experiences of the live wcdma network in stockholm, · pdf filetpc transmission power control...

12
Ericsson’s WCDMA evaluation system Ericsson’s WCDMA evaluation system was developed to meet the demands of wireless mobile communication in a true multi- media environment, where high-speed packet data and the Internet bearer service play major roles. It has been supplied to sev- eral telecommunications operators all over the world to give them hands-on experience of WCDMA technology. At Ericsson, the system has been used for purposes of evalu- ation and demonstration. The WCDMA evaluation system is composed of several radio network nodes: The mobile switching center (MSC) offers simplified call control and mobility man- agement. Its main task is to set up and re- lease calls to and from mobile stations. • The radio network controller (RNC) houses functions for controlling the radio network—the node establishes and re- leases connections, manages handover, and controls transmission power—and for managing radio resources. Also located in this node are the codec and diversity- combining devices, for use during soft handover. The radio network controller supports the A-interface connection to the mobile switching center in a GSM net- work, which enables voice calls to be made between the GSM system and a WCDMA terminal. Roaming to and from GSM has been demonstrated using this system. The structure of the base transceiver sta- tion (BTS) allows resources to be pooled. This provides the flexibility that is re- quired for packet-switched services and for the dynamic allocation of bandwidth. The base transceiver station has the de- gree of flexibility that is necessary in an evaluation system—for example, the number of rake fingers and the searcher characteristics can be changed. This flex- ibility is also carried over into the radio network algorithms, in order to facilitate experiments with alternative solutions. • The mobile station simulator (MS-SIM) is only used for testing purposes, and has not been optimized in terms of size or weight. In principle, it is based on the same technologies as the base transceiver station. The WCDMA evaluation system in Stockholm The WCDMA evaluation system in Stock- holm was composed of one field system and one lab system (Figure 1). The field system 204 Ericsson Review No. 4, 2000 Experiences of the live WCDMA network in Stockholm, Sweden Peter Almers, Anders Birkedal, Seungtai Kim, Anders Lundqvist and Anders Milén The joint Ericsson and Telia wideband code-division multiple access (WCDMA) evaluation project has given valuable experience in terms of live system performance and general system operation. It has also been extremely valuable in providing early feedback and validating working assumptions for the ongoing development of Ericsson’s commercial third- generation WCDMA system. Furthermore, a number of important system parameters have been fine-tuned to optimize capacity and coverage. The project has also presented a great opportunity for gathering the technical knowledge necessary to support future business cases, and to provide input to radio network planning activities. The authors describe a small part of the tests and experiments con- ducted during the course of the project. Ericsson’s WCDMA evaluation system was built according to a draft of the Association of Radio Indus- tries and Businesses (ARIB) WCDMA specifications, which has since evolved into the international WCDMA standard by the Third-generation Partnership Project (3GPP). The system’s lower layers are more consis- tent with current 3GPP technical specifications than are the higher layers, and that is why the physical layer has been the focus of the tests. 3GPP Third-generation Partnership Project ARIB Association of Radio Industries and Businesses BER Bit error rate BTS Base transceiver station C/I Carrier-to-interference ratio CRC Cyclic redundancy check DTX Discontinuous transmission E b Energy per bit FER Frame error rate GPS Global positioning system GSM Global system for mobile communication I o Interference energy ISSI Interference signal strength indica- tor MSC Mobile switching center MS-SIM Mobile station simulator RNC Radio network controller RSSI Received signal strength indicator TPC Transmission power control (power control command) WCDMA Wideband code-division multiple access WOS WCDMA operating system BOX A, TERMS AND ABBREVIATIONS

Upload: phamnhu

Post on 10-Mar-2018

220 views

Category:

Documents


3 download

TRANSCRIPT

Page 1: Experiences of the live WCDMA network in Stockholm, · PDF fileTPC Transmission power control (power control command) WCDMA Wideband code-division multiple access WOS WCDMA operating

Ericsson’s WCDMAevaluation systemEricsson’s WCDMA evaluation system wasdeveloped to meet the demands of wirelessmobile communication in a true multi-media environment, where high-speedpacket data and the Internet bearer serviceplay major roles. It has been supplied to sev-eral telecommunications operators all overthe world to give them hands-on experienceof WCDMA technology. At Ericsson, thesystem has been used for purposes of evalu-ation and demonstration. The WCDMAevaluation system is composed of severalradio network nodes:• The mobile switching center (MSC) offers

simplified call control and mobility man-agement. Its main task is to set up and re-

lease calls to and from mobile stations.• The radio network controller (RNC)

houses functions for controlling the radionetwork—the node establishes and re-leases connections, manages handover,and controls transmission power—and formanaging radio resources. Also located inthis node are the codec and diversity-combining devices, for use during softhandover. The radio network controllersupports the A-interface connection to themobile switching center in a GSM net-work, which enables voice calls to be madebetween the GSM system and a WCDMAterminal. Roaming to and from GSM hasbeen demonstrated using this system.

• The structure of the base transceiver sta-tion (BTS) allows resources to be pooled.This provides the flexibility that is re-quired for packet-switched services andfor the dynamic allocation of bandwidth.The base transceiver station has the de-gree of flexibility that is necessary in anevaluation system—for example, thenumber of rake fingers and the searchercharacteristics can be changed. This flex-ibility is also carried over into the radionetwork algorithms, in order to facilitateexperiments with alternative solutions.

• The mobile station simulator (MS-SIM)is only used for testing purposes, and hasnot been optimized in terms of size orweight. In principle, it is based on thesame technologies as the base transceiverstation.

The WCDMA evaluationsystem in StockholmThe WCDMA evaluation system in Stock-holm was composed of one field system andone lab system (Figure 1). The field system

204 Ericsson Review No. 4, 2000

Experiences of the live WCDMA network inStockholm, SwedenPeter Almers, Anders Birkedal, Seungtai Kim, Anders Lundqvist and Anders Milén

The joint Ericsson and Telia wideband code-division multiple access(WCDMA) evaluation project has given valuable experience in terms of livesystem performance and general system operation. It has also beenextremely valuable in providing early feedback and validating workingassumptions for the ongoing development of Ericsson’s commercial third-generation WCDMA system. Furthermore, a number of important systemparameters have been fine-tuned to optimize capacity and coverage.

The project has also presented a great opportunity for gathering thetechnical knowledge necessary to support future business cases, and toprovide input to radio network planning activities.

The authors describe a small part of the tests and experiments con-ducted during the course of the project. Ericsson’s WCDMA evaluationsystem was built according to a draft of the Association of Radio Indus-tries and Businesses (ARIB) WCDMA specifications, which has sinceevolved into the international WCDMA standard by the Third-generationPartnership Project (3GPP). The system’s lower layers are more consis-tent with current 3GPP technical specifications than are the higher layers,and that is why the physical layer has been the focus of the tests.

3GPP Third-generation Partnership ProjectARIB Association of Radio Industries and

BusinessesBER Bit error rateBTS Base transceiver stationC/I Carrier-to-interference ratioCRC Cyclic redundancy checkDTX Discontinuous transmissionEb Energy per bitFER Frame error rateGPS Global positioning systemGSM Global system for mobile communication

Io Interference energyISSI Interference signal strength indica-

torMSC Mobile switching centerMS-SIM Mobile station simulatorRNC Radio network controllerRSSI Received signal strength indicatorTPC Transmission power control (power

control command)WCDMA Wideband code-division multiple

accessWOS WCDMA operating system

BOX A, TERMS AND ABBREVIATIONS

Page 2: Experiences of the live WCDMA network in Stockholm, · PDF fileTPC Transmission power control (power control command) WCDMA Wideband code-division multiple access WOS WCDMA operating

Ericsson Review No. 4, 2000 205

included omnidirectional and multisectorbase transceiver station sites. In the lab sys-tem, the base transceiver stations were con-nected to mobile station simulators via ca-bles and instrumentation that simulate theradio environment.

As part of the test equipment, all nodesin the lab and field systems, including themobile station simulator, were fitted withglobal positioning system (GPS) equip-ment, to provide accurate location and timeregistration for the test data.

Obviously, measurement capabilitiesmust be present in an evaluation system.The measurement data accumulated by thesystem was collected at appropriate pointsby the WCDMA operating system (WOS),which also provided the operation and main-tenance interface.

System services and capacity The test system offered voice and data ser-vices as well as multicall capability (Tables1 and 2). Therefore, besides making simul-taneous voice and data retrieval possible,tests could be conducted with new and in-teresting multimedia applications.

Voice service: 8 kbit/s using a codecCircuit-switched data: 64 to 384 kbit/sPacket-switched data: up to 470 kbit/sMultiservice support for mobile terminals

TABLE 1, SYSTEM SERVICES ANDCAPACITY

Downlink frequency band 2110 to 2130 MHzUplink frequency band 1920 to 1940 MHzMaximum BTS output power 20 W per sectorPower control cycle 0.625 ms Power control step 1 dBChip rate 4.096 McpsSpreading factors 16 to 256

For testing purposes, the Perch and DPCH32 channels were defined and used throughout the eval-uation. The Perch channel is a combined broadcast and synchronization channel; DPCH32 is adedicated physical channel at 32 kilosymbols/s that carries speech.

TABLE 2, SYSTEM CHARACTERISTICS

MS-SIMs

BTSRNC MSC

RNC MSC

BTS 2

BTS 1

BTS 3

Lab system

Field system

PSTNInternet

MS-SIM with WOS and GPS

Lab view

Lab LAN

Field LAN

WOS

WOS

GPSInstrumentation

Figure 1Overview of the WCDMA evaluation sys-tem in Stockholm.

Page 3: Experiences of the live WCDMA network in Stockholm, · PDF fileTPC Transmission power control (power control command) WCDMA Wideband code-division multiple access WOS WCDMA operating

Detector performance testsLaboratory testing

Test configuration

A channel emulator was used to simulate theradio interface using five channel models(Table 3). Different mobile speeds could besimulated with the channel emulator, thuschanging the fading speed.

Uplink performance within receiverdynamic range

A noise generator was used to introduce in-terference to test uplink performance with-in the receiver’s dynamic range. Figure 2shows the test set-up.

The output from the mobile station sim-ulator was fixed, whereas the backgroundnoise was varied using the noise generator.The noise effect was measured when theframe error rate (FER) was 2%. The carrier-to-interference ratio (C/I) could then be cal-culated.

For signal levels –90, –100, –110 and–120 dBm, FER = 2% was obtained at C/I = –20 dB using receiver diversity in theradio base station. Without diversity, thecorresponding value was –17 dB. The the-oretical value of the C/I needed for aDTCH32 channel can be calculated as fol-lows:

With a processing gain of 18 dB, the C/Ishould be –21 dB. This is in agreement withthe measurements. The experiment con-firmed that no unexpected limitations ex-isted in the system.

Influence of available rake fingers onperformance

Because the maximum number of rake re-ceiver fingers set in a base transceiver sta-tion affects the cost of the equipment, it isa good idea to minimize their number. Inthe test system, eight rake fingers could beused for each channel. Provided diversitywas used, on the average, four fingers couldbe used per channel. Two tests were con-ducted: a sensitivity test and a test to de-termine the effects of power control.

The sensitivity test measured sensitivityin the uplink as described above—exceptthat the maximum number of rake fingerwas altered and the power control was dis-abled. Figure 3 shows the sensitivity degra-dation relative to an AWGN channel for dif-ferent settings of the maximum number of

Eb – coding gain – diversity gain =4.7–4.7–3=–3dBIo

206 Ericsson Review No. 4, 2000

Channel DescriptionAWGN Averaged white Gaussian noise,

“ideal channel”RAY4 Four fading rays with equal aver-

age amplitudeHA Vehicle test environment. High

antenna IO Indoor office test environmentIP Outdoor-to-indoor and pedestri-

an test environment

TABLE 3, CHANNEL MODELS

RAY4 IPIOHA0

1

2

3

4

5

6

dB

Channel type

6 rays 4 rays8 rays

Figure 3Sensitivity degradation relative to theAWGN channel.

MS-SIM

Constant output power

BTS

Noise generator

Measured UER=2%

Figure 2Test set-up for uplink performance withinthe receiver’s dynamic range.

rake fingers. The fading velocity was 100km/h. The decrease in sensitivity, when thenumber of available rake fingers was reducedfrom eight to six, was within the measure-ment error. However, when the number ofavailable fingers was reduced further, say,down to four, the RAY4 channel indicateda 2 dB decrease in sensitivity. The othermodels, which had fewer rays, were basical-ly unaffected.

In the second test, power control was usedwith a relatively high Eb/Io target of 10 dB.The power was set so that the power controlcould work in its dynamic range, that is,without saturation. The maximum numberof rake fingers allowed was then set, and theuplink Eb/Io was measured. Table 4 showsthe outcome of the second test. The mea-sured Eb/Io is shown for different settingsof the maximum number of rake fingers.

The last column shows the mean numberof rake fingers actually used. With low fad-ing speed, the spread in the measured Eb/Iovalues was almost independent of the num-ber of available rake fingers.

With fast fading signals, more rake fin-gers were required. Moreover, we saw thatthe Eb/Io spread increased when the num-ber of available fingers was reduced. The in-crease was small between eight and six, butlarger between six and four. There was alsoa slight increase in the average Eb/Io valuemeasured. Because variations in the averageand deviation influenced the power used perchannel, capacity was also influenced.

Note that the readings (Table 4) apply toa dispersion model in which the total energywas equally distributed between four rays.Models with fewer rays will exhibit less sen-sitivity to the number of rake fingers allowed.

Effects of downlink loading

A test was conducted to determine how theperch channel and the DPCH32 channel af-fect one another. The power control was dis-abled and Eb/Io was measured on differentcombinations of perch and DPCH32 power.

Figure 4 shows the measured Eb/Io values.The numbers shown outside of the bracketsstand for the power (in dB) of the measuredchannel; the numbers shown inside thebrackets stand for the level of load. The solidlines show measurements on the DPCH32channel; the dashed lines represent the cor-responding values of the Perch channel. Thetheoretical value of the expected Eb/Io canbe calculated asEb = Gp ⋅ Pchannel ⋅ PloadIo

Page 4: Experiences of the live WCDMA network in Stockholm, · PDF fileTPC Transmission power control (power control command) WCDMA Wideband code-division multiple access WOS WCDMA operating

Ericsson Review No. 4, 2000 207

where Gp is the processing gain, Pchannel is thepower of the measured channel, and Pload isthe power of the loading channel. Table 5shows the calculated and measured values.The measured levels confirmed our expec-tations. That is, the measured Eb/Io valuewas dependent on downlink loading. Thisrelationship has been taken into considera-tion in the algorithms that use Eb/Io mea-surements (power control, cell selection, andso on).

Field tests Measurements in the field system were madefrom a van travelling at a velocity of 50 to60 km/h. These measurements were then an-alyzed in terms of • self-interference from base transceiver sta-

tion output power;• uplink coupling loss; and• actual and target Eb/Io measurements.

Self-interference from BTS output power

We obtained a correlation factor of 0.78when we correlated the measured outputpower with the measured received interfer-ence level (interference signal strength in-

Maximum Eb/Io Eb/Io Eb/Iono. of rake Velocity mean standard median Eb/Io Multipathfingers km/h value deviation value 90%† mean8 3 10.00 0.57 10.05 1.60 4.996 3 10.01 0.46 10.05 1.50 4.684 3 10.00 0.56 10.05 1.60 3.758 100 10.43 0.75 10.45 2.40 7.346 100 10.46 0.79 10.45 2.60 6.004 100 10.68 0.96 10.65 3.20 4.00† spread for 90% of all measured data

TABLE 4, Eb/Io VERSUS NUMBER OF RAKE FINGERS

Eb/Io Eb/Io Eb/Io Eb/IoPerch DPCH Perch Perch DPCH DPCHpower power measured calculated measured calculateddB dB dB dB dB dB35 26 >30 34 17 1435 15 >30 43 3 326 15 28 34 12 1226 26 24 23 23 23

TABLE 5, Eb/Io VALUES FOR DIFFERENT LOAD CASES

Eb/Io [dB]Eb/Io for CH1 with CH2 present [Level signal (level load)]

Input power [dBm]

0

5

10

15

20

25

30

–120 –110 –100 –90 –80 –70 –60 –50 –40

35(15)

26(15)

26(26)

26(26)

26(35)

15(26)

15(35)

35(26)

Figure 4The effects of load on Eb/Io.

Page 5: Experiences of the live WCDMA network in Stockholm, · PDF fileTPC Transmission power control (power control command) WCDMA Wideband code-division multiple access WOS WCDMA operating

dicator, ISSI) in the base transceiver station(Figure 5). Although the uplink and down-link frequency bands are separated by 190MHz, the transmitter output interferedwith the input. The leakage from the out-put was suppressed by approximately 120dB. Because the received signal strength in-dicator (RSSI) level in the uplink was heldfairly constant, the interference due to leak-age affected capacity.

Uplink coupling loss

Figure 6 shows the coupling loss between themobile station simulator and the base trans-ceiver station—that is, the difference be-tween transmission power (dBm) of the mo-bile station simulator and RSSI. The varia-tion in RSSI was small: –119.8 to –125.3dBm. Thus, the graph also describes the mainfeatures of uplink power control. Due to ahigh correlation between the long-term fad-ing channel for the uplink and downlink fre-quency bands, the variations in transmissionpower of the mobile station simulator werequite similar to that of the base transceiverstation (Figure 5).

Eb/Io—actual and target values

The mobile station power control serves toreduce the difference between measured andtarget Eb/Io. We measured the average Eb/Ioduring one second, and over an averagelength of at least 90 wavelengths (1s, 50km/h, λ = 0.15 m). The measured averageEb/Io was 0.44 dB higher than the targetEb/Io. The standard deviation was Σ = 0.75dB. Note that the short-term variationscould be much higher (Figure 7).

The number of active rake fingers indicatesthe channel quality. Many active rake fingersmeans that many reflections are present. Fig-ure 8 shows that the target Eb/Io value in-creased with the number of active rake fin-gers. If the same channel is used for a differ-ent number of rake fingers, Eb/Io will increaseas more fingers are assigned in the receiver.That is, the greater the number of rake fin-gers in the receiver (increased receiver com-plexity), the better the performance.

Power control tests

General investigation of the power-control function (field test) To avoid interference from other cells, theoutput power from other base stations wasreduced to a minimum level. During thetests, the velocity of the mobile station sim-

208 Ericsson Review No. 4, 2000

dBm, dBµV

Time [s]0 200 400 600 800 1000 1200

5

10

15

20

25

30

ISSI [dBµV]

BTS Tx power [dBm]

Figure 5Transmission power and received interference in the base transceiver station, power leak-age between the output and input signals.

Time [s]0 200 400 600 800 1000 1200

80

90

100

110

120

130

140

Uplink coupling loss [dB]

Min CL = 84.8 dB Max CL = 131.3 dB Mean CL = 103.8 dB

Figure 6The uplink coupling loss, transmission power of the mobile station less RSSI +113 [dB].

Page 6: Experiences of the live WCDMA network in Stockholm, · PDF fileTPC Transmission power control (power control command) WCDMA Wideband code-division multiple access WOS WCDMA operating

Ericsson Review No. 4, 2000 209

ulator varied between 50 and 80 km/h. Theinner-loop and outer-loop power controlfunctions were activated for the uplink aswell as the downlink. Receiver antenna di-versity was used on the base transceiver sta-tion side. The discontinuous transmission(DTX) function was deactivated for the up-link and downlink.

The uplink Eb/Io value and the number ofmultipaths were measured in the base trans-ceiver station; the uplink transmission out-put power was measured in the mobile sta-tion simulator; and the target Eb/Io valuewas measured in the radio network con-troller. All data was measured once a sec-ond.

The parameters were set to immediatelyincrease the target Eb/Io value if a single badframe was detected. There was only a one-frame delay before the system restarted mea-suring the next frame error rate (FER) value.The report cycle was reduced to 67 frames,which means that the target Eb/Io value de-creased if there were no checksum errors (re-vealed by a cyclic redundancy check, CRC)during 67 frames.

Measured Eb/Io [dB]

4

5

6

7

8

3.5

4.5

5.5

6.5

7.5

8.5

3 3.5 4 4.5 5 5.5 6 6.5 7 7.5 8Target Eb/Io [dB]

Figure 7Eb/Io versus target Eb/Io in base transceiv-er station.

3

3.5

4

4.5

5

5.5

6

6.5

7

7.5

8

0 1 2 3 4 5 6 7 8Number of rake fingers

Uplink target Eb/Io [dB]

Figure 8Relationship of number of rake fingers to target uplink Eb/Io.

Page 7: Experiences of the live WCDMA network in Stockholm, · PDF fileTPC Transmission power control (power control command) WCDMA Wideband code-division multiple access WOS WCDMA operating

The scale on the left in Figure 9 shows themeasured and target Eb/Io values (dB) andthe number of multipaths. The scale on theright shows the transmission power of themobile station (dBm). The measured Eb/Iovalue usually followed the target Eb/Io value,which means that the inner-loop power con-trol functioned well. As expected, the Eb/Iovalue ranged from 4 to 8 dB, and the varia-tion was not significant. In the middle ofthe graph, the Eb/Io value was very low, sincethe mobile station simulator was in the lineof sight of the base transceiver station.Sometimes there were high peaks in thetransmission power, but there was no highpeak in the target Eb/Io value at the sametime, which further indicates that the inner-loop power control worked well in the testenvironment. There was no major variationin the target Eb/Io value. In this test, the dy-namic range of the outer-loop power controlwas about 5 dB.

Quality control during discontinuoustransmission (lab test) Discontinuous transmission was used inWCDMA systems to increase system ca-pacity. When there is no data to transmit,the user data being transmitted over the airinterface was suspended, thus decreasing in-terference in the cell or increasing the ca-

pacity of the air interface. For example, dur-ing voice service, the typical speaker is silent60% of the time; thus, with discontinuoustransmission, we can increase voice capaci-ty. Even though the transmission of userdata was suspended, some control informa-tion was always transmitted, such as pilotbits and power control commands (TPC).However, if the discontinuous transmissionfunction was not properly systemized, itcould cause problems for quality control inthe radio network controller.

In a laboratory test, the quality controlfunction reduced the target Eb/Io value by 2to 3 dB when discontinuous transmissionwas active during voice service, thereby re-ducing the output power of the mobile sta-tion. This effect might generate transientbehavior when speech is resumed. Similar-ly, it might be difficult to establish newradio links for soft handover when the tar-get Eb/Io value is reduced. The quality of thesignal to the new base station was inade-quate when the path loss exceeded that tothe current serving cell. The uplink channelto the base transceiver station was then com-manded to produce barely acceptable syn-chronization. Consequently, a handovercandidate (BTS) with higher path loss haddifficulty in synchronizing with the mobilestation, which increased the chances of adropped call.

Outer-loop power control (or quality con-trol) during soft handover might also beproblematical because the radio connectionwith the highest path-loss could take overthe power control. If soft handover has beenestablished, the outer-loop power controlgenerates a target Eb/Io value that is suffi-cient to obtain the low Eb/Io value in theweakest leg. The power levels transmittedduring soft handover might therefore in-crease to undesirable levels in proportion tothe difference in path loss between the legs.This is because the outer-loop power con-trol in the radio network controller used thefollowing quality information:• measured frame error rate on user data;

and• synchronization status on the radio link.During discontinuous transmission, theframe error rate could only be estimatedfrom a few frames carrying backgroundnoise data (comfort noise). Thus, the syn-chronization status information on theweakest channel (highest path-loss) was in-stead used for outer-loop power control.When only a few speech frames were sent,the outer-loop power control reduced the

210 Ericsson Review No. 4, 2000

0 200 400 600 800 1000Time [s]

0

1

2

3

4

5

6

7

8

9

–60

–50

–40

–30

–20

–10

0

dB, Multipaths dBmEb/Io Target Eb/Io Number of multipaths MS Tx power

Figure 9Eb/Io and target Eb/Io, mobile station transmission power, and number of multipaths.

Page 8: Experiences of the live WCDMA network in Stockholm, · PDF fileTPC Transmission power control (power control command) WCDMA Wideband code-division multiple access WOS WCDMA operating

Ericsson Review No. 4, 2000 211

target Eb/Io value until synchronizationproblems occurred.

To correct this, we will either increase thenumber of comfort noise frames during dis-continuous transmission, or use anotherreading for outer-loop power control (in-stead of checksum errors on the user data)—for example, during the silent period, we canread the pilot bit error rate (BER).

Performance of fast power control withextensive time dispersion (lab test) Power control can be unstable when the rakereceiver does not resolve all the multiplepaths. By attempting to increase the powerto compensate for the decreased Eb/Io value,the fast power control also increased thepower of the interfering reflection and esca-lated transmission power. For example,when a reflection was located outside thesearcher window of the rake receiver, therake receiver was not able to take advantageof this delayed multipath component. Thepower control tried to compensate for thelower Eb/Io value caused by undetected re-flection, by increasing the transmissionpower of the mobile station simulator (inthe uplink). The increase in transmissionpower increased the power of the reflectionas well as interference. The purpose of thetest, which was performed in the lab system,was to investigate the effect of the powercontrol’s divergent behavior.

We used a DPCH32 channel with aspreading factor of 32. The fast power con-trol loop was activated with the target Eb/Iovalue arbitrarily set at 15 dB. The maximumoutput power of the mobile station simula-tor was set at 20 dBm, or 100 mW. Themeasurements started without a reflection.The late reflection was added after about 30seconds, then deleted after an additional 30seconds. The first test used a reflectionpower that was 3 dB stronger than the di-rect path. The reflection power was 5 dBstronger in the second test.

In Figure 11, we see that the fast power con-trol had trouble regaining the initial Eb/Iovalue. The reflection power was increased by2 dB (compared to Figure 10), which causedthe transmission power to increase by about13 dB. Increased transmission power increas-es the bit energy (Eb) and the reflection ener-gy (Iref):

Thus, in theory, the possibility of re-establishing Eb/Io by increasing the trans-mission power of the mobile station depends

Eb =Io

EbIref +N

Tx power [dBm], Eb/Io [dB]

Time [s]

0

2

4

6

8

10

12

14

16

18

20

0 10 20 30 40 50 60 70 80 90 100

Eb/Io [dB]

Tx power [dBm]

Figure 10For a DPCH32 channel: Eb/Io in the base transceiver station, average transmission powerfrom the mobile station simulator—with and without a reflection that is 3 dB stronger thanthe direct path.

Tx power [dBm], Eb/Io [dB]

Time [s]

0

2

4

6

8

10

12

14

16

18

20

0 10 20 30 40 50 60 70 80 90 100

Eb/Io [dB]

Tx power [dBm]

Figure 11For a DPCH32 channel: Eb/Io in the base transceiver station, average transmission powerfrom the mobile station simulator—with and without a reflection that is 5 dB stronger thanthe direct path.

Page 9: Experiences of the live WCDMA network in Stockholm, · PDF fileTPC Transmission power control (power control command) WCDMA Wideband code-division multiple access WOS WCDMA operating

212 Ericsson Review No. 4, 2000

3 km/h

Single Rayleigh Average Minimum Maximum Std. dev. Median[dBm] [dBm] [dBm] [dBm] [dBm]

BTS1 –5.5 –16.5 19.5 6 –6.5BTS2 –5.1 –16 20 6 –6.5BTS1&2 –8.5 –18 14 3.6 –9Gain [dB] 3.0 to 3.4

Single RicianBTS1 9.3 1 20 3.9 8BTS2 9.5 0.5 20 4.2 8.5BTS1&2 6.9 0.5 18.5 2.7 6.5Gain [dB] 2.4 to 2.6

High antennaBTS1 9.9 0 20 4.3 9BTS2 9.7 –1.5 20 4.2 9BTS1&2 7.2 –1 18.5 3.2 7Gain [dB] 2.5 to 2.7

100 km/h

Single Rayleigh Average Minimum Maximum Std. dev. Median[dBm] [dBm] [dBm] [dBm] [dBm]

BTS1 –5.8 –12 3 2.6 –6BTS2 –5.5 –12.5 6.5 2.6 –5.5BTS1&2 –8.5 –14 –1.5 1.9 –8.5Gain [dB] 2.7 to 3.0

Single RicianBTS1 15.1 9.5 19.5 1.7 15BTS2 14.8 17.5 19.5 1.8 15BTS1&2 12.8 7 17.5 1.7 13Gain [dB] 2.0 to 2.2

High antennaBTS1 10.8 4 18 2.5 11BTS2 10.2 4 18.5 2.5 10BTS1&2 7.7 2.5 14.5 1.9 7.5Gain [dB] 2.5 to 3.1

TABLE 6, SOFT-HANDOVER GAIN TEST RESULTS

Page 10: Experiences of the live WCDMA network in Stockholm, · PDF fileTPC Transmission power control (power control command) WCDMA Wideband code-division multiple access WOS WCDMA operating

Ericsson Review No. 4, 2000 213

on the ratio between the reflection energy(Iref ) and the noise power density (N). If theratio is less than one, then the fast powercontrol will be able to re-establish the tar-get Eb/Io value, otherwise, it will fail.

Although an interference level of 5 dBstronger than the direct path seemed to behigh at higher data rates, these levels werereduced by the lower spreading factors.Given a spreading factor of 4 instead of 32,interference limits were 9 dB lower. This ef-fect (Figure 11) is evident when reflectionpower was 4 dB less than the direct path(Figure 8).

Other mobile stations in the cell experi-enced increased interference and tried tocompensate for it by increasing their trans-mission power. If this effect is not compen-sated for when system algorithms andsearcher window sizes are designed, then asingle user could increase the interferencelevel in the entire cell.

Soft handover testsIn soft handover, the mobile station simu-lator was connected to more than one basetransceiver station at the same time. The sig-nal structure of WCDMA is well suited forthe implementation of soft handover. In theuplink, two or more base transceiver stationscould receive the same signal. In the down-link, the mobile station simulator could co-herently combine the signals from differentbase transceiver stations, since the signalswere considered to be additional multipathcomponents. This provided the additionalbenefit of macro diversity. However, in thedownlink, soft handover created more in-terference in the system, because the newbase transceiver station transmitted anoth-er signal to the mobile station simulator.

Soft-handover gain (lab test) Soft-handover gain was tested in the lab sys-tem. To include the effect of power control,soft handover gain was defined as the re-duction in the transmission power of themobile station simulator when soft han-dover is in progress. The same attenuationwas used for two uncorrelated channels (basetransceiver station no. 1, BTS1, and basetransceiver station no. 2, BTS2). The aver-age transmission power from the mobile sta-tion simulator was measured for • only BTS1 active;• BTS1 and BTS2 (BTS1&2) active (soft

handover); and • only BTS2 active (Table 6).

The greater the fading, the greater the soft-handover gain that could be obtained. Thisrelationship was found by comparing thechannel that was characterized by the mostsevere fading (single Rayleigh) to that char-acterized by the least severe fading (Ricianchannel).

Power control commands (TPC) weretransmitted every 0.625 ms, and the powerwas regulated in steps of 1 dB. The powercontrol was not fully able to cope with thefast fading channel.

In high-speed channels (faster variations),the effect was even more severe, resulting inlower soft-handover gain than at lower ve-locities. The effect is evident if we comparethe 100 km/h environment with the 3 km/h(see standard deviation in Table 6). The soft-handover gain was less when the transmis-sion power could not keep pace with the ac-tual fading channel. Slow power control alsogave rise to an increase in average transmis-sion power, in order to compensate for fad-ing tops that were missed. An increase inthe average transmission power increasedthe level of interference in the system.

For the high antenna channel (a GSMchannel model with six Rayleigh fadingpaths) the rake receiver used diversity in theRayleigh reflections, thus reducing the fad-ing detected by the receiver, which result-ed in a lower gain (compared with only oneRayleigh fading path).

Handover offset (field test) The handover offset parameter could be usedin the handover evaluation algorithm, tomove the cell border of a sector. Handoveroffset was employed to avoid hot spots at cellborders. This was evaluated in the field sys-tem.

Figure 12 shows the mean time duringwhich a handover branch was in differentsectors as a function of the handover set-tings. BTS1, BTS2 sector 1, and BTS2 sec-tor 2 refer to the three sectors designated inthe test (one omnidirection base transceiverstation site and one multisector base trans-ceiver station site). Handover branch timealso constitutes a measure of how well thecell borders could be controlled by the hand-over offset parameter.

The mean handover branch time of BTS2sector 1 was proportional to the handoveroffset parameter, while that of BTS2 sector2 had a much smaller dynamic range. Thisis because the antenna of BTS2 sector 2 cov-ered the test route, while that of BTS2 sec-tor 1 pointed in a different direction. The

Page 11: Experiences of the live WCDMA network in Stockholm, · PDF fileTPC Transmission power control (power control command) WCDMA Wideband code-division multiple access WOS WCDMA operating

signal from BTS2 sector 2 was thus verystrong in the mobile station (compared tothat of BTS2 sector 1), indicating that thevalue range of the parameter was too small.However, since our purpose was to controlthe cell border, the signals at the cell bor-der should have been relatively weak, re-sembling those from BTS2 sector 1. This,finally, indicates that the handover offset pa-rameter had a sufficiently wide value range.

Power drift during soft handover (lab test) During laboratory tests, it was found thatthe transmission power of base stationsdrifted unintentionally, causing the mobilestation to increase its transmission powermore than wanted. This phenomenon be-came even more significant when the dif-ference in path loss between two base sta-tions increased during soft handover. If thedifference was significant, there was somevariation in transmission power of the mo-bile station during soft handover.

A variable attenuator was used to vary the

path loss between two base transceiver sta-tions in the uplink and downlink connec-tions between the base transceiver stationsand a mobile station simulator. During softhandover, variable attenuators were used tochange the path loss by 0.1 to 4 dB.

A large dynamic range of 80 dB was usedin the uplink. The downlink range was 20dB. The maximum transmission power inthe mobile station simulator was 29 dBm;in the base transceiver station, 35 dBm. Theminimum transmission power value in themobile station simulator was –50 dBm; inthe base transceiver station, 15 dBm. Inner-loop and outer-loop power control were ac-tivated in the uplink and downlink, but thediscontinuous transmission function wasdeactivated.

As shown in Figure 13, the transmissionpower of BTS2 was frequently much high-er than the transmission power of BTS1. Attimes, the difference in transmission powerbetween BTS1 and BTS2 was greater thanthe difference in path loss (about 3.5 dB) be-

214 Ericsson Review No. 4, 2000

Mean handover branch time [%]

Handover offset for BTS2 sector 1 and sector 2

0

20

40

60

80

100

120

–5,+5 –4,+4 –3,+3 –2,+2 –1,+1 –0,+0 +1,–1 +2,–2 +3,–3 +5,–5+4,–4

BTS2 sector 2 BTS2 sector 1

BTS1

Figure 12Mean handover branch time.

Page 12: Experiences of the live WCDMA network in Stockholm, · PDF fileTPC Transmission power control (power control command) WCDMA Wideband code-division multiple access WOS WCDMA operating

Ericsson Review No. 4, 2000 215

tween the base stations. This means that thetransmission power of the mobile stationwas regulated by BTS2, whose uplink wasinferior and which sent power control com-mands that caused the mobile station to in-crease its transmission power. This hightransmission power could produce an un-necessarily high Eb/Io value in the uplink.

The transmission power of BTS2 was toohigh due to erroneous power control com-mands in the uplink. When the path lossbetween BTS1 and BTS2 became substan-tial, the uplink condition of BTS2 was quitebad, causing many commands to be inter-preted incorrectly. Consequently, an algo-rithm that reduces this effect has been in-troduced into system design.

Conclusion The tests described in this article are a sam-ple of the tests jointly conducted by Ericsson and Telia in the WCDMA evalua-tion system in Stockholm. Ericsson has con-

ducted similar tests in several differentcountries together with other operators. Theparticipating companies have gained veryvaluable practical experience of theWCDMA technology.

Several important system characteristicswere evaluated in real radio environmentsand have given rise to optimized algorithmsin the design of the commercial WCDMAproducts currently under development atEricsson. The system performance that wasobserved could also be used to create guide-lines for setting parameters in a real system.The effect should be visible in terms of bet-ter capacity, coverage, performance and re-liability for the real commercial systems thatwill be deployed in the near future.

We have shown that WCDMA technol-ogy lives up to the industry’s expectations.Thanks to the tests made using theWCDMA evaluation system, we have beenable to identify and correct unforeseenproblems, and in so doing, improved on thesystem design of commercial products.

17

19

21

23

25

27

29

0 2 4 6 8 10 12 14

Tx power [dBm]

BTS1 Tx power

BTS2 Tx power

Time [s]

Figure 13Illustration of power drift during soft handover.

The authors wish to thank the staff within theWCDMA evaluation project at Telia, Swiss-com, Centro Studie e Laboratori Telecomuni-cazioni (CSELT) and KPN Research for theircontributions.

ACKNOWLEDGMENT