7/30/2019 EUCAP2011.pdf

1/5

Consideration of MIMO in the Planning of LTE

Networks in Urban and Indoor ScenariosOliver Stbler, Reiner Hoppe, Gerd Wlfle, Thomas Hager, Timm Herrmann

AWE Communications GmbH

Otto-Lilienthal-Strae 36, 71034 Bblingen, Germany

mail@AWE-Com.com

Abstract Broadband wireless access is emerging as one of the

hottest areas of growth within mobile communications. It enables

users to enjoy the same QoS they have at home, in the office or

wherever they go. 3GPP Long Term Evolution (LTE) is the latest

standard in the mobile cellular network technology. Innovative

wireless communication systems, such as LTE, are expected to

offer highly reliable broadband radio access in order to meet the

increasing demands of emerging high speed data and multimedia

services. For the planning of LTE networks, the investigation of

radio transmission in urban areas, but also within and into

buildings is getting more important.This paper introduces a deterministic approach for the

simulation and performance evaluation of LTE networks in

urban and indoor scenarios. Besides signal levels the expected

MIMO capacity is evaluated. Comparisons with two

measurement campaigns verify the high accuracy of the

presented prediction model.

Keywords LTE, MIMO channel, deterministic channel model,ray tracing, comparison with measurement data.

I. INTRODUCTIONAfter the great success of wireless communications used in

land and personal mobile radio networks, the growing demand

for high data rates and high reliability becomes more andmore important. Not only WLAN and WiMAX, but also 3G

and LTE networks with their wireless multimedia services

(e.g. video terminals) are used inside and outside buildingsand rely on the same high Quality of Service (QoS)

everywhere. 3GPP Long Term Evolution is able to cope with

those requirements using the MIMO technology in order to

guarantee high data rates and sufficient QoS. The LTE

specification provides downlink peak rates of at least 100

Mbit/s, uplink rates of at least 50 Mbit/s and RAN round-triptimes of less than 10 ms. As LTE uses OFDMA in downlink

and SC-FDMA for the uplink, scalable carrier bandwidths,

ranging from 1.4 MHz to 20 MHz are supported as well.

The performance of such wireless communication systemsdepends in a fundamental way on the mobile radio channel.As a consequence, predicting the propagation characteristics

between two antennas belongs to the most important tasks for

the radio planning of LTE networks. In order to calibrate and

evaluate the accuracy of the prediction models, comparisons

with measurement data are inevitable. The next paragraph

gives an overview of the deterministic ray tracing simulationapproach, which is used to predict LTE systems in a time

efficient and highly accurate way. The following sections

show comparisons of measured LTE channel data with

prediction results obtained from the 3D ray tracing

propagation model within an urban area and inside a building.

II. SIMULATION METHOLOGYA. 3D Ray-Optical Propagation Model

The mobile radio channel in urban and indoor areas ischaracterized by multi-path propagation. Dominant

propagation phenomena in these scenarios are the shadowing

behind obstacles, the reflection at the walls of buildings, the

wave guiding effects (due to multiple reflections) in street

canyons or corridors, and the diffractions at vertical and

horizontal wedges. The ray tracing algorithm used for thedeterministic modeling is based on the evaluation of 3D

building data representing the evaluated environment [1].

Figure 1: 3D vector database with propagation paths between transmitter and

receiver in an urban environment

The propagation model is fully three dimensional and

computes all rays with up to three interactions (incl. double

diffractions and combinations of reflections and diffractions).

The prediction of the path loss along the ray is computed with

the uniform theory of diffraction (UTD) and with the Fresnel

coefficients for the reflections. In order to accelerate the time-

consuming path determination the Intelligent Ray Tracing

(IRT) model can be utilized. The IRT is based on a pre-

processing of the building data, thus combining high accuracy

with short computation time [2], [3].

B. Post-Processing for Computation of MIMO ChannelsThe main disadvantage of the deterministic wave

propagation models is their excessive computation time. The

most time-consuming part is the determination of all relevant

paths between transmitter and receiver. To avoid large

computation times, the IRT (Intelligent Ray Tracing) is used

to predict the SISO channel impulse response between the

centers of the transmitter and the receiver antenna arrays. As

the spacing between the antenna elements of the arrays is

7/30/2019 EUCAP2011.pdf

2/5

rather small it can be assumed that the same propagation paths

exist for all antenna elements of the array and only the signalphases are changing from one element to the other (assuming

planar incidence of the waves) [4]. Considering such a

modular approach avoids re-computing the ray tracing

between all the antenna elements of the transmitter and the

receiver station during the generation of the MIMO channel

matrices.

III.MEASUREMENT COMPARISON IN URBAN ENVIRONMENTIn order to verify prediction results in urban environments,

results from a measurement campaign provided by the

University of Ilmenau [5] have been used for comparison.

A.Modeling of Simulation ScenarioThe urban simulation scenario [5] was modeled using a 3D

CAD model of the buildings located in the city center of

Ilmenau as well as the corresponding digital terrain elevation

database. Figure 2 depicts the simulation scenario together

with the location of the base station and the two receiver

trajectories, which have been taken into account for the

measurement comparison. Further details of the computationparameters and databases are summarized in Table 1.

Figure 2: Simulation scenario with transmitter location and receiver

trajectories

TABLE 1SIMULATION SCENARIO

Simulation area 1000 m x 1000 m

No. of vector buildings 4506

Min. building height 0.8 m

Max. building height 27.3 mMin. elevation 474.2 m

Max. elevation 519.7 m

Std. dev. of elevation 12.1 m

Resolution of prediction 5.0 m

The uniform linear antenna array of the base station is

located 26.5 meters above the ground level with an azimuthal

adjustment of 315 degrees and a down tilt of 10 degrees. The

two antenna elements are spaced 0.49 apart and radiate at a

centre frequency of 2.53 GHz with a total transmission power

of 46 dBm. The mobile receiver is equipped with a uniform

linear antenna array, which is oriented always perpendicularto the direction of movement. It is located on top of a vehicle

1.9 meters above street level. The car drives with a nearly

constant velocity along two different trajectories, which are

show in Figure 2. The first route (10b-9b) is approximately

123 meters long and covers locations with and without direct

line-of-sight between transmitter and receiver. The second,

upper trajectory (41a-42) is about 54 meters long and hasalways no line-of-sight between transmitter and receiver (cf.

Figure 3).

B.Results of ComparisonThe measurement routes coincide with 30 prediction pixels

for route 10b-9b and 14 prediction pixels for route 41a-42.

Since the predictions are computed for each pixel separately,

the measurement data is averaged and mapped to the

prediction pixels using the following procedure: First, each

measurement snapshot is mapped to the pixel, whose center is

closest to the location where the measurement snapshot was

recorded. In a second step, the median values of the received

power, delay spread and channel capacity values of allsnapshots mapped to the same pixel are calculated in order to

obtain one measured value [5] for each pixel to compare it

with the predicted results.

Figure 3: Predicted LOS status along the two receiver routes

The following graphs show the comparison betweenmeasured and predicted values of received power, delay

spread and channel capacity along the receiver trajectories

considering a 2x2 MIMO system. The two curves of the

graphs represent the measurement values (blue line) and theprediction values (red line), respectively. Prediction values are

depicted additionally in the scenario map together with thebuilding vector database of the surrounding.

The curves of the received power prediction show a good

agreement with the corresponding measurement values.

As the delay spread is a measure of the multi-path richness

of a wireless channel, it is a further crucial parameter for thecharacterization of MIMO channels used in broadband LTE

networks. Figure 7 and Figure 8 show the comparison of the

occurring delay spread along the two receiver trajectories.

7/30/2019 EUCAP2011.pdf

3/5

Figure 4: Received power along receiver route 10b-9b

Figure 5: Received power along receiver route 41a-42

Figure 6: Power predictions along the two receiver routes

Figure 7: Delay spread along receiver route 10b-9b

Figure 8: Delay spread along receiver route 41a-42

Figure 9: Delay spread predictions along the two receiver routes

The comparison of the delay spread values also turned out

to look very promising, especially for the receiver trajectory

10b-9b. The values of the statistical evaluation and the mean

prediction errors can be found in Table 2 and Table 3,

respectively.

Channel capacities of the 2x2 MIMO channels shown inFigure 10 and Figure 11 have been computed using a post-

processing step for the 3D ray tracing results of a SISO

channel. For the prediction a fixed mean signal-to-noise-ratio

of 15 dB was assumed.

Figure 10: Channel Capacity along receiver route 10b-9b

7/30/2019 EUCAP2011.pdf

4/5

Figure 11: Channel capacity along receiver route 41a-42

Figure 12: Channel capacity predictions along the two receiver routes

TABLE 2

STATISTICAL EVALUATION FOR MEASUREMENT (M) AND PREDICTION (P)

Mean Std. Dev.Channel

Parameter

Route

M P M P

Rx Power

[dBm]

10b-9b -50.8 -50.9 6.2 5.3

Rx Power

[dBm]

41a-42 -62.4 -62.5 2.2 2.1

Delay Spread

[ns]

10b-9b 173.4 172.4 75.5 70.6

Delay Spread

[ns]

41a-42 195.3 208.8 17.1 37.5

Channel Capacity

[bit/s/Hz]

10b-9b 6.1 6.3 0.2 0.3

Channel Capacity

[bit/s/Hz]

41a-42 6.3 6.5 0.1 0.2

TABLE 3MEAN PREDICTION ERRORS AND STANDARD DEVIATIONS (MEASUREMENT

VALUES HAVE BEEN SUBTRACTED FROM PREDICTION VALUES)

Channel Parameter Route Mean Std. Dev.Rx Power dBm] 10b-9b 0.0 1.7

Rx Power [dBm] 41a-42 0.1 0.7

Delay Spread [ns] 10b-9b 0.9 27.2

Delay Spread [ns] 41a-42 13.5 33.3

Channel Capacity [bit/s/Hz] 10b-9b 0.1 0.2

Channel Capacity [bit/s/Hz] 41a-42 0.2 0.2

Table 2 summarizes the results of the statistical evaluation

of the measurement data as well as of the prediction results

along the two receiver trajectories. The mean prediction errors

and the corresponding standard deviations are listed in Table 3.

These numbers refer to a case where measurement results

have been subtracted from ray tracing predictions. Thegraphical comparisons as well as the statistical evaluation

indicate a good matching between the predicted values and the

measurement data for both receiver trajectories.

IV.MEASUREMENT COMPARISON IN INDOOR ENVIRONMENTThe availability of high data rates and QoS inside buildings

becomes more and more important nowadays. The indoorcoverage provided by macro cellular LTE networks can be

improved substantially using indoor MIMO antenna systems,

as shown in the measurement campaign presented in [6].

A.Modeling of Simulation ScenarioThe results of these measurements taken in the 2.6 GHz

LTE test setup of the Fraunhofer Heinrich-Hertz-Institute in

Berlin [6] have been used for evaluation of the MIMO

prediction module [7]. The corresponding simulation scenario

is depicted in Figure 13. Further details are given in Table 4.

Figure 13: 3D vector database of simulation scenario

TABLE 4

SIMULATION SCENARIO

Simulation area 30 m x 30 m

No. of vector objects 164

Height of floor 4.0 m

Prediction height 1.5 m

Resolution of prediction 1.0 m

The indoor antenna system was fed from an urban macro

cell eNodeB, which is located on the left hand side of the

building, approximately 500 meter apart. Four different indoor

antenna configurations have been investigated in order to

identify the best solution for maximum radio coverage and

data rate on the depicted floor of the building. The following

sub sections show the comparisons between predicted (left

part) and measured (right part) maximum achievable data

rates for four different antenna configurations.

B. Scenario A: Single Antenna PairA simple approach to obtain radio coverage at indoor

locations, where no coverage from the outdoor macro cell is

available, is to deploy indoor antennas in these areas. As the

outdoor eNodeB is placed on the left hand side of the building,

Scenario A introduces two MIMO antenna elements in the

shadowed part of the building (cf. Figure 14) in order to

enhance the coverage in this area.

7/30/2019 EUCAP2011.pdf

5/5

Figure 14: Measurement comparison for maximum achievable data rate in

scenario A

C. Scenario B: Distributed AntennasDue to the floor plan of the building and the anticipated

wave guiding effects in the corridors, a system with two

distributed antennas radiating one spatial stream each was

evaluated as well. The comparison between prediction and

measurement is depicted in Figure 15 and shows a good

agreement.

Figure 15: Measurement comparison for maximum achievable data rate in

scenario B

D. Scenario C: Distributed Antenna PairsRadiating both spatial streams from two separated locations

further enhances the maximum achievable data rate in large

parts of the floor to a nearly optimum level as depicted in the

following figure.

Figure 16: Measurement comparison for maximum achievable data rate inscenario C

E. Scenario D: Interleaved AntennasThe scenario with four interleaved antennas provides the

best configuration regarding the maximization of the

achievable data rate. With this configuration, maximum

achievable data rates above 100 Mbit/s can be reached all over

the building floor.

Figure 17: Measurement comparison for maximum achievable data rate in

scenario D

The simulation results for the maximum achievable data

rate depicted on the left hand side of the Figures 14-17 show a

good agreement between measured and predicted values and

therefore proof the high simulation accuracy of WinProp [7]

in indoor environments.

V. SUMMARYThe baseline of the paper introduces a deterministic

approach for the simulation and performance evaluation ofLTE networks in urban and indoor scenarios.

In the second part simulation results predicted with a 3D

ray tracing model are compared to measurement data takenwithin an urban city center and inside a building. Based on the

deterministic simulation approach presented in this paper,

received power, delay spread and data rate predictions in

urban macro cellular and in indoor pico cellular propagation

environments are achieved with high accuracy in very short

simulation times.

VI.ACKNOWLEDGEMENTSThis work has been supported by the German Ministry for

Education and Research (BMBF) within the projectSIMPLON, which is kindly acknowledged.

REFERENCES

[1] R. Hoppe, G. Wlfle, P. Wertz, F. M. Landstorfer, Advanced Ray-Optical Wave Propagation Modelling for Urban and Indoor

Environments Including Wideband Properties, European Transactionson Telecommunications (ETT), January/February 2003 (Number

01/2003), January 2003.

[2] T. Rautiainen, G. Wlfle, and R. Hoppe: Verifying Path Loss andDelay Spread Predictions of a 3D Ray Tracing Propagation Model in

Urban Environments, 56th IEEE Vehicular Technology Conference

(VTC) 2002 - Fall, Vancouver, Canada, September 2002.[3] H. Zhang, O. Mantel, M. Kwakkernaat, M. Herben, Analysis of

Wideband Radio Channel Properties for Planning of Next-Generation

Wireless Networks, 3rd European Conference on Antennas andPropagation (EuCAP) 2009, Berlin, Germany, March 2009.

[4] O. Staebler, R. Hoppe, MIMO Channel Capacity Computed with 3DRay Tracing Model, 3rd European Conference on Antennas andPropagation (EuCAP) 2009, Berlin, Germany, March 2009.

[5] C. Schneider, G. Sommerkorn, M. Narandi, M. Kske, A. Hong, V.Algeier, W.A.Th. Kotterman, R. S. Thom, C. Jandura, Part I:Reference Campaign Description and Application, ITG WSA,

Berlin, Germany, 2009.

[6] O. Braz, J. Stefanik, T. Wirth, L. Thiele, T. Haustein, Mimo bringtMobilfunk ins Haus, Funkschau 07/2010, pp. 44 47, July 2010.

[7] AWE Communications: WinProp Software Package. Free evaluationversion of a 3D ray tracing tool for urban and indoor environments.Available: http://www.awe-communications.com.