Impact of Base Station Locations and Antenna Orientations on UMTS Radio Network Capacity and Coverage Evolution

Impact of Base Station Locations and Antenna Orientations on UMTS Radio Network Capacity and Coverage Evolution Jarno Niemelä and Jukka Lempiäinen Tam...
0 downloads 4 Views 263KB Size
Impact of Base Station Locations and Antenna Orientations on UMTS Radio Network Capacity and Coverage Evolution Jarno Niemelä and Jukka Lempiäinen Tampere University of Technology, Institute of Communication Engineering P.O.BOX 553, FIN-33101 TAMPERE, FINLAND [email protected] and [email protected]

Abstract The aim of this paper is to study the impact of the base station site locations and antenna orientations on UMTS radio network performance under different traffic distributions, and to find an optimum deployment strategy from 3-sectored to 6-sectored site solutions. A radio network planning tool has been used to verify the radio network system performance of different network scenarios. The results show that UMTS network is fairly robust for small base station site location and antenna orientation deviations. Furthermore, the results indicate that in UMTS radio network planning, the concentration should be paid rather on the base station site and antenna configurations than on base station location. According to the simulations, the most efficient 6sectored capacity and coverage network configuration can be achieved by utilizing horizontally 33° half power narrow beam antennas.

1. Introduction The required network coverage and system capacity together with sufficient quality of service and implementation costs are the most essential elements that define the site density and configuration for a planning area. Site density has a direct impact on the required site locations, which moreover have an important role in radio network performance. The overall network performance is easily deteriorated by having a base station site location farther away from a traffic congestion. This could occur, especially, in interference-limited WCDMA-based UMTS network, where dense site locations create easily additional interference rise in the cell coverage areas, and thus reduce system capacity. Several contributions affect the selected base station site locations of an operator. A hexagonal grid planning would be an attractive choice for a network layout, but typically nonhomogenous terrain and traffic distribution restricts this deployment strategy. Moreover, in urban environments, where the base stations are often located on the top of the buildings, the physical space required for the hardware of the planned site solution may not be enough. At some places, the authority constraint may prevent an operator to use a certain site location. Also economical reasons could persuade to select another place for a site. Therefore, the search of candidate site locations may be a challenging task due to limited number of available site locations, and thus an optimal site configuration is often more critical task than site location [1]. The effect of a small randomness in base station site locations has been found to have a negligible impact on the performance of the cellular network under homogenous environment and traffic distribution [2]–[5]. On the contrary, in [6], which concentrated especially on the UMTS system, the base station site locations were concluded to have a notable impact on downlink (DL) and uplink (UL) performance with relatively large cell range.

Antennas of a sectored base station are sometimes directed by having equal spacing. In practice, due to obstacles close to the base station site location, or due to errors, e.g., in the base station antenna implementation, the antenna directions may change, and thus affect the network performance. Hence, the effect of antenna direction deviation on the network performance should be considered. 3-sectored sites are a practical solution for a macro cellular network of typical capacity requirements. When capacity requirements are increasing, and 3-sectored sites are not able to provide enough capacity, more sectors have to be implemented. 6-sectored UMTS sites provide significant capacity enhancements compared to 3-sectored networks [7]–[8]. Hence, in the radio network planning process, an optimum site evolution strategy has to be defined. The aim of this paper is to evaluate the impacts of the base station site location and antenna direction deviations on the UMTS network performance under non-homogenous environment and different traffic models (a homogenous traffic distribution with and without indoor users and a weighted number of indoor users). Moreover, the aim is to find an optimal base station antenna configuration for 6-sectored site solutions; both from network coverage and system capacity point of view.

2. Deviation of base station site location and antenna direction Without terrain variations and traffic congestions, a hexagonal grid planning would be an efficient way to deploy a macro cellular network. In Figure 1, an impact of non-hexagonal base station site locations is illustrated. BS1

BS3

BS2

Figure 1.Illustration of non-hexagonal site locations. As a consequence of the new base station locations, coverage overlapping is increasing between the sites BS2 and BS3. Hence, the other-cell interference level is also rising. On the other hand, the overlapping between the sites BS1 and BS2 is decreasing, thus reducing other-cell interference. However, the mobiles located in the area between BS1 and BS2 may have

coverage problems. Thus, the effect of deviations in the site locations is closely related to the average distance between base stations. The new base station location results also modified soft handover (SHO) areas, since the received pilot power levels change. According to this analysis, a maximum deviation in a site location has to be defined that avoids a significant reduction in the network performance. Previous analysis was made by assuming a homogenous traffic distribution. Intuitively, in a non-homogenous traffic distribution, if the new (randomized) base station locations would be farther away from a traffic hot-spot, e.g., business building, the performance of the network would deteriorate. The effect base station antenna direction deviations is illustrated in Figure 2. If the directions of the antennas belonging to the same site are changed like at BS1, it results an increase in the number of softer handover (SfHO) connections. These additional SfHO connections consume the air interface capacity. The number of additional SfHO connections depends strongly on the horizontal beam width of the base station antenna. On the other hand, if the antennas of different base stations are directed more towards to each other, the coverage overlapping increases, and may result an increase in SHO connections. BS1

Increased number of SfHO connections

deviation was considerations.

assumed

to

be

enough

for

practical

Table 1. General simulation parameters. Base station Maximum TX power CPICH CCCH SCH Noise figure Required Eb/N0 Mobile station Maximum TX power Dynamic range Noise figure Required Eb/N0 Other STD of slow fading Inter-cell correction of slow fading Intra-cell correction of slow fading UL noise rise DL orthogonality SHO window

43 dBm 33 dBm 33 dBm 33 dBm 5 dB 5 dB 21 dBm 70 dB 9 dB 8 dB 10 dB 0.5 0.8 6 dB 0.6 4 dB

Increased number of SHO connections BS2

Figure 2.Illustration of antenna direction deviations.

3. Simulation parameters In the simulations, a static simulator using Monte-Carlo simulations were utilized to verify the system performance of different network scenarios. An accurate digital map (including morphological and topographical data of the simulation area together with the building rasters) was given as an input for the simulator for the coverage predictions. Okumura-Hata propagation model was applied together with some topographical correction factors. Propagation slope was kept constant (35dB/dec) through simulations, and an average area correction factor was set to -6.7 dB corresponding to light urban/suburban environment. The base and the mobile station antenna heights were 25 m and 1.5 m, respectively. The user profile consisted only of speech users (12.2kbit/s). Other relevant simulation parameters are gathered in Table 1. In the reference network scenario, 17 sites were located in a regular hexagonal grid having equal site spacing of 1.5 km (exact reference site locations are marked with red balls, Figure 3). Base stations were configured with three sectors and 65° antennas, whose directions were 0°, 120°, and 240°. The effect of irregular grid was studied by giving an error for each base station site location by vector, thus describing its new location in the irregular grid. The maximum deviation from the hexagonal location was one-quarter of the site spacing. This

Figure 3. A part of the simulation area of reference base stations in ideal hexagonal grid (with red ball in the middle), and base stations with deviation in their location.1.5 km. In the antenna orientation (antenna directions) deviation study, the base station site locations followed ideal hexagon (site spacing 1.5 km), but a random deviation was added for each antenna direction as illustrated in Figure 4. The antenna direction deviations were selected according to normal distribution of different weights. In the first simulations, the average deviation was 9.1° and 18.2° in the second. Also these deviations were based on practical considerations.

Table 3. Results of the base station location deviation with homogenous traffic distribution (indoor users included) and 1.5 km site spacing. Parameter Service probability [%] SHO probability [%] SfHO probability [%] UL load [%] UL i DL and UL throughput [kbit/s/sector] DL power [dBm]

Figure 4. A part of the simulation area illustrating the base stations antennas with direction deviation of 18.2°.

4. Simulation results 4.1 Base station locations and antenna directions The simulation results of the base station site location deviations in case of 1.5 km site spacing are presented in Table 2. The results of irregular grid are an average value of five different simulation scenarios. Table 2. Results of the base station site location deviation with homogenous traffic distribution (no indoor users) and 1.5 km site spacing. Parameter Service probability [%] SHO probability [%] SfHO probability [%] UL load [%] UL i DL and UL throughput [kbit/s/sector] DL power [dBm]

Reference grid 98.1 24.9 3.9 43.2 0.79

Irregular grid 98.3 25.0 3.9 43.4 0.80

315

318

37.2

37.3

A small randomness in the base station site locations does not deteriorate the network performance in a practical urban network layout when indoor coverage is required. On the contrary, the average service probability of the irregular grid networks is slightly improved compared to reference network, which shows the effect of terrain fluctuations. The UL other-toown-cell interference is slightly increased as well as the DL transmit power and the UL load. Moreover, the change in SHO and SfHO connections is negligible. The same scenario was simulated also including the indoor users (Table 3). In the Monte-Carlo simulations, standard deviation of slow fading for indoor users was 15 dB, and an additional indoor penetration loss of 15 dB was added for indoor mobiles. In general, after including indoor users, the service probability deteriorates. This is caused by UL coverage limitations. Altogether, introducing indoor users does not affect a lot because of the high indoor coverage threshold. Moreover, the inter-related results of the regular and the irregular networks are only slightly changed.

Reference grid 97.5 24.4 3.8 42.6 0.71

Irregular grid 97.5 24.6 3.8 42.7 0.72

314

315

37.2

37.3

In Table 4, the results of the simulations of non-homogenous traffic distribution (70% indoor and 30% outdoor users) are presented. The simulations were performed using half the number of users than in previous scenarios. Despite of that, the service probability in both cases remains under 95%, since the networks are highly UL coverage limited. However, the network performance of the regular and the irregular hexagonal scenarios is almost the same. Table 4. Results of the base station location deviation with nonhomogenous traffic distribution and 1.5 km site spacing. Parameter Service probability [%] SHO probability [%] SfHO probability [%] UL load [%] UL i DL and UL throughput [kbit/s/sector] DL power [dBm]

Reference grid 94.0 23.7 3.4 19.2 0.77

Irregular grid 94.1 23.9 3.5 19.9 0.78

149

149

32.6

32.6

The relation between site location and site distance was analyzed by doubling the site spacing into 3.0 km. The effect of randomness in the site locations begins to some extent affect (Table 5). Compared to reference network, the service probability has dropped over 2%, and moreover the differences in other network parameters are distinguishing more than in 1.5 km site spacing scenario. Table 5. Results of the base station location deviation with homogenous traffic distribution (no indoor users) and 3.0 km site spacing. Parameter Service probability [%] SHO probability [%] SfHO probability [%] UL load [%] UL i DL and UL throughput [kbit/s/sector] DL power [dBm]

Reference grid 96.2 17.2 4.6 37.6 0.58

Irregular grid 94.0 18.0 4.9 37.1 0.60

289

285

36.3

36.3

Table 6 shows the pilot coverage probabilities for indoor (-84 dBm) and outdoor (-104 dBm) thresholds of 1.5 km and 3.0 km site spacings. The effect of randomness in the site locations does not affect, if the network is planned to have sufficient

indoor coverage probability (larger coverage overlapping). However, if high indoor coverage probabilities are not required and larger cell dominance areas are accepted, the deviation in the site location becomes more crucial. Table 6. Indoor and outdoor coverage probabilities of 1.5 km and 3.0 km site spacings. Site spacing

1.5 km

3.0 km

Reference grid

Indoor 87.9

Outdoor 99.9

Indoor 54.9

Outdoor 96.6

Irregular grid1

90.2

99.9

50.6

96.6

Irregular grid2

86.4

99.6

51.6

95.5

Irregular grid3

87.9

99.6

52.7

96.9

Irregular grid4

84.0

98.9

53.6

95.9

Irregular grid5

87.5

99.9

52.5

95.9

The impact of antenna direction deviation was studied using 1.5 km site spacing. The results in Table 7 show that the antenna deviations cause only a small degradation in the network performance. Probably the most remarkable change is observed in the SfHO probability. The amount of the change in service probabilities and in DL and UL performance is similar to the base station site location study. Moreover, the results of the different base station antenna direction deviation scenarios were about constant even if indoor users were added. Table 7. Results of the antenna direction deviation with homogenous traffic distribution. Scenario (1) refers to 9.1° average deviation, and scenario (2) to 18.2° average deviation. Parameter Service probability [%] SHO probability [%] SfHO probability[%] UL load [%] UL i DL and UL throughput [kbit/s/sector] DL power [dBm]

98.1 24.9 3.9 43.2 0.79

Antenna direction (1) 98.0 25.1 4.1 43.3 0.80

Antenna direction (2) 98.0 25.0 4.9 43.6 0.81

315

318

319

37.2

37.4

37.3

Reference grid

Table 8. Results of the antenna direction deviation with nonhomogenous traffic distribution and 1.5 km site spacing. Parameter Service probability [%] SHO probability [%] SfHO probability [%] UL load [%] UL i DL and UL throughput [kbit/s/sector] DL power [dBm]

Reference grid 91.7 24.6 3.6 38.8 0.73

Antenna direction (1) 91.0 24.6 3.8 38.8 0.74

292

292

36.7

36.7

In second scenario, a non-homogenous traffic distribution (70% indoor and 30% outdoor users) was used. The results in Table 8 show that the service probability has dropped over 6% compared to the corresponding scenario in Table 7. Nevertheless, the difference between the reference network and

the antenna direction deviation scenarios in the network performance is negligible. 4.2 Base station site configuration evolution from 3-sectored to 6-sectored Altogether, four different 6-sectored site solutions were compared to each other and to 3-sectored site. The first two configurations were 6-sectored sites of 65° and 33° antennas with fixed antenna directions. The third 6-sectored configuration was 33° antenna with 30° shift in antenna directions in the first tier of the base stations. This was utilized in order to fill in the coverage holes caused by the narrow beam antennas. In the fourth configuration, the original 65° antennas of the 3-sectored sites were attained and 33° antennas were added between them. The results of pilot coverages of different thresholds are presented in Table 9. The thresholds correspond to urban in-building (-75 dBm), suburban in-building (-84 dBm), and outdoor coverage (-104 dBm) thresholds. In general, the coverage performance of the 33° antennas in 6-sectored sites is superior compared to 65° antennas. On the other hand, the best suburban in-building coverage is achieved with a combined use of 65° and 33° antennas. This is partly caused by the fact that 65° antennas are not so directive as 33° antennas, and thus the coverage is smoother. Due to the indoor planning thresholds, the selected site spacing was 1.5 km, and thus nearly 100% outdoor coverage probability was achieved in all scenarios. Table 9. Pilot coverage probabilities of 3- sectored and 6sectored site solutions. Site type 3-sec/65 6-sec/65 6-sec/33 6-sec/33 (1st tier antennas) 6-sec/65&33

-75 46.1 61.9 64.9 67.8 65.7

Thresholds [dBm] -84 -104 87.9 99.8 91.5 99.9 90.7 99.9 91.0 99.9 91.6 99.9

The best pilot urban in-building coverage is achieved with 33° antennas when the antennas in the first tier of the base stations are shifted 30° (referred as 1st tier antennas). The typical pilot coverages of 65° and 33° antennas are depicted in Figure 5. The coverage area of 65° antennas reminds coverage of omnidirectional antennas, whereas the coverage of 33° antennas reminds a ‘flower’. However, using narrower antenna, the signal is directed more precisely to the sector serving area, and less interference is radiated toward other sectors. The results of Table 10 show the performance and capacity analysis of 3- and 6-sectored sites. The simulations were made with homogenous traffic distribution, and the number of mobiles was selected based on the approximately 95% service probability in the network of 3-sectored sites. As was expected, the service probability of the 6-sectored networks is nearly 100%. Moreover, mean number of mobiles in SHO is quite equal in all 6-sectored scenarios, which is a couple of percent less than in the 3-sectored network. The UL other-to-own-cell interference of the 6-sectored sites of 65° antennas is increased due to the wider beam width antennas. This results quite huge sector overlapping and hence extremely large SfHO areas (probability). The number of SfHO connections can be easily controlled by using narrower antennas or combined antenna solution (65° and 33°). Based on the antenna direction deviation

study, it was partly expected that shifting base station antennas (33° scenarios) does not improve the network performance.

An optimum deployment strategy for 6-sectored sites is based on the narrow beam antennas from coverage, capacity and network performance point of view. The performance reduction of wider antenna beam widths is related to larger sector overlapping. This causes more other-cell interference (increased loads and transmit powers), and thus the network of widebeam antennas can not meet the future capacity requirements as the narrow beam antennas can.

6. Acknowledgements Authors would like to thank European Communications Engineering (ECE) Ltd for helpful comments concerning simulation parameters and simulation environment, Nokia Networks for providing NetAct Planner tool for simulations, FM Kartta for providing the digital map, and the National Technology Agency of Finland for funding the work. Figure 5. Examples of typical pilot coverages of 65° and 33° horizontal beam width antennas in 6-sectored sites. Table 10. Results of capacity and performance analysis of 3-and 6-sectored solutions. 33* denotes shift of the first tier antennas. Parameter Antennas [°] Service probability[%] SHO [%] SfHO[%] UL load [%] UL i DL throughput [kbit/s/site] DL power [dBm]

7. References [1] Edoardo Amaldi, Antonio Capone, Federico Malucelli, Francesco Signori, UMTS Radio Planning: Optimizing Base Station Configuration, IEEE 56th Vehicular Technology Conference, vol. 2, 2002.

3-sec

6-sec

6-sec

6-sec

6-sec

65

65

33

33*

65&33

94.9

99.7

99.8

99.8

99.7

[2] Li-Chun Wang, Kapil Chawla, Larry J. Greenstein, Performance Studies of Narrow-Beam Trisector Cellular Systems, IEEE 48th Vehicular Technology Conference, vol. 2, 1998.

26.7 4.2 48.0 0.79

23.0 22.5 29.7 1.19

23.8 3.8 25.6 0.80

23.8 3.7 25.5 0.79

23.5 10.6 27.3 0.96

[3] Vladan M. Jovanović, James Gazzola, Capacity of Present Narrowband Cellular Systems: Interference-Limited or Blocking-Limited?, IEEE Personal Communications, vol. 4, issue 6, 1997.

1082

1252

1099

1099

1154

38.4

34.5

33.4

33.4

33.8

[4] Timothy X Brown, Cellular Performance Bounds via Shotgun Cellular Systems, IEEE Journal on Selected Areas in Communications, vol. 18, no. 11, 2000.

5. Discussion and conclusions In this paper, the impacts of UMTS network base station site location and antenna direction deviations were studied. Moreover, an optimum deployment strategy for network evolution from 3-sectored to 6-sectored sites from coverage and capacity point of view was found. The base station site location deviation was compared to regular hexagonal grid structure, and a small deviation was found to have a negligible impact on the overall performance of UMTS network; especially, if higher indoor coverage thresholds were required. Thus, it can be concluded that the base station site configurations should be optimized rather than base station site locations. However, if only a lower indoor coverage probability is enough (i.e., rural planning and less coverage overlapping), the deviation in the base station site locations is more crucial because it strongly affects the coverage. Also, the deviation in base station antenna directions was found to have a negligible impact on the network performance. The impact could have been more significant if wider antenna beam widths or 6-sectored sites were utilized.

[5] A.G. Spilling and A.R. Nix, Aspects of Self-Organisation in Cellular Networks, The Ninth International Symposium on Personal, Indoor and Mobile Radio Communications, PIMRC 1998, vol. 2, 1998. [6] Maciej J. Nawrocki, Tadeusz W. Wieckowski, Optimal Site and Antenna Location for UMTS – Output Results of 3G Network Simulation Software, 14th International Conference on Microwaves, Radar and Wireless Communications, MIKON-2002, vol. 3, 2002. [7] Jaana Laiho, Achim Wacker, Tomáš Novosad, Radio Network Planning and Optimisation for UMTS, John Wiley & Sons Ltd, 2002. [8] Jarno Niemelä, Jukka Lempiäinen, Impact of the Base Station Antenna Beamwidth on Capacity in WCDMA Cellular Networks, IEEE 57th Vehicular Technology Conference, vol. 1, 2003.

Suggest Documents