IRAT Ericsson 2G DT reference without 2G to 3G CS Handover for Cingular market deployment

.. 5

1 Introduction

This document provides DT (Data Transcript) template for Ericsson 2G nodes to introduce IRAT handover feature. 2G to 3G handover CS call is not planned to be implemented in the network. Example is given where possible.

2 Revision Information

A – First version

B – updated on 10-04-05. Changed CGI with LAI in MGOCI

C – updated on 10-04-05

3 2G BSC DT

3.1 Activate features in the BSC to support IRAT

Activate IRAT feature:

DBTSP:TAB=AXEPARS,NAME=SUPCOEXUMTS;

SYPAC:ACCESS=ENABLED,PSW=PSW2PAR;

DBTRI;

DBTSC:TAB=AXEPARS,SETNAME=CME20BSCF,NAME=SUPCOEXUMTS,VALUE=1;

DBTRE:COM;

SYPAC:ACCESS=DISABLED;

DBTSP:TAB=AXEPARS,NAME=SUPCOEXUMTS;

RAEPP:ID=COEXUMTS;

RAEPC:PROP=COEXUMTS-1;

RAEPP:ID=COEXUMTS;

Set the interval at which the load in the GSM cell is measured:

RAEPP:ID=COEXUMTSTINT;

RAEPC:PROP=COEXUMTSTINT-1000;

RAEPP:ID=COEXUMTSTINT;

3.2 Define data for 2G to 3G cell reselection

Define UTRAN measurement frequency information in all IRAT 2G cells:

RLUMP:CELL=gsmcell;

RLUMC:CELL=gsmcell,ADD,UMFI=fddarfcn-scramblingcode-NODIV,LISTTYPE=IDLE;

RLUMP:CELL= gsmcell;

Example:

RLUMP:CELL=C3D32A1;

RLUMC:CELL=C3D32A1,ADD,UMFI=487-128-NODIV,LISTTYPE=IDLE;

RLUMP:CELL=C3D32A1;

3.3 Add UTRAN data to System Information for 2G cells participating in IRAT HO

RLSBP:CELL=gsmcell;

RLSBC:CELL= gsmcell,ECSC=YES;

RLSBP:CELL= gsmcell;

RLSUP:CELL= gsmcell;

RLSUC:CELL=gsmcell,FDDMRR=1,QSC=15,QSCI=1,QSI=7,SPRIO=YES,FDDQMIN=6,FDDQOFF=0; ! with Cingular recommended values with NO 2G to 3G CS Handover!

RLSUP:CELL=gsmcell;

Example:

RLSBP:CELL=C3D32A1;

RLSBC:CELL=C3D32A1,ECSC=YES;

RLSBP:CELL=C3D32A1;

RLSUP:CELL=C3D32A1;

RLSUC:CELL=C3D32A1,FDDMRR=1,QSC=15,QSCI=1,QSI=7,SPRIO=YES,FDDQMIN=6,FDDQOFF=0;

RLSUP:CELL=C3D32A1;

4 2G MSC DT

4.1 Activate features in the MSC to support IRAT

DBTSP:TAB=AXEPARS,SETNAME=GSMMMSF,NAME=MSCNF946;

SYPAC:ACCESS=ENABLED,PSW=PSW2PAR;

DBTRI;

DBTSC:TAB=AXEPARS,SETNAME=GSMMMSF,NAME=MSCNF946,VALUE=1;

DBTRE:COM;

SYPAC:ACCESS=DISABLED; DBTSP:TAB=AXEPARS,SETNAME=GSMMMSF,NAME=MSCNF946;

BSC to be defined as GSM99:

MGBSP:BSC=bsc;

MGBSC:BSC=bsc,BSCDATA=PHASE2-1;

Example:

MGBSC:BSC=RDMNDC6,BSCDATA=PHASE2-1;

4.2 Define outer Cells LAC of the 3G network

MGOCI:CELL=anyname,LAI=mcc(3g)-mnc(3g)-lac(3g),MSC=3gmsc;

Example:

MGOCI:CELL=3GLAI,LAI=310-980-47,MSC=3GALMSC;

4.3 MSC Routing, Signalling data

Cingular is taking responsibility for defining the data

Beginning of Cingular task:

Define Data for intra-MSC handover to 3G MSC:

Incoming Handover route:

EXROI:R=routename,DETY=MHOC,FNC=2;

Neighboring 3GMSC:

MGNMI:MSC=3gmsc,MSCADDR=3gmsc_address,R= routename;

Neighboring 3GVLR:

MGCVI:VLR=3gmsc,VLRADDR=3gvlr_address,MAPV=MAP-3,LAI=locarea_in_3gmsc;

S7 Data towards 3GMSC & 3GVLR:

S7TZI;

S7TCI;

S7TSI:GTS=3gmsc_address,GTRC=x;!3GMSC!

S7TSI:GTS=3gvlr_address,GTRC=x; !3GVLR!

S7TAI;

S7NPI:SP=3gmsc_SPC;

S7NSI:SP=3gmsc_SPC,SSN=7;

S7NSI:SP=3gmsc_SPC,SSN=8;

Routing analysis on 3G handover number (network dependent info):

EXROI on intermsc_trunkroute

ANRSI to define routing case

ANBSI to define B-nr analysis

LATA connection to 3GMSC (network dependent info):

MGLXI

MGLXP

Example:

EXROI:R=7MHOC1,DETY=MHOC,FNC=2;

MGNMI:MSC=3GALMSC,MSCADDR=4-19078319870,R=7MHOC1;

MGNMP:MSC=ALL;

MGCVI:VLR=3GALMSC,VLRADDR=4-19078319870,MAPV=MAP-3,LAI=310-980-31; !MSC & VLR addresses are same in Ericsson CN!

MGCVP:VLR=ALL;

S7TZI;

S7TCI;

S7TSI:GTS=10-19078319870,GTRC=9;!3GMSC & 3GVLR!

S7TAI;

S7NPI:SP=229-99-70;

S7NSI:SP=229-99-70,SSN=7;

S7NSI:SP=229-99-70,SSN=8;

4.4 Define equivalent PLMN if PLMNs are different in 3G and 2G

MGELP:EPLMNL=ALL;

MGCPP: EPLMNL=ALL;

MGELI:EPLMNL=eplmnl,PLMN=plmn...;

MGCPI:EPLMNL=eplmnl,LAI=lai...; ! connect equivalent PLM to Loc Area!

End of Cingular task.

5 2G SGSN DT

Activate the nodeproperty for the usage of DNS for LAC-RAC lookup. Activate equivalent PLMN feature in the SGSN.

Define equivalent PLMN if PLMNs are different in 3G and 2G network.

Verify the new LAC-RACs with quiry to DNS.

The following commands are used to configure Ericsson SGSNs for IRAT:

gsh set_nodeproperty Gn_UseDnsMapRaToSgsnAddress –val true

gsh create_imsins ImsiNumberSeries -rs Home|Visitor -dn DomainName -np NumberingPlan [-na NatureOfAddress] [-rd NoOfDigitsToRemove] [-ad DigitsToAdd] [-misc1 String] [-misc2 String] [-misc3 String] [-phase3 CamelPhase3] [-qpmw QoSPolicyMap] [-qpmg QoSPolicyMap]

set_feature equivalent_plmns ok

gsh create_epl_plmn EquivalentPlmnList -plmn “Plmn”

gsh create_epl_imsins ImsiNumberSeries -eplp EquivalentPlmnListPointer

gsh create_epl_lai Lai [-epl0 EquivalentPlmnList0] [-epl1 EquivalentPlmnList1] [-epl2 EquivalentPlmnList2] [-epl3 EquivalentPlmnList3] [-epl4 EquivalentPlmnList4] [-epl5 EquivalentPlmnList5] [-def DefaultEquivalentPlmnList]

gsh create_csgsn SgsnName -ip SgsnAddress (Not needed, since DNS solutionis used for LAC-RAC lookup)

gsh create_cra Rai -sgsn SgsnName (Not needed, since DNS solutionis used for LAC-RAC lookup)

How to quiry and verify a new LAC-RAC from SGSN?

Login to SGSN, go to “/tmp/DPE_SC/ApplicationData/dnsApp/”, type ./test_resolve and enter the LAC-RAC.

e.g.

=== root@eqm01s14p2 ANCB ApplicationData/dnsApp # ./test_resolv

Enter hostname or 'quit' to exit: rac0001.lacCB28 mnc410.mcc310.gprs

Official hostname: rac0001.lacCB28.mnc410.mcc310.gprs

address: 66.102.182.65

Smith Chart™ for Excel™

New Edition Allows Entering Complex Impedance Values Rather Than S-Parameters

here to Download> (Note: IE8 sometimes has problems with the ZIP. Please use Chrome or Firefox, or, send me an e-mail)

A situation arose at work where we needed to be able to plot complex impedance points taken off the network analyzer display and plot them on a Smith Chart. A modification to the original Smith Chart for Excel permits doing so. If you download the new edition, you will need to enable the Analysis ToolPak (included with Excel as an Add-In) to perform the complex math.

The original Smith Chart for Excel that takes s-parameters as input is detailed below. ↓ ↓ ↓ ↓ ↓ ↓ ↓ ↓ ↓ ↓

This example Excel workbook demonstrates how easy it is to implement a Smith Chart using only a standard x-y scatter chart and coordinate conversions. The workbook shown below used data imported from a typical S-parameter file (in this case an RF2321 amplifier, from RF Micro Devices) and plotted on a chart that uses an image file that contains a Smith Chart. Version 2.0 adds equivalent denormalized impedance with equivalent resistance and capacitance/inductance values. Version 2.1 corrects a graphical equation, but does not affect the accuracy of the previous versions (thanks to Peter for alerting me).

Step-by-step instructions are presented below. Click here to download the example workbook. Click here to download just the background Smith Chart graphic. Finally, click here if you would like a fully-detailed Smith Chart, created in Visio, with impedance and admittance lines.
(Note: IE8 sometimes has problems with the ZIP. Please use Chrome or Firefox, or, send me an e-mail)

If you appreciate the effort it took to develop this workbook, please consider making a donation to RF Cafe by clicking here (at the bottom of the list).

Engineers use spreadsheets for a myriad of applications from calculating cascaded chains of components to PLL phase noise prediction, but I can never recall seeing S‑parameters plotted in a spreadsheet using a Smith Chart¹. If a Smith Chart is included in a spreadsheet, it is usually an image pasted in from some other application. This article describes an extremely simple method of implementing a Smith Chart using the built‑in graphing capability of any modern spreadsheet program (Excel is used in this example). All that is required is an accurate graphic of a Smith Chart for use as the chart background image, and a rectangular‑to‑cylindrical coordinate conversion.

Example Spreadsheet




RF2321 Datasheet Excerpt

Although the example given here is used to plot S‑parameters from a file, the possibilities are great for generating any sort of Smith Chart application such as for impedance matching.
A general-purpose amplifier (RF2321) manufactured by RF Micro Devices is used in this example, and its S-parameter file was downloaded from the RFMD website. A copy of the datasheet Smith Charts are given for results comparison. Here are step-by-step instructions for generating your first Smith Chart. Experienced Excel users might want to skip down to the image loading and calibration section.

• Open a new workbook in Excel.
  
• Click the "File/Open..." menu selection and locate the S‑parameter file to be plotted (in this case, “23212725.s2p”). Set the window to display "All Files (*.*)," since the S‑parameter file will most likely not end in an Excel extension.
  
• The Text Import Wizard will open. Select the "Delimited" option, then click "Next."
  
• Unclick the "Tab" checkbox and select "Space." Scroll down into the data area and verify that the data is separated by vertical lines at the appropriate points (lined up in columns), then click "Next."
  
• Click "Finish." You will now have all the data imported into a worksheet. Now would be a good time to save the workbook under a new name (be sure to save it as an Excel worksheet).
  
• The data column labels might need to be shifted to line up with the data (results will not be affected if left as is). The data cannot be plotted as imported and must be translated into equivalent circular coordinates (very simple).
  
• Click the "Insert/Worksheet" menu selections.
  
• Refer to the example spreadsheet as a suggested format for the plotting data.
  
• In the "Freq (MHz)" column, use the equation ="Freq"/1e6, where "Freq" is referenced from the S‑parameter import worksheet. This column is not plotted, but is provided as a reference for the S‑parameter.
  
• In the "S11x" column, use the equation ="|S11|"*cos("∕ S11"*PI()/180), where "|S11|" is the magnitude and "
• In the "S11y" column, use the equation ="|S11|"*sin("∕ S11"*PI()/180).
  
• In the "S22x" column, use the equation ="|S22|"*cos("∕ S22"*PI()/180).
  
• In the "S22y" column, use the equation ="|S22|"*sin("∕ S22"*PI()/180).
  
• That creates the first row of equations. Now, highlight all five cells and grab the "handle at the lower right corner of the highlighted area and drag it down by the number of rows of imported data (201 in this case). You can cut out whatever data you do not want to plot.
  
• Use the "Format/Cells..." menu selection to format the numbers to your preference.
  
• Highlight the entire group of S‑parameter data (201rows by 4 columns), then click the "Insert/Chart..." menu selection. Click the "XY (Scatter)" chart type and then select the "Scatter with data points connected by lines." picture in the lower left. Do not worry that the preview looks meaningless at this point. Click "Next."
  
• Select the "Series" tab. Highlight "Series 1" in the list, then place the cursor in the "Name" edit box and type in S11. The name in the list will change to "S11."
  
• Click "Series 2" in the list and then click the "Remove" button.
  
• Click "Series 3" in the list and rename it to S22. In the "X Values" edit box, change the "$B" to "$D" on both sides of the colon, then click "Next."
  
• Click the "Gridlines" tab and uncheck everything, then click the "Legend" tab and select the "Corner" option. Click "Next," then "Finish."
  
• Move the chart to a convenient place on the worksheet, and reshape it to as close to a square as possible. Not being a perfect square will not affect the accuracy of the plotted points, but will make a nasty looking Smith Chart.
  
• Click an open area of the graph (the "Plot Area") and use the "handles" to resize the graph to fill the graph window (it will not go all the way to the edge).
  
• LOADING THE SMITH CHART IMAGE
• Right-click in the Plot Area and select the "Format Plot Area..." menu selection, then click the "Fill Effects..." button. Next, click the "Picture" tab and click the "Select Picture..." button.
  
• Navigate to where your favorite Smith Chart image is located and select it. The one used in this example can be downloaded from the RF Cafe web site. If you are creating your own version, the best results can be had using a vector image creator (such as Visio) and then saving it in WMF or EMF format. Doing so preserves the sharpness of lines when resizing. It is also necessary to provide white space around the edge of the image to allow for the Excel plot area not being able to extend all the way to the edges. Click Insert. Click the "OK" buttons to close all the formatting windows.
  
• CALIBRATING THE SCALES
  
• Somewhere on the worksheet enter the numbers -1, 0, and 1 in separate cells. These will be used to set the scale to correspond with the outer circle.
  
• Right-click in the Plot Area and choose the "Source Data..." menu selection, then click the "Series" tab.
  
• Click the "Add" button and type "-1+j0" in the "Name" edit area. Place the cursor in the "X Values" edit area and select the cell with the "-1" in it. Place the cursor in the "Y Values" edit area and select the cell with the "0" in it. Note that any default values in the edited areas must be overwritten.
  
• Click the "Add" button and type "1+j0" in the "Name" edit area. Place the cursor in the "X Values" edit area and select the cell with the "1" in it. Place the cursor in the "Y Values" edit area and select the cell with the "0" in it.
  
• Click the "Add" button and type "0+j1" in the "Name" edit area. Place the cursor in the "X Values" edit area and select the cell with the "0" in it. Place the cursor in the "Y Values" edit area and select the cell with the "1" in it.
  
• Click the "Add" button and type "0-j1" in the "Name" edit area. Place the cursor in the "X Values" edit area and select the cell with the "-1" in it. Place the cursor in the "Y Values" edit area and select the cell with the "0" in it. Click "OK."
  
• Right-click the y-axis and select the "Format Axis..." menu selection, then click the "Scale" tab.
  
• Set the "Minimum" value to -1.02, the "Maximum" value to 1.02, and the "Major unit" and "Minor unit" values to 5. Click OK."
  
• Right-click the x-axis and select the "Format Axis..." menu selection, then click the "Scale" tab.
  
• Set the "Minimum" value to -1.02, the "Maximum" value to 1.02, and the "Major unit" and "Minor unit" values to 5. Click "OK."
  
• If the calibration marks do not line up with the unit circle of your Smith Chart, go back and adjust the scales until they do. After calibration, the marks and axis lines and labels can be removed to eliminate clutter.
  

That's all there is to it. As you can see, the results are identical to the published Smith Chart in the RFMD datasheet. Once you do the first one, the rest will be really easy. Of course, if you do not want to go to the trouble of carrying out the above procedure, you can simply go to the RF Cafe web site (http://www.rfcafe.com) and download the "Smith Chart for Excel" file free of charge. This exact example workbook is what you will be getting. - Enjoy!

REFERENCES

1. Smith Chart is a registered trademark of Analog Instruments Company, New Providence, NJ
   

2. “Field and Wave Electromagnetics,” by David K. Cheng, Addison Wesley, 198

ANTENNA PARAMETERS

ANTENNA PARAMETERS

GAIN

Since an antenna is passive, the only way to obtain gain in any direction is to increase the directivity by concentrating the radiation in the wanted direction. For a loss free antenna the directivity can be given with the same number as the gain if the latter is given with respect to an isotropic antenna. Hence, in this chapter the distinction between gain and directivity is not always strictly maintained. The directivity can be increased by reflectors or by stacking dipoles on the same vertical line. The latter method can be used because a number of coherent radiation sources interfere constructively (in directions where they radiate in phase) and destructively (in directions where they are in “anti phase” and more or less cancel each other out). Each doubling of the number of dipole elements (corresponding to a doubling in length) increases the gain in the main direction by 3 dB. Figure 1 shows some different antenna arrays. The gain is different in different directions. However, when the antenna gain is quoted it is usually given for the direction of maximum radiation.

Figure 1 Antenna Arrays

Since the concentration of radiation is inversely proportional to the solid angle of the beam, the gain can be estimated if the beamwidths are known:

G = 10 x log 31000/(V3 x H3)

G = Antenna gain relative isotropic antenna (dBi)

V3 = Vertical beamwidth relative -3 dB points (degree centigrades)

H3 = Horizontal beamwidth relative -3 dB points (degree centigrades)

BEAMWIDTH

Vertical Beamwidth

Since the concentration of radiation is proportional to L/l, the vertical beamwidth decreases as the gain increases. The vertical beamwidth can be estimated if the length of the antenna is known:

V3 = 15300/(F x l)

V3 = Vertical beamwidth relative – 3 dB points (degrees centigrades)

F= Frequency (MHz)

l= Antenna length (meter)

ANTENNA DOWN TILTING

The vertical beam of an antenna is normally directed towards the horizon, assuming the antenna is correctly mounted. Lowering the beam below the horizon is known as “down tilt” (Figure 2). Consequently, if the beam is directed above the horizon, “up tilt” is achieved. Below is a description of the methods used to achieve down tilt and a discussion on how down tilt can improve the performance of a system. Up tilt will not be discussed further.

Figure 2 Antenna Down Tilt (Basic Geometry)

ELECTRICAL TILT

Electrical down tilt requires an antenna with a number of vertically stacked dipoles. (Here, the word “dipole” represents other radiating elements as well.) The individual dipoles can be oriented vertically, which is the most common orientation in cellular systems. They can also be oriented horizontally or at a slant (±45°) position. If all dipoles are fed with the same phase, the main beam of the vertical pattern will be perpendicular to the mechanical axis of the antenna (towards the horizon). A phase difference between the dipoles will result in a beam that deviates from the horizontal. Different tilt angles are available, depending on the antenna manufacturer. Typical values are 2° and 6°. An advantage of using electrical tilt is that the antenna is always mounted in a vertical position irrespective of tilt. A disadvantage is that the antennas must be ordered with a certain tilt angle. (Antennas with adjustable electrical tilt are available on the market to avoid the disadvantage of fixed tilt values. The antennas have a limited gain and are expensive.)

MECHANICAL TILT

Mechanical tilt is achieved by changing the mechanical alignment of the antenna. All antenna manufacturers have adjustable brackets designed for this purpose. It is possible to combine the electrical and mechanical methods.

CELL PLANNING ASPECTS ON DOWN TILT

Down tilt can be used to overcome coverage and/or interference problems. To be able to discuss down tilt from a general point of view, some special applications must be excluded, i.e. antennas on extreme hill tops, the “Manhattan syndrome”, etc. In these cases, tilt can always be motivated. As a general rule, to reduce co-channel interference, three criteria must be fulfilled:

1. Short site-to-site distances (small cells)

2. High mounted antennas

3. High gain antennas (narrow vertical beam)

Figure 3 Schematic of a Regular Network With Site to Site Distance of 1 km α=1° β=2°

Let us start with a case based on medium values (Figure 3). Site-to-site distance: 1 km; antenna height: 25 m; and an antenna with 14° vertical beamwidth (approximately at the -3 dB point). As a starting point, let us reduce the signals from the interfering site (Alpha) towards the interfered site (Bravo) by 7 dB. The diagram in Figure 3-4 shows that a tilt of 10° is needed to achieve a reduction of 7 dB towards the horizon. However, to reduce the signal by 7 dB at the cell border, a tilt of 11° is needed (10 + 1°) since the angle ( a) towards the cell border is 1°. Note that the gain reduction at the cell border for no tilt is almost zero.

We started with site Alpha which is a potential interferer to site Bravo. As we down tilted Alpha by 11°, the interference situation in site Bravo is improved by 7 dB. But if the network is regular (in a reasonable sense) site Bravo is also an interfering site to site Cairo. Now we have to down tilt Bravo as well with the same values as Alpha and the result in a regular network is that almost all sites must be down tilted The next step is to see what happens in the own site area when the antenna is down tilted. The angle (b ) between the horizontal and a mobile on street level on the cell border is 2° (Figure 3). It is obvious that the mean vertical beam is pointing somewhere inside the cell border. 11° corresponds to a distance of 129 m. From the same antenna diagram, it can be seen that the signals at the cell border are reduced by 5 dB, found in the diagram at 11 – 2° = 9°. Note that the gain reduction at the cell border for no tilt is almost zero.

The net result regarding C/I increase is only 2 dB — at the expense of 5 dB coverage loss!

Figure 4 Typical gain reductions as a function of tilt angle for three different antennas (beamwidths are 7, 14, and 28 degrees)

Note that the figures are not drawn to scale (i.e. that the horizontal scale is different from the vertical scale). It is common to make figures in this way but it can be misleading. It is obvious that the calculations so far are based on a network in open terrain as no obstacles can be seen between the base station and the mobile. A more realistic case with respect to co-channel interference problems is in urban or suburban areas with buildings in-between (Figure 5).

Figure 5 This figure illustrates the fact that there is seldom line-of-sight between two antennas in an urban environment

It is unlikely that the radio signals follow the direct line between the base station antenna and the mobile, passing all the buildings in-between. It is more realistic to see the signals coming from (by reflection and diffraction) the roof tops down to the street. The angle to the cell border can then be calculated from the base station antenna height above roof tops (e.g. 5 m). Assuming a site-to-site distance of 1 km, it is an angle of 0.4° to the cell border. The conclusion is: Signals from the interfering site and the interfered site arrive at the cell border with a very small difference in the vertical angle – regardless of how much down tilt is applied. However, down tilting means that less radiation is transmitted across the roof tops and the coverage might decrease.

Returning to the three requirements in this section:

1. Short site-to-site distances (small cells)

2. High mounted antennas

3. High gain antennas (narrow vertical beam)

It can be seen that the first requirement (small cells) gives the possibility to achieve a difference in the two vertical angles towards the roof tops on the cell border and towards the roof tops on the interfered site. The second requirement helps to increase that difference. Finally, with a narrow vertical beam, a C/I increase by 2-3 dB is possible if 1° angle difference can be achieved (e.g. by mounting the antennas 20 m above the roof tops) and that not more than 5 dB coverage reduction is acceptable. For example, if a 7° antenna is tilted 5°, the gain reduction towards the roof tops for the interfered site is 3 dB (found at 5° – 2° = 3° in Figure 4); whereas for the interference, it is 5 dB (found at 5° – 1° = 4°) i.e. a 2 dB increase in C/I.

NULL FILL-IN

As previously mentioned, antenna gain is different in different directions (Figure 6). This means that areas at a certain distance (depending on the antenna height) from the antenna will be radiated by the first null rather than the main direction. Hence, the signal level will not decrease monotonically as the distance between the transmitting antenna and the receivers increases, but more as it is illustrated in Figure 7 and Figure 8. For parallel-fed collinear arrays, it is possible to reduce the gain reduction in the direction of the first null by simply adjusting the power fed to the different antenna elements slightly. This gives a small reduction in gain in the main direction but this is compensated for by much more predictable signal strengths in areas closer to the transmitting antenna.

Figure 6 Gain Reduction as a Function of Vertical Angle

Figure 7 High gain antenna at 25 m height

Figure 8 High gain antenna at 75 m height

DIVERSITY

There is a need for receiver diversity in cellular systems to improve the uplink. Space diversity is the conventional method used where the two RX antennas are separated by a certain distance. Based on experience from measurements and simulations (and because of installation advantages) polarization diversity is used in standard configurations. The signals from the two RX antennas are later combined in the base station. The result is an increase in signal strength of three to six dB. (The exact value depends on the similarity between the signals received from the two antennas where the two receiving antennas are separated by 90 degrees in the polarization plane.)

SPACE DIVERSITY

Figure 9 shows a traditional configuration with space diversity. The horizontal space needed for the antennas is dependent on the required diversity separation.

Figure 9 Antenna configuration with space diversity

POLARIZATION DIVERSITY

A dual-polarized antenna is an antenna device with two arrays within the same physical unit. The two arrays can be designed and oriented in different ways as long as the two polarization planes have equal performance with respect to gain and radiation patterns.

Figure 10 Dual polarized antennas

The two most common types are vertical/horizontal arrays and arrays in +/-45 degree slant orientation (Figure 10). The two arrays are connected to the respective RX branches in the BTS.

The two arrays can be used as combined TX/RX antennas (Figure 11) and then the number of antenna units is reduced compared with space diversity. The use of a duplex filter reduces the number of antenna units to only one per cell depending on configuration.

Figure 11 Antenna configuration with polarization diversity

The diversity gain obtained from polarization diversity is slightly less then the gain from space diversity. In the most critical environments (such as indoors and inside a car) the gain is, however, almost as good as if space diversity were used. A dual polarized antenna offers very low correlation between the two received signals, but the power reception of each branch is slightly better with space diversity. This implies a small benefit for space diversity in noise-limited environments. For most applications, the difference is negligible. In interference limited environments on the other hand, the low correlation obtained by polarization diversity is advantageous. Due to slightly different propagation characteristics for different kinds of polarization, the downlink from a +/-45 degree dual polarized antenna suffers from about 1.5 dB extra loss compared to a vertically polarized antenna. This loss only affects the downlink.

The isolation between the two polarization planes needs to be 30 dB. The size of the antenna must remain small, as the intention with polarization diversity is to reduce the outlook of the antenna installation.

INTERMODULATION (IM)

When two signals of a different frequency mix in a non-linear device, the result is InterModulation (IM). The non-linear devices can be, e.g. antennas, combiners, connectors, and duplex filters. IM can be a problem at any site that has two or more transmitters. The IM problems can be caused by a transmitter in the same system or by a transmitter in another system that is cosited or has a site in the neighborhood. Finding the intermodulation source can be time-consuming since the problem is often intermittent.

Second order products are given by the formula f1 +/- f2 . Both these frequencies are outside the receiver passband. In fact, all the even-order products will be well outside the receiver passband. Third order products are given by the formulae 2 x f1 – f2 and 2 x f2 – f1. These frequencies fall inside the band. All oddorder products can cause problems (Figure 12). However, higher order products (usually 7th order and higher) decrease rapidly in power and therefore do not cause any problems.

Figure 12 Intermodulation in 900 MHz cellular system

The allocated frequency band and the duplex distance are what determines if the IM will cause problems (Table 1 and Table 2). IM3 products are strong enough to degrade the receiver sensitivity even though there is no combining; just backwards coupling from one antenna to the other. IM5 is a problem if the frequencies are combined before entering the duplex filter.

Table 1 The maximum band (B) to avoid intermodulation. Based on worst case scenario, which is IM from the lowest and the highest frequencies in the allocated band

Table 2 Worst case relations for IM in the RX band. D= Duplex distance (MHz), B = allocated band (MHz)