definition of rural areas is found in section It is also proposed that the regulator requests information about TMA:s used pilot power settings for all cells from the operators, and verify the assumptions in this report, before changing the requirement. Information about TMA:s and pilotpower settings could be requested on a per base station basis as part of the normal data reported by the operators. 4.2 Changed probability Calculating the margins for a given area probability is standard procedure for most radio planners. The maths required include translating the required area probability to the corresponding probability at the cell border, and then under the assumption that signal variations has a log normal distribution, applying the normsinv function to arrive at the margin required for a certain required probability. Assuming for a a rural environment, that signal propagates proportionally to -30log(d) and fading has a standard deviation of 6 db, 95 % area probability corresponds to cell edge probability of 85 %, requiring a planning margin of 6.2 db, while 90 % area probability corresponds to 74 % cell border probability and a planning margin of 3.9 db, i.e. a difference in margin of 2.3 db Using the more common assumptions applicable for city environment, signal propagates proportional to -35 log(d) and fading has a standard deviation of 8 db. The required margins from 95 and 90 % respectively is 8.7 and 5.5 db, i.e. a difference in margin of 3.2 db. The difference of changing the area probability from 95 % to 90 % is the same as lowering the signal strength requirement with approximately 2-3 db, a little depending on the environment. When planning cellular networks, it is often the indoor probability that is targeted. When a mobile is indoors, the building penetration loss will vary significantly depending on whether a window is present in the direction to the base station or not. The normal way of dealing with this is to combine the standard deviation for the slow fading and the building loss, i.e. σ tot = σ fad 2 * σ pen 2

16 16 (18) and then apply a margin to correspond to whatever percentage is required. σ fad normally varies between 6 db (rural open areas) to 12 db (cities, large buildings) while σ pen varies between 4 db (small wooden houses) to 8 db (large buildings). In our rural environment the σ fad is approximately 6dB and σ pen is 4 db. That gives σ tot = 7.2 The 95 % outdoor area probability will in this case correspond to 92 % indoor probability (corrected for the average penetration loss) due to the higher to signal variation in the indoor environment. Changing the outdoor requirement to 90 % probability would mean indoor probability will go down to approximately 87 %. In order to achieve a 90 % indoor probability, outdoor probability needs to be around 93 %. For this reason a change in the outdoor area probability requirementto 90 % is not supported. 4.3 Measurements The operators have very briefly presented some measurements that they suggest prove their point that 50 dbuv / m should be a relevant signal strength requirement. The exact point that they want to make is not totally clear, except showing that it is possible to maintain a 64 kbit/s uplink at a measured cpich level of 36 dbuv / m (3 db noise rise included). This corresponds fairly well to the TMA link budget above if we correct for the following: The link budget assumes 11 db penetration loss (+11) The link budget assumes 1 db noise rise instead of 3 (-2) The link budget assumes a 144 kbit / s instead of 64 (+3) = 48 dbuv / m. From the operators presentation of the measurement it is not clear exactly how the base station was configured, but this information has been submitted later on condition parts of it will be kept confidential. According to the information available and not under confidentiality the basestation on which the measurement was conducted was not equipped with TMA, which obviously makes the comparison with the TMA link budget invalid. What is also surprising is that the base station seems downlink limited even when not equipped with a TMA (downlink bit rate drops to 128 and 64 kbit /s at times, while uplink is not (?) affected). With so little measurement data and background information, it is difficult to draw any conclusions from the data presented.

17 17 (18) 5 IMPLICATIONS OF CHANGED REQUIREMENTS 5.1 Definition of rural areas By rural area we mean 1. Traffic / interference is low 2. Buildings are small, thus having low penetration loss and contributing little to shadow fading In Sweden the boundaries for tätort could be used to define populated versus rural areas. The definition of tätort means that more than 200 people live in houses separated less than 200 meters. However, when looking at the smaller tätort, it is obvious that they do not have any high buildings, or that 200 inhabitants will create any substantial load / interference on the network. A comparison between areas classified as high building by Gröna Kartan and the list of tätorter in Sweden shows that the number of inhabitants needs to be before the tätort shows any high buildings. It is difficult to estimate the traffic that will be generated. In 2G networks each subscriber is often assumed to generate merlang voice traffic in busy hour. Translated to a 3G network this will correspond to: people * 90 % penetration / 2 networks * merlang / sub * 12.2 kbit / s = kbits/s, which is fairly low, but not insignificant, load on a cell. Traffic can be expected to be higher on a 3G network than 2G due to more advanced services. Simulations show that an uplink load of approximately 300 kbits /s is shown to generate a noise rise of 1 db [2], which is the noise rise assumed in the link budget. The increased interference level around the city will not be confined to the city itself, but will affect areas served by base stations in the city. Range of city cells outside the city itself is estimated to be a couple of kilometres. Based on this it is suggested to define rural as areas outside a 3 km border around tätorter with more than 1000 inhabitants.

18 18 (18) 5.2 Operator Infrastructure Path loss in rural areas is approximately proportional to -30 log(d) in rural areas. This means a reduction in the signal strength requirement with 8 db, cell range increases with a factor 1,8. Theoretically this would mean the number of base stations in rural areas required would be 1/1,8 2 = 31 % of the original numbers. If a change in coverage requirements is confined to rural areas only, the calculated percentage can not be applied to the whole country There are 716 cities in Sweden with more than 1000 inhabitants (i.e. not classified as rural in the assumptions), thus not being affected by a changed requirement in rural areas. Without knowing how many base stations are planned in or around the 716 cities, it is difficult to calculate the number of base stations saved by the operator if requirements changes. It also depends on whether the operator has already planned and to some extent built for the 58dBuV / m requirement. Given that many base stations already planned and built in some cities, and that the coverage requirement may only change in rural areas, the saving in number of base stations will be significantly less than the 69% indicated above, probably closer to 30-40%. These numbers are based on very loose assumptions. To come to a better estimate, it is proposed to make a tentative network plan with both original and changed requirements and compare the two. Such a plan could be generated by a computer program, at least for comparison purposes. Unfortunately that is outside the scope of this report. The number of RNC:s will be affected but not to the same extent as the number of base stations, since placing of these is done also to act as transmission network nodes. It may also be difficult to change the planning of RNC:s at this point in time, since many RNC.s will already be operational. The number of MSCs and GSN:s should be virtually unaffected by changed requirements, since traffic is the major driver for the number of these elements. 6 REFERENCES [1] Tillståndsgivningen för UMTS I Sverige, 27 juni 2001 [2] WCDMA for UMTS. Harri Holma and Antii Toskal [3] Chapter 4.6, COST 231 final report [4] Microwave Mobile Communications, William C Jakes [5] Ansökan om Andrade tillståndsvillkor för tillhandahållande av nätkapacitet för mobile rteletjänster av UMTS/IMT-2000 standard. PTS, Dnr /10