WCDMA Network Performance in Variable Repeater Hotspot Traffic Cases P. Lähdekorpi, J. Niemelä, J. Borkowski, J. Lempiäinen Institute of Communications Engineering Tampere University of Technology P.O. Box 553 FI-33 TAMPERE FINLAND Tel. +358 3 35 4552, Fax. +358 3 35 388 Email: {panu.lahdekorpi, jarno.niemela, jakub.borkowski, jukka.lempiainen}@tut.fi Keywords: hotspots, repeaters, WCDMA. Abstract The target of the paper is to illustrate the impact of hotspot traffic distributions on the WCDMA network performance when repeaters are used for providing service to the hotspot users. The paper concentrates on providing a detailed systemand cell-level analysis of the impact of repeaters by using static Monte Carlo simulations. Moreover, the effects of different repeater configurations are presented. Simulations were made by using different repeater gains, different hotspot traffic densities, and two different repeater distances to find out a planning guideline for WCDMA repeaters. The results show that the optimum repeater gain does not depend on the amount of traffic in hotspots, but mainly on the repeater configuration, and on the distance from the mother cell. Moreover, cell-level analysis reveals that repeaters have remarkable effect on the interference levels in cells surrounding the repeater cells. The results also show, how the downlink capacity can considerably be improved by using repeaters. Introduction WCDMA (Wideband Code-Division Multiple Access) repeaters are used as an amplifier unit between the mother base station and mobile stations. Correct installation of a repeater results in increased signal level under repeater service area. Repeaters are deployed to help mobile stations together with the mother base station to use lower transmit powers and thus to reduce interference propagated to the surrounding cells. Hence, repeater is a very attractive choice, e.g., when hotspots (areas with increased traffic density) are introduced. Repeaters, and their effect on the network performance in hotspot cases, have been studied in [] and [2]. Clear increase in downlink capacity was observed in both cases. However, the uplink direction seems to be more problematic in sense of capacity and interference as illustrated in []. In addition, repeater field measurements also indicate the increase in downlink capacity at properly adjusted repeater gains [3]. This paper presents the expected WCDMA network performance (in system-level and in cell-level), when repeaters are installed near hotspot areas. Moreover, the effects of repeaters on the nearby cells are analysed. This paper also clarifies what happens in the network especially in cases with different hotspot traffic loads, and what is the effect to the overall network performance in these cases compared to the case without repeaters. Finally, the effects of different repeater distances are studied to illustrate the importance of correct repeater configuration. 2 Simulations in brief Analysis in the following chapters is based on Monte Carlo simulations with same simulation parameters already presented in []. Support for repeaters and hotspots was implemented in the static network simulator (NPSW [4]). Simulations were performed with two repeater configurations and by using voice users only. In the scenario, repeaters were located at the distance of 5 m (path loss value db) from the mother base station. In the scenario 2, repeaters were brought to the distance of 333 m (path loss value 96 db). Figure shows the simulation scenario 2. Table defines the terms used in the cell-level analysis. Figure : Network layout with 9 3-sectored sites and 6 repeaters serving 6 hotspots in the centre. Site spacing is m and repeaters are located 333 m away from the mother base station (scenario 2).
3 2.5. (regular cells) (regular cells) 6 (regular cells) (regular cells). (repeater cells) (repeater cells) 6 (repeater cells) (repeater cells) 2 i UL.5.5 Figure 2: UL other-to-own cell interference in different hotspot traffic densities when using the scenario (repeater distance 5 m) with 2 homogenous overall users. Solid lines indicate neighbouring regular cells (averages) and dashed lines indicate repeater cells (averages). i UL 3 2.5 2.5.5. (regular cells) (regular cells) 6 (regular cells) (regular cells). (repeater cells) (repeater cells) 6 (repeater cells) (repeater cells) Figure 3: UL other-to-own cell interference in different hotspot traffic densities when using the scenario 2 (repeater distance 333 m) with 2 homogenous overall users. Solid lines indicate neighbouring regular cells (averages) and dashed lines indicate repeater cells (averages). The repeater loss (G T ) value is used to define the repeater configuration from the mother base station antenna to the repeater service antenna [5]: G = G L + G + G [db], () T BS DONOR where G BS is the mother base station antenna gain and G DONOR is the repeater donor antenna gain. Different repeater configuration scenarios can be simulated by using different path loss (L) between the repeater and mother base station and varying repeater gain (G REP ). Hotspot traffic density (the density of the users in hotspot area) is defined by the hotspot density factor (). Hotspot traffic (T HS ) is defined by the traffic density in the overall network area per square km multiplied by the factor: T HS A HS REP = D, (2) where D is the user density of the whole network area (users/km 2 ) and the A HS is size of the hotspot area (km 2 ). Figure 4: Downlink throughput in cell-level in the scenario 2. Network parameters: = 6, repeater gain = 75 db. White color indicates blocked cells. Name Definition Related cells Repeater cell A cell with repeater installed BS2, BS3, BS8, BS, BS8, BS2 Regular cell A cell with no repeater installed BS, BS4, BS7, BS4, BS5, BS9 Table : Cell definitions. values between. and 2 were used in the simulations. Very small value (such as.) means that no hotspot traffic is present. A special traffic case was also included, where the homogenous overall traffic layer was removed, including user traffic only in the hotspots. In this case the amount of users in the whole simulated area equals to the total number of users in the hotspots. 3 Cell-level analysis Cell-level analysis was performed with different hotspot traffic densities and with both simulation scenarios. Repeater cells and regular cells were analysed separately. Figures 2 and 3 show averaged uplink other-to-own cell interferences (i UL ) as a function of the repeater gain for the scenarios and 2. They illustrate how other cell interference of the regular cells increases rapidly at high repeater gains. At the same time, own cell interference in regular cells is reduced because of smaller cell dominance area and thus smaller number of users in the cell. These two figures also indicate that the repeater gain can not be increased after certain limit without heavily affecting interference levels in the surrounding cells. Dashed lines in Figure 2 and 3 indicate lower uplink other-to-own cell interference in repeater cell at high repeater gains because the cell dominance area expands (own cell interference increases) when repeater gain increases. Figure 2 indicates also that when the density of traffic in hotspots is increased, the effect to the uplink other-to-own cell interference is more sensitive. Use of too high repeater gains can be observed as the cell blocking phenomenon as in Figure 4.
3 UL load limit BS TX Power limit.9 25.8 2.95.7.6 Dropped users 5 Service probability.5.9.4 3 users. REP ON (REP OFF value:.97) 3 users REP ON (REP OFF value:.969) 3 users 6 REP ON (REP OFF value:.848) 2 users REP ON (REP OFF value:.896) users 2 REP ON (REP OFF value:.92) users Only HS users REP ON (REP OFF value:.569).3 Figure 5: Overall network service probability in different hotspot traffic cases when the scenario is used. 5.85 Figure 7: Outage statistics using scenario with 3 homogenous overall users and with no hotspots..9.8.7.6.5 3 users. REP ON (REP OFF value:.97).4 3 users REP ON (REP OFF value:.969) 3 users 6 REP ON (REP OFF value:.86) 2 users REP ON (REP OFF value:.94) users 2 REP ON (REP OFF value:.928) users Only HS users REP ON (REP OFF value:.565).3 Figure 6: Overall network service probability in different hotspot traffic cases when the scenario 2 is used. The effect of varying repeater distance (i.e., varying G T in Equation ()) can be clearly seen from Figures 2 and 3. The curves are shifted to the left by 4-5 dbs. This comes from the path loss reduction and can be seen as a change in the optimum repeater gain value. 4 System-level analysis 4. Service probability It is now seen how the hotspot traffic cases and the repeater configuration affect uplink interference in cell-level. Systemlevel analysis is also needed to see the changes in overall network capacity. The results in Figure 5 indicate that the overall service probability is increased a little by using repeaters for high traffic hotspots compared to the case without repeaters. Figure 5 also shows the importance of setting the repeater gain correctly in high hotspot traffic density cases (=6). The raise in the service probability (2-4 percent units) at repeater gains 6-7 db can be explained by the high value. High amount of hotspot users (using low transmit powers) will result in noticeable decrease in uplink interference although the repeater is amplifying the interference to the other cells. This phenomenon is insignificant with lower values. The rapid decrease in service probabilities (Figures 5 and 6) in low value cases is due to the higher overall network load and interference (case with 3 users). By looking at the low- cases in Figures 5 and 6, a shift of 4-5 dbs in optimum repeater gain value can be observed as expected. Outage statistics presented in Figure 7 show what happens in the network when increasing the repeater gain. Solid curves in Figure 7 present the number of dropped users for that particular reason. The two main reasons for dropped users were: UL load limit and BS total maximum transmit power limit. Network configuration for the results in Figure 7 included 3 homogenously distributed users and no hotspots. This represents highly loaded network scenario, since the service probabilities were less than %. Scenario was used with repeater distances of 5 m. It is clear from the Figure 7, how the base station transmit powers are decreasing when increasing the repeater gain, and how the uplink interference dominates the overall service probability. Network capacity could be increased in downlink direction by using even higher repeater gains. High repeater gains quarantee lower BS transmit powers, and thus, increased network capacity. Low downlink other-cell interference levels in repeater network scenarios can be explained by the use of directional BS antennas and good site planning. Finally, repeaters are more effective in increasing WCDMA network performance in downlink direction. However, the uplink direction will limit the overall capacity when using high repeater gain values. This rapid increase in uplink interference in the repeater equipped areas can be explained by repeater donor antenna leakage at high repeater gains. It should also be considered, that UEs are assumed to have omni-directional antennas, i.e., they are radiating to all directions. The uplink interference limitation makes it necessary to consider setting the repeater gain in both directions separately. It might be reasonable to use the repeater amplification in downlink direction more than in uplink direction.
2 3 6 5 4. 6 2 MS TX Power [dbm] 4 5 6 7 Capacity gain [%] 3 2 Downlink 8 9 2 3 users. REP ON (REP OFF value: 2.475) 3 users REP ON (REP OFF value: 2.4) 3 users 6 REP ON (REP OFF value: 2.32) 2 users REP ON (REP OFF value: 3.) users 2 REP ON (REP OFF value: 3.9) users HS users only REP ON (REP OFF value: 3.92) 2 Figure 8: Averaged MS transmit powers in different hotspot traffic cases for the scenario. 2 Repeater gain [db] Figure : Evaluated uplink and downlink capacity gains for the scenario. Uplink 2 5 4. 6 2 Downlink 3 3 MS TX Power [dbm] 4 5 6 7 Capacity gain [%] 2 8 Uplink 9 2 3 users. REP ON (REP OFF value: 2.447) 3 users REP ON (REP OFF value: 2.29) 3 users 6 REP ON (REP OFF value: 2.45) 2 users REP ON (REP OFF value: 3.54) users 2 REP ON (REP OFF value: 4.727) users HS users only REP ON (REP OFF value: 3.288) 2 Figure 9: Averaged MS transmit powers in different hotspot traffic cases for the scenario 2. 4.2 MS transmit power As mentioned before, mobile station transmit power is a crucial parameter in determining the efficiency of a repeater configuration. Figures 8 and 9 present the network-level averaged mobile station transmit powers for the scenario and scenario 2 when different hotspot traffic cases are simulated. In Figures 8 and 9, the impact of repeaters on the mobile station transmit powers is most clearly seen from the traffic case with only hotspot traffic present. When using the case with only hotspot traffic, the resulting transmit power value is still an average from the whole network. However, this value represents quite well an average taken only from hotspot users (i.e., repeater cell users). This is due to the fact, that most of the connections are made by using a repeater cell. A reduction of almost 5 db in MS transmit power average is observed in this particular case when using the scenario. In cases of low hotspot traffic present, the phenomenon is mainly insignificant, because the uplink interference dominates and raises the MS transmit power average value. This, together with the service probability results from Figure 5 and 6, basically indicates that repeaters are useless in case of low hotspot traffic situations. However, in case of high 2 3 Figure : Evaluated uplink and downlink capacity gains for the scenario 2. hotspot traffic, maximum average gain of 5 db is expected in mobile station transmit powers. When comparing the Figures 8 and 9, the change in the G T value (also in the optimum repeater gain value) is again clearly visible. MS transmit power curves rise 4-5 dbs earlier when using the scenario 2 instead of using the scenario, thereby indicating major uplink interference at high gains. 4.3 Capacity evaluation Repeater gain [db] Network-level capacity evaluations were made for scenarios and 2 to determine the optimum repeater gain setting and DL capacity gain for each hotspot traffic case. The results from these evaluations are presented in Figures and. Capacity gains in Figures and are calculated by comparing the averaged network throughputs when repeaters are swithced on and off at a certain network load point. These calculations were made separately to each of the hotspot traffic cases. The optimum repeater gain values and capacity gains are finally put in Table 2. The optimum repeater gain values are extracted from Figures and by allowing a loss of 5 percent units in uplink capacity. Finally, the DL capacity gains are selected by using these optimum repeater gain values.
Scenario Opt. G REP (db) DL capacity gain (%) Scenario 2 Opt. G REP (db) DL capacity gain (%). 6 Results from the capacity evaluations (Figures and, Table 2) show, that the optimum repeater gain is not largely changing, when changing the hotspot traffic density. The path loss reduction of 4 dbs is again visible when comparing the optimum repeater gain values between the scenarios and 2. When looking at the capacity gain values in Table 2, the downlink capacity is clearly increased in both scenarios when using higher values. Up to 35 percent unit increase is observed in DL capacity gain when comparing the lowest and the highest hotspot traffic density case for the scenario. In case of the scenario 2, the difference is not that large due to the increased repeater noise and interference issues. The results in Table 2 show the importance of lowering the users transmit powers in hotspots with high number of users, e.g., by using a repeater. Although the small uplink capacity loss is still present, the downlink can be now more efficiently used to, e.g., serve higher speed data users. 5 Conclusions and discussion 2 72 72 72 73 72 3 6 27 35 48 68 68 67 69 66 3 6 22 28 39 Table 2: Optimum repeater gain values and downlink capacity gain values for the scenarios and 2. The simulations have indicated that repeaters are very effective in increasing the overall downlink capacity in all hotspot traffic cases. However, uplink direction has proven to be the bottleneck for the system performance, when using repeaters. With repeater gains larger than 7 db, the rapid increase in uplink interference levels starts to deteriorate the system performance. Optimum repeater gain for these simulation scenarios has been seen to depend mostly on the repeater configuration (repeater distance). The smaller is the repeater distance, the smaller is the optimum repeater gain (See Table 2). Improvement up to 5 percent units in overall service probability has been observed in case of high hotspot traffic. If allowing 5 % decrease in uplink capacity, a gain of 6 % in downlink capacity is observed even when no hotspots are introduced. When the amount of hotspot traffic is increased, much higher DL capacity gains are observed. Due to the differing behaviour of the uplink and downlink directions, the repeater gain could be set separately for uplink and downlink. Higher growth in downlink capacities could be observed by using more gain in downlink than in uplink, without simultaneously raising the uplink interference. An interesting application for outdoor repeater is the connection to indoor WCDMA networks. Repeaters could be located in between outdoor network and indoor network in such a way, that the penetration loss from the wall materials is bypassed. Acknowledgments Authors would like to thank European Communications Engineering (ECE) Ltd for helpful comments and the National Technology Agency of Finland for funding the work. References [] J. Niemelä, P. Lähdekorpi, J. Borkowski, J. Lempiäinen, Assessment of repeaters for WCDMA UL and DL performance in capacity-limited environment, In Proc. IST Mobile & Wireless Communications Summit, Dresden, June 25. [2] M. Rahman, P. Ernström, Repeaters for hotspot capacity in DS-CDMA, IEEE Trans. Vehicular Technology, issue 3, vol. 53, pp. 626-633, May 24. [3] J. Borkowski, J. Niemelä, J. Lempiäinen, Applicability of repeaters for hotspots in UMTS, In Proc. IST Mobile & Wireless Communications Summit, Dresden, June 25. [4] J. Laiho, A. Wacker, T. Novosad, Radio network planning and optimisation for UMTS, CD-ROM, John Wiley & Sons, Ltd., 22. [5] Qualcomm white paper, Repeaters for Indoor Coverage in CDMA Networks, [online]. Available: http://www.repeaterone.com.