HSUPA Performance in Indoor Locations Pedro Miguel Cardoso Ferreira Abstract This paper presents results of HSUPA performance tests in a live network and in various indoor environments. Tests were performed in locations with different indoor coverage solutions: indoor locations covered by outdoor sites, dedicated indoor site, optical repeater and RF repeater. to the mobile network. The test setup is shown in Figure 1. Keywords HSUPA, indoor, real live performance, throughput. I. Introduction This paper collects the results of several tests were data transfers were performed using UMTS HSUPA feature. HSUPA also known as Enhanced Uplink was defined by 3GPP [1] in Release 6 with the goal of improving UMTS uplink data transfers in a somewhat similar way HSDPA did for downlink. The main innovations which came with HSUPA were: fast Node B based scheduling, fast physical layer HARQ and optional 2ms TTI. These features together with the possibility of using SF2 it will allow in the near future peak data rates of 5.76MBps[2]. The tests were performed in four different indoor locations, each one with different characteristics and different solutions for indoor coverage. In the first test scenario the indoor coverage was provided by the outdoor network whereas in the rest a dedicated indoor system was present. The second test scenario was in a location with a dedicated indoor site, the third in an indoor location with an optical repeater and finally the fourth test scenario was located in a building with a RF repeater. From those tests a number of different metrics were collected to evaluate the performance of the data transfer, namely: Throughput [kbps], Ec [dbm], Ec/I0 [db] and Noise Rise [db]. III. Test scenarios Figure 1: Test Setup In the first test scenario the indoor coverage was provided by an outdoor site, which is by far the most common situation in indoor scenarios. In this scenario three test points were chosen. P1 next to the window with potentially the best received signal strength but also the highest interference form other sites. P2 in a standard indoor location with some walls in the direct path to the serving site. And finally P3, in a deep indoor location with the worst radio conditions. II. Test Description In each test scenario three test points were chosen with different radio characteristics. For each of the test points, consecutive uploads of a 5MB file were performed with intervals of 10s in between each upload. These tests try to emulate usual tasks which use HSUPA, as photo transfers or work files uploads, etc. Uploads were performed from a laptop equipped with a HSUPA compatible UE to a ftp server directly connected to the GGSN therefore avoiding throughput fluctuations not due Figure 2: Outdoor site The second test scenario was located in a building with a dedicated site, which is a solution normally used in buildings where a large number of users are expected either/or special service demands are forecasted. The coverage was in this case guaranteed through a distributed antenna system, which is partially shown in Figure 3for the serving antenna.
Figure 3: Dedicated site diagram The serving antenna was located in a room of the building where three test points were chosen: the first one by the window, a second one in the direct vicinity of the serving antenna and a third one in a standard indoor position; which can be seen in Figure 4. Test points with the same characteristics were chosen in the other test scenarios with dedicated systems. Figure 6: Optical repeater diagram Finally the last test scenario was located in a building with coverage guaranteed by a RF Repeater connected to a distributed antenna system. A floor plan of the test scenario where serving antenna and test points are marked can be seen in Figure 7. Figure 4: Dedicated site test layout For the third test scenario a building in which dedicated coverage is provided by an optical repeater was chosen. The building layout with the location of the serving antenna and the three test points is shown in Figure 5. Figure 7: RF repeater test layout RF repeaters receive signal from an outdoor site via a donor antenna and, in normal indoor cases, feed a distributed antenna system. The repeater itself only filters and amplifies the incoming signal and noise. RF repeater deployment is the cheapest way to boost indoor capacity, so they are used in locations whenever no capacity upgrade is foreseen and a limited area is to cover, due to its limited output power. A diagram showing the serving antenna integrated in the RF repeater system can be seen in Figure 8. Figure 8: RF repeater diagram Figure 5: Optical repeater test layout Optical repeaters are used in similar locations as dedicated sites. Optical repeaters allow however larger distances between the base station and antennas, because the larger part of this distance is connected via fibre optics which presents lower losses than coaxial cable. Moreover the signal is regenerated before being converted back to RF. An excerpt of the block diagram containing the serving antenna is shown in Figure 6. IV. Results All tests were performed according to the methodology already presented, using a power class 3 UE. Regarding HSUPA capability a category 3 UE was used, meaning a maximum of 2 SF4 and a 10ms TTI was supported, therefore the maximum theoretical physical throughput was limited to 1.45Mbps [2]. All the metrics were collected with the tool TEMS TM Investigation [3].
In the test scenario where the indoor coverage was provided by an outdoor site one can see in Figure 9 that, as expected, the average received signal strength decreases as we move indoor. Figure 12: RTWP [dbm] outdoor site Figure 9: RSCP outdoor site Finally in Figure 13 one can observe that despite the previous differences in some collected statistics the uplink application throughput was practically constant in all test points. Nevertheless this reduction as little impact on the signal quality as can be seen in Figure 10, only test point 3 shows a small degradation. Figure 13: UL Throughput outdoor site Figure 10: Ec/I0 outdoor site The UE transmitted power increase is, as expected, proportional to the received signal strength decrease, as can be seen in Figure 11. For the dedicated site scenario the received signal strength was as foreseen, stronger on the antenna test point and almost equal in the other two points. Figure 14 presents the collected values. Figure 14: RSCP dedicated site Figure 11: UETxPwr outdoor site Regarding the noise rise due to the tests, Figure 12 shows that its value is small. It varies between 0.3dB and 0.8dB in test points 2 and 3 respectively. However the difference towards the average RTWP is smaller or even inexistent in P2. This RTWP_AVG values was calculated based on hours adjacent to the ones the tests were performed when mostly R99 services were carried by the network. Figure 15: Ec/I0 dedicated site
In Figure 15, one can see that despite the received signal being almost equal in P1 and P3 due to the fact that the window location is a bit more exposed to outside interference the received signal quality is slightly worse in P1. This can explain as well the difference seen in Figure 16 for the UE transmitted power, where the average value collected in P1 is higher than P3. Figure 19: RSCP optical repeater Regarding the signal quality, Figure 20 shows that it is good in all test points with slightly better values in the test point by the antenna, as should be expected. Figure 16: UETxPwr dedicated site Once again, as seen in Figure 17, the impact on noise rise of HSUPA is reduced and in this case similar in the three test points. Figure 20: Ec/I0 optical repeater The UE transmitted power follows closely the received signal strength, as the obtained results in Figure 21 demonstrates. Figure 17: RTWP [dbm] dedicated site There almost no difference in the throughput between the three test points in the dedicated site scenario, as can be observed in Figure 18. Figure 21: UETxPwr optical repeater Figure 18: UL Throughput dedicated site In the optical repeater scenario the RSCP for the three test points follows a more natural pattern with the highest value close to the antenna, followed by the window location and the standard indoor location, as can be seen in Figure 19. Figure 22: RTWP [dbm] optical repeater
Figure 22 shows that noise rise is in this scenario much higher than in previous ones, reaching values around 3dB. Nevertheless, differences between the three test points remain small. Figure 26: UETxPwr RF repeater Figure 23: UL Throughput optical repeater Albeit the differences in noise rise, one can see in Figure 23 that the uplink application throughput values remain high. In this case P3 shows a small but noticeable worse value than the other points. Regarding the noise rise, Figure 27 shows that also in this scenario the tests caused a bigger impact than in the solutions with no repeaters. Although in this case the value is about 1dB which is significantly less than in the optical repeater scenario. Once again the difference between test points is barely visible. The RSCP in the RF repeater scenario is shown in Figure 24. The values follow what should be expected according to the test locations. Figure 27: RTWP [dbm] RF repeater Once again Figure 28 shows very good and even results in terms of throughput in all test points. Figure 24: RSCP RF repeater This is not the case for the Ec/I0, seen in Figure 25, where the window location strangely presents the best values. A possible explanation is a coherent combination of the signal from the indoor antenna and the repeated outdoor site at this point. Figure 28: UL Throughput RF repeater Figure 25: Ec/I0 RF repeater Again there is in this case a match between the RSCP and the UE transmitted power, which can be observed in Figure 26. In Figure 29, one can observe the average radio conditions and standard deviations values in which the tests were performed, for the four scenarios. The presented values sum up the results from the three test points in each scenario. As can be seen the scenarios with indoor dedicated coverage systems present, as expect, higher received signal strength than the case where indoor is covered by an outdoor site. Conversely, the signal quality shows higher values for the outdoor scenario than the dedicated systems scenarios.
Despite the differences in the radio conditions on the several scenarios the achieved uplink throughput values are quite good for all of them, as seen in Figure 32. This results are also possible because the UE is capable of compensate the worst radio conditions with extra power. Despite the values being all good there are slight differences that are inversely proportional to the noise rise values. Figure 29: Radio conditions Ec vs Ec/I0 The average UE transmitted power and standard deviation values are shown in Figure 30. The achieved results follow closely the ones from the received signal strength, with the exception of the optical repeater. This scenario despite the fact of presenting the highest RSCP shows a higher UE transmitted power value. This should be caused by the attenuator used to reduce the downlink signal due to the input power limitations of the optical system, which also attenuates the uplink signal. Figure 32: Uplink throughput V. Conclusions Figure 30: UE transmitted power Rise over thermal average values for the four test scenarios are presented in Figure 31, showing a significant difference between the solutions with and without repeaters. The dedicated site and outdoor site scenarios presented small noise rise values during the tests, therefore is expectable that a large number of users might use HSUPA simultaneously in the cell. The tests impact in the noise rise is much more visible in the repeater scenarios, especially in the optical repeater. This behaviour might put a more strict limitation on the number of HSUPA users allowed in these cases. From the results collected in the conducted tests one can conclude that HSUPA is a major improvement in the uplink data transfer performance. These results were nevertheless achieved in good radio conditions and low load. Moreover, the tests were performed with HSUPA limited to 2 SF4 and 10ms. There is also no significant difference in uplink throughput between the different indoor coverage solutions. However indoor coverage solutions based on repeaters present a higher noise rise due to HSUPA which can be an important limitation to the cell capacity. REFERENCES [1] http://www.3gpp.org/specifications [2] Holma,H. and Toskala,A., HSDPA/HSUPA For UMTS, High Speed Radio Access formobile Communications, John Wiley & Sons, UK, 2006. [3]http://www.ericsson.com/solutions/tems/index.s html Figure 31: Rise over thermal