Sei sulla pagina 1di 5

Comparing RSRP, CQI, and SINR measurements with Predictions for Coordinated and Uncoordinated LTE small cell Networks

Jay Weitzen

Department of Electrical and Computer Engineering University of Massachusetts Lowell Lowell, Ma, USA 01854 Airvana, Chelmsford Ma 01824

Abstract- This paper presents results from the first measurements comparing the coverage and coverage quality of a network of coordinated LTE small cell radio points compared to a cluster of independent standalone LTE small cells. The measurements are compared to theoretical predictions and show that the coordinated small cell provides significant improvement in the overall quality of LTE coverage as measured by RSRP, SINR, CQI and MCS. The indoor measurements described in this paper were made with a new autonomous robotic system designed for making indoor wireless coverage measurements.

Keywords-LTE; wireless networks; small cells

I.

INTRODUCTION

Airvana "OneCell" is a hybrid technology incorporating elements of classic small cells and Distributed Antenna Systems. It consists of a cluster of LTE small cell radio points that are connected via Ethernet switches and cat-5 Ethernet cable to a coordinating central control unit so that they function as if they were a single distributed LTE eNodeB. The cluster of radio points provides uniformity of indoor coverage and in-cell spatial diversity and multiplexing instead of inter-cell handoff regions and interference associated with standalone small cells. The distributed single cell combined with the ability to re-use subcarriers on spatially separated radio points creates macro-level spatial reuse, increasing the capacity of the single eNodeB cell over a conventional discrete cell. The advantages of coordinated small cells have been discussed theoretically in the literature [1], but to date there are no measurements to verify some of the key elements of the technology. The experiments described in this paper demonstrate the uniformity of high quality coverage due to the fact that signals combine rather than interfere in the regions between radio points as they would with discrete small cells.

This work was sponsored by Airvana, Chelmsford Ma

Rachel Wakim, Erin Webster

Department of Electrical and Computer Engineering University of Massachusetts Lowell Lowell, Ma, USA 01854

II. PREDICTIONS OF COVERAGE QUALITY FOR COORDINATED AND UNCOORDINATED LTE SMALL CELLS

To help investigate and understand the properties of coordinated small cells, an indoor test network was set up consisting of up to 3 radio points as shown in Figure 1. The location of the radio points was chosen to create conditions of relatively close-spaced radio points to investigate interference and coverage overlap and other radio points that are RF isolated to allow for carrier reuse experiments. The single story office building test bed consists of a mix of standard cubicles plus walled offices and laboratories in a multiple­ wing building. Each radio point (RP) uses 2x2 cross polarization MIMO with the transmitter powers set to 10 dBm per antenna to prevent leakage and interference outside of the building (each transmitter is capable of 23 dBm). The LTE metrics of interest in this paper are overall received signal strength metrics, primarily RSRP (Reference Signal Received Power), and signal quality metrics including overall SINR (signal to noise plus interference ratio both reference signal and broadband), CQI (Channel Quality Index), MCS (modulation and coding scheme), and physical layer throughput (PDSCH throughput). RSRP is LTE signal strength metric in dBm that measures the average received power over the resource elements that carry cell-specific reference signals. RSRP measurements are applicable in both idle and connected modes and measurements are independent of loading and interference while received signal strength (RSSI) can vary with load, CQI is a number between 0 and 15 which is related to the Signal to noise plus interference ratio and influences the modulation and coding used (Modulation and Coding Scheme, MCS, is a number between 0 and 30 used and in essence determines the effective throughput). In this paper we focus on RSRP, reference signal SINR, CQI, MCS and PDSCH throughput because they are indicative of

978-1-4799-7473-3115/$31.00 ©2015 IEEE From IEEE International Conference on Microwaves, Communications, Antennas and Electronic Systems

(COMCAS 2015), 2 - 4 November 2015, Tel Aviv, Israel

user experience and relatively easy to measure using industry standard equipment.

In the series of experiments described in this paper, the individual LTE radio points were programmed to transmit in LTE band 25 (PCS G block, 1995 MHz) using standard 5 MHz LTE signals and were connected to the controller via standard copper cat-5 Ethernet. Timing synchronization within the cluster of radio points is provided via IEEE-1588 from a centralized GPS source. The centralized controller schedules packets for each radio point (RP) and simulcasts them over the downlink stream. The controller also provides uplink combining from mUltiple radio points. A data server provides continuous UDP data streaming to keep the handset in connected mode and scheduling buffers full. In the experiments described we considered only one UE handset device that is using the entire bandwidth of the system. This series of experiments were designed to help us understand the quality and uniformity of coverage provided by coordinated smallcells. We first modeled and predicted coverage and coverage quality using the AWE Radio Propagation Tool [2] and then made a series of measurements to compare to predictions. We considered the following scenarios and experiments:

1) Predictions and comparison with measurements of basic coverage (RSRP) and signal quality (SINR and CQI) for three individual radio points (RP1, RP2, RP3), one at a time. These measurements were used to calibrate the materials data base in the AWE

prediction tool, to verify that the individual radio points were functioning correctly, and that the key metrics could be accurately predicted. 2) Predictions and measurements of basic coverage (RSRP) and signal quality (SINR, MCS and CQI) for

3 radio points operating in uncoordinated mode as 2

individual standalone individual eNodeB (3 RP, 2 Controllers, two different PCI). The two different PCI (index1=200 and index2=206) were set so that they were the same modulo 6 to ensure that the SINR measured on the reference channels would be independent of system loading. In this configuration

Reference signals are sent on the same resource elements of all units independent of which controller they are homed to. 3) Predictions and measurements of basic coverage (RSRP) and signal quality (SINR, MCS and CQI) for

3 radio points operating in coordinated mode with a

single controller and single PCI simulcasting on the downlink stream.

Coverage and signal quality calculations were performed using the AWE radio propagation tool [2] that was configured using the "dominant path" indoor propagation model [3] with a general propagation exponent of 2.1. One half square meter pixels and a UE height of 0.5 meter for the receiver were used. Figure 2 plots the predicted RSRP in dBm for a) three RP

operating as a unified coordinated small cell and b) three RP operating as independent small cells. Compared to the uncoordinated mode, it is observed that in the region between the two radio points, the RSRP is approximately 3 dB higher for the radio points operating as a unified coordinated small cell. It is in this region that handoff would be occurring when operating as standalone units. The primary difference between deploying a cluster of conventional standalone LTE small cells in which each small cell functions as an independent eNodeB with a different physical cell identity (PCI), and coordinated small cells forming one large virtual cell with a single PCI, is that there are no interior cell boundaries or inter­ cell handoff regions. Figure 3 shows predicted signal quality as measured by SINR for uncoordinated and coordinated small cells assuming high loading on the system. The background noise plus interference level is measured at approximately -87 dBm with a bit more interference observed near the windows, and slightly less interference in the building interior.

1 .-
1
.-

r - -

Figure 1 Layout of 3 radio points (RPI through RP3) in single story mixed use office building.

OneC,_
OneC,_

Coordinated

Sma
Sma

0eII

mixed use office building. OneC,_ C o o rd in at ed Sma 0eII Standalone Small
Standalone Small C�II Cluster
Standalone Small C�II
Cluster

Figure 2. Predicted RSRP in dBm calculated for 3 in coordinated mode, uncoordinated modes (it is approximately the same)

l OneCell Coordinated Small Cell Standalone Small Cell Cluster
l
OneCell Coordinated
Small Cell
Standalone Small Cell
Cluster

Figure 3 Predicted SINR for (a) coordinated and (b) uncoordinated small cells. Note the difference in SINR in the regions between the smallcells.

Comparing Figures 3a and 3b we observe that the predicted SINR for the coordinated small cell is significantly greater because signals combine instead of interfere as they do with stand alone small cells. Signal quality determines the LTE physical layer data rate and as shown in Figure 3b, in the handoff region between radio points, the SINR of the standalone small cell cluster is on the order of 0 dB, while in the coordinated small cell shown in Figure 3a, it is almost uniformly greater than 15 dB.

III. INDOOR WIRELESS COVERAGE MEASUREMENT SYSTEM BASED ON AN AUTONOMOUS ROBOT

Numerous repetitive measurements in the in-building test bed are required as the product is developed. The conventional method for making indoor coverage measurements [4, 5] is to create a series of waypoints on a building map. As a tester walks a linear track from waypoint to waypoint, the tester enters the waypoints into the logging system. The measurements are then positioned along the line connecting the waypoints using linear interpolation. This measurement process is time consuming, labor intensive, and it is hard to reproduce conditions from test to test. It is also hard to create more complicated or involved coverage mapping routes because interpolation between waypoints is required. To create a system that provides a significant improvement in terms of both accuracy and repeatability over manual measurement techniques, we designed an autonomous robotic coverage mapping system with the following capabilities.

Explore an unknown area.

Accurately localize itself relative to an origin, creating a map that can be compared to predictions.

Collect wireless measurement data in real-time while the system is navigating. Produce high-resolution maps of coverage and evaluations of signal quality metrics.

The system used collect the data described in this paper is designed to simultaneously eliminate many of the shortcomings of labor-intensive manual methods while improving the overall accuracy of the indoor measurement process. The architecture of the autonomous measurement system is shown in Figure 4. The robotic system is designed around a commercially available two-wheeled, battery­ operated, circular robot platform. In addition to wheeled odometry, the robot contains a built-in gyroscope to assist with localization and an external laser for range finding and mapping. Mounted on the robot are a series of pillars holding up platforms for placing a standard wireless measurement system that is integrated with the "Gmapping" localization software [6] in the robot platform.

localization software [6] in the robot platform. Figure 4. Architecture of the Robotic Coverage mapping

Figure 4. Architecture of the Robotic Coverage mapping system.

The wireless measurement sub-system consists of a standard commercial LTE mobile phone device (UE) that has been laboratory calibrated and is connected via USB to a notebook computer running purpose-built data logging software. The temporal resolution of the data logging is up to 2 ms and the resolution of the robotic localization is 1 second or approximately 1 to 1.5 feet. Post-processing of the measured data uses the process described in Weitzen and Grosch [5]. The post processing software allows data to be imported into the AWE tool [2] for comparison with predictions or displayed on maps produced by the robotic mapping functionality. The system uses a local [x,y] coordinate system that can then be geocoded for input into other types of tools such as IBwave.

uses a local [x,y] coordinate system that can then be geocoded for input into other types
Figure 5. Robotic Autonomous Coverage Measurement Platform IV. MEASUREMENTS OF COVERAGE QUALITY FOR COORDINATED AND

Figure 5. Robotic Autonomous Coverage Measurement Platform

IV. MEASUREMENTS OF COVERAGE QUALITY FOR

COORDINATED AND UNCOORDINATED SMALL CELLS

Using the autonomous wireless coverage tool, data sets were collected for each of the scenarios described in Section I and compared to the predictions described in Section II. In each set of measurements, the robot traversed the same area of the building. Each data measurement set consisted of about 150,000 individual measurement points (RSSI, RSRP, SINR MCS, PDSCH throughput, and CQI) along with other data. All measured points that reside within the same 0.5 square meter prediction bin are averaged to form a bin average that is plotted. Figures 6a and 6b show RSRP measured over the coverage area of the two radio points for uncoordinated and coordinated small cell modes. The average error between the predictions and measurements was observed to be approximately +2.3 dB (measurements higher than predictions), with the error log normally distributed. In coordinated small cell mode, there was a single controller for the cluster of radio points. All downlink packets were simulcasted. In uncoordinated mode, multiple controllers allowed each radio point to function as a standalone small cell with an independent Primary Cell Index (PCI). Figures 7a and 7b plot the measured SINR for the uncoordinated and coordinated small cells respectively. The figure shows that the coordinated small cell has a consistently high signal quality since there are no cell boundaries in between the two radio points. Figure 8 plots and compares the cumulative distribution function (CDF) of the measured SINR over the route traversed for both modes of operation: three radio points operating as a single coordinated small cell, and three radio points as individual small cells. The measurements show a very high probability of good signal quality for the coordinated mode and significantly poorer coverage quality for uncoordinated mode. For the coordinated small cell mode

of operation, the distribution of measured SINR values over the covered area was normally distributed with mean 24.3 dB and standard deviation of 4.87 dB. For the uncoordinated small cell mode, the average SINR was 15.3 dB with a standard deviation of7.5 dB.

Finally, Figure 9 shows a series of static application layer measurements in which the system stopped for 2 minutes to do simultaneous uploads and downloads. It shows the uniformity of data rates, even in the region between radio points that in a conventional deployment would have high interference and low data rates.

-., (c5m) -6J.00 -65.00 -70,00 -75.00 �.oo -85.00 - 100.00 -90,00 - -95.00 - ·105,00
-.,
(c5m)
-6J.00
-65.00
-70,00
-75.00
�.oo
-85.00
- 100.00
-90,00
-
-95.00
-
·105,00
-110.00
-115.00
-120,00
·125,00
-130,00
- ·105,00 -110.00 -115.00 -120,00 ·125,00 -130,00 Figure 6 a and b. Measured RSRP for 2

Figure 6 a and b. Measured RSRP for 2 RP operating in coordinated and uncoordinated modes.

lJ.oo ;SOO (� ( "tJ11J 1511J - 10.00 5,00 000 -5 00 -10 11J
lJ.oo
;SOO (�
(
"tJ11J
1511J
-
10.00
5,00
000
-5
00
-10 11J

Figure 7 a and b. Measured received SINR for coordinated and uncoordinated small cells

03 0.1 0 1 0.& 05 OJ O� 02 01 . 1\ ·10 . ,
03
0.1
0 1
0.&
05
OJ
O�
02
01
. 1\
·10
. ,
,
10
1\
10
�H'Id81

Figure 8. Cumulative Probability Distribution of measured SINR for 3 RP over the coverage area for coordinated and uncoordinated modes.

the coverage area for coordinated and uncoordinated modes. Figure 9. Single User Throughput measurements as a

Figure 9. Single User Throughput measurements as a function of location throughout the coverage region using udp streaming server.

V.

CONCLUSIONS

The measurements presented represent the first measurements of a new type of indoor coverage solution consisting of a coordinated cluster of smallcell radio points that eliminate the conventional intercell boarder. The experiments verify experimentally that the coordinated smallcell eliminates the cell edge interference effects seen with a cluster of conventional standalone small cells and provides high quality, uniform coverage. With -87 dBm average background interference and 10 dBm transmitters, an average SINR of over 16 dB was observed throughout the measurement area, especially in the region between the radio points that would have close to 0 dB SINR. Highly uniform coverage quality as measured by SINR in the inter radio point region was observed with the coordinated small cell, whereas significant inter-cell interference was observed when radio points operated in standalone mode.

References

[1] Jungnickel, V., Manolakis, K. ; Zirwas, W. ; Panzner, B. ; Braun, V. ; Lossow, M.; Sternad, M. ; Apelfrojd, R. ; Svensson, T, The role of small cells, coordinated multipoint, and massive MlMO in 5G, IEEE Communications Magazine, May 2014, 44 - 51,

[2]

AWE Communications, Germany. Software tool WinProp for the planning of mobile communication networks,; 2013

[3]

Woelfle, G, .Wertz, P, Landsdorfer F. M, Performance, Accuracy and Generalization Capability of indoor propagation models in different types of buildings, 10 th IEEEE international Symposium on Personal Indoor and Mobile Radio Communications (PIMRC) 1999, Sept 1999, Osaka Japan

[4]

Morton Tolstrup, "Indoor Radio Planning, a practical guide for GSM, DCS, UMTS, HSPA and LTE", John Wiley. 2n d Edition, 2011.

[5]

Weitzen, Jay A. and Grosch, Theodore, "Comparing Coverage Quality for Femtocell and Macrocell Broadband Data Services", IEEE Communications Magazine, January 2010

[6] Giorgio Grisetti, Cyrill Stachniss, and Wolfram Burgard: Improved Techniques for Grid Mapping with Rao-Blackwellized Particle Filters, IEEE Transactions on Robotics, Volume 23, pages 34-46, 2007