|
Spectrum Management: |
[an error occurred while processing this directive]
Contents
1. Software installation on Sun 3 workstations.
The software comes on a magnetic tape cartridge (quarter‑inch QUIC-type) and includes the object code for RFDESIGNER, its associated topographic and geographic databases, a runtime version of GKS graphics libraries and associated files. It also contains files that can be useful in setting up the UNIX and SunView environment. There may be issues described in this section with which you are not familiar, in which case you are advised to initially contact your own Unix systems administrator.
It is recommended that only a system administrator use the higher access rights of the UNIX operating system on the workstation and that he installs the software. It is necessary to be logged into the system at this higher level in order to install the software.
This is old software and it will now only work on Sun Microsystems Sparc hardware with the sunc or sunm architecture running SunOS 4.0 up to Solaris 2.2. This software will not run on other Unix hardware or Intel‑based systems running Solaris, Linux, MacOS or Microsoft Windows. Support for the obsolete SunView windowing system, which RFDESIGNER uses was discontinued in later versions of Solaris, so this software will not run in versions later than Solaris 2.2. X-Windows emulators, or terminals cannot be used to run RFDESIGNER, as SunView predates X-Windows and is rather different in implementation. The chances of finding hardware still running pre-SunOS 4.0 operating systems are now thankfully remote, but RFDESIGNER wont run on them either. RFDESIGNER will also only run on machines that have the operating system kernel configured for shared memory. Details of how to check this are given in Appendix C. Refer to Suns own documentation for the general setting up of the workstation itself for items such as adding users, setting up passwords, and creating new user directories. Again, this is probably best done by a Unix system administrator.
The software will only run on the machine whose Hardware ID and MAC address details you originally gave to the JRC Secretariat. None of the data or code on the tapes should be passed to other parties and because of the hardware lock its highly unlikely they would be able to run the software anyway. This lock can also cause problems. Often, early Sun Hardware if stored for some time, loses its Hardware ID and MAC address as the battery on the NVRAM chip where this information is stored loses its charge. So long as you still have that information, then replacement of the dead NVRAM chip is the cheap and easy task detailed in Appendix E.
It is a straightforward task to install the software and it is described in detail here to avoid ambiguity. Before reading the tape it is necessary to create a number of working directories. The tape is then read and checks can be made that the new files exist.
To load the software,
1. Ensure that sufficient free space exists in the partition on which /usr is mounted. 7 MB is needed to install the software. Use the UNIX df command to establish the available disk space.
df
2. Check that /usr/filters exists. If the /usr/filters, directory doesnt exist then, as superuser create a new directory /usr/filtersusing mkdir:
mkdir /usr/filters
3. Create a new directory /usr/RFDESIGNER, using mkdir:
mkdir /usr/RFDESIGNER
4 Change directory to /usr/RFDESIGNER, using cd:
cd /usr/RFDESIGNER
5. Insert the tape into the tape drive and wait for it to stabilise.
6. Extract the application from the tar archive on the tape by issuing the command:
tar -xvfb /dev/rst0 126
The names of the filesbeing extracted from the tarball will appear on the screen as they are being written to the hard drive.
7. Files will be read into /usr/RFDESIGNER, /usr/filters, and /usr/liblgks. The contents of these directories are listed in Appendix A.
8. For each user, create a directory e.g. pmr, or jrcband or RFDESIGNER within their home directory. RFDESIGNER should then be run from this directory and the results will reside there.
9. If the Canon Laser Beam Printer LBP8II has not previously been set up, i.e. for use with RFPLANNER, then reference should be made to Appendix D for information on setting up. If you have a different printer, then the information in Annex D will be a useful guide, but you may need to refer to your printers user manuals.
Note: The software will be in a separate read‑only directory while the results of predictions will be in the directories of individual users and can be altered or deleted at will.
2. The background to RFDESIGNER.
The JRC has had a long involvement with computers for the prediction of mobile radio propagation, with the original concept going back to 1968. A key decision was taken early on in that original work to produce a topographical database of the UK. This was based on 25,000:1 scale maps and was done in such a way as encode key features such as maximum and minimum heights of peaks and valleys, so that a model of sufficient accuracy for vhf mobile radio could be built with a spacing interval of 0.5 km. The increased accuracy of profile reconstruction with a database of finer resolution was considered not to significantly improve prediction accuracy and would increase storage requirements and lengthen computer run times. The argument is still considered valid and the original topographic database is at the core of the RFDesigner software. The use of this topographic database allows a prediction of the mean value of the signal strength to be made for areas of 0.25 km. Since the topographic database does not contain details of embankments, cuttings, bridges, buildings, etc, the effects of local shadowing and fading have to be borne in mind by the system designer. The fine detail variations in propagation are often described mathematically by a Rayleigh distribution, and it has to be emphasised that this prediction model, like its predecessors, does not include this short period fading effect. In the particular case of the JRC band the specified 19 dB protection ratio includes a factor to take account of this local variation and allow mean values to be used in planning.
The propagation algorithms used in RFDESIGNER have been modified to provide improved modelling of losses due to ground reflection, and have been validated against some 1,800 km of measured route, using five base station sites and three frequencies.
RFDESIGNER produces predictions for circular areas, which the user can specify to suit particular requirements. This approach is well suited to the regular re-use planning of JRC band, and the software has a considerable degree of assistance for planning in that band. It is possible to overlap these circular prediction areas to give overall predictions for extensive areas. There are limits placed on the distance over which predictions can be made and these are principally determined by the distances over which the algorithms have been validated (~150 km).
Geographic information is included in the system and the extent to which this is incorporated into the display is a matter of user choice. Its purpose is to give the signal level predictions a relationship with geography that was previously only achieved by using transparent overlays of the prediction with conventional maps.
3. Comparison with RFPLANNER.
For those who have been using RFPLANNER, it is worth emphasising the differences with RFDESIGNER. The propagation model has been revised so that the effects of ground reflection losses and frequency are now more accurately modelled. The signal level predicted by the two models will vary between sites and their terrain, but the mean error and deviations are a significant improvement over RFPLANNER.
The revision of the algorithm has been coupled with revision of the control and scanning processes resulting in the time to compute predictions being halved. A greater reduction in computation time would have been achieved had the propagation algorithm not been made significantly more complex.
The data input process is now more fully integrated and there is a choice of resolution of the predictions which gives the opportunity to carry out faster preliminary predictions at increased spacing intervals on the ground.
The detailed functionality has increased significantly and it is expected that the improved user interface will make that functionality as easy to use as, if not easier than, RFPLANNER.
The new system supports investigation of the results of using more than one site to serve an area by allowing the combination of predictions.
The structure of the results files is different between the two systems, so that results computed by RFPLANNER cannot be displayed directly on RFDESIGNER. However, since RFDESIGNER is considered more accurate, it should be used for all future work. The storage requirements for a given area are halved for RFDESIGNER.
4. Using RFDESIGNER
4.1. How to start the program
It is assumed that the installation instructions in Section 1 have been followed and a directory has been set up for the storage of the results of the computations of predictions. The program will store its results in the directory from which the program is run. If your .cshrcfile has been amended to contain the alias rfd, then the program can be started by typing rfd, otherwise it will require the full command:
/usr/RFDESIGNER/RFDESIGNER
AS RFDESIGNER uses a major part of the machines resources and because of possible interactions between the windows of this program (see Appendix D) and other programs, it is recommended that RFDESIGNER is run alone.
4.2. Initialisation
When the program is first run from a new user directory, it will start up with an outline map display of the whole country together with 100 km grid lines and an invitation to select the area of interest as below.
Figure 1 Start-up screen on initial use
The upper limit on the area to be defined as the default area and displayed in the main window is 400 x 400 km and the lower is 100 x 100 km. The area is in whole 100 km squares, and this allows approximate positioning of the pointer when selecting the area, since pointing anywhere within a 100 km square will result in that square being at a corner of the area selected. It is envisaged that the area will be selected to cover the area normally of interest to the users organisation. To select the area, use the mouse to move the pointer to point to a corner of the area which you want and click the left mouse button.
|
Figure 2 |
Figure 3 |
Then move the pointer to the opposite corner of the area you require. You will see a box being drawn as you move the pointer, and clicking of the left mouse button will confirm the area of the map you have chosen. The map window display will change to show the 100 km squares corresponding to your selection and this area will become the start up or default area for RFDESIGNER. It is possible to change this default area at any time by the use of the modifyview area button in the control panel, which produces the screen shown in Figure 1.
Figure 4 - The screen after initial selection of area as in Figures 2 & 3
Once the selection of the start-up area has been made then the Control Panel will appear at the top right of the screen, giving access to the various functions of RFDESIGNER, and will be the starting point for subsequent use.
Figure 5 Control Panel
4.3. Computing predictions.
The control panel gives access to the major ways in which RFDESIGNER can be used. Messages relating to the overall status of the system appear in a status line at the top of the screen to the left of the control panel.
The selection of the perform predictions button (by means of the pointer and clicking the left mouse button) will display a data entry panel below the control panel, as shown below in Figure 6.
|
|
Figure 6 New Prediction panel
|
To enter data into this New Prediction panel it is necessary for the mouse pointer to be hovering overthe panel, but not necessarily pointing to any particular line. This is the default window focus policy for Unix windowing systems and varies from the Microsoft Windows convention. The caret (triangular cursor) will show where any keyed input will go (i.e. site name in Figure 6above). Items are entered into the system by hitting return. Entries in individual fields can be altered prior to keying return at the completion of that item. If the data that has been entered is subsequently judged to be incorrect then the abortbutton should be selected and the complete data re-entered. Moving the caret about with the mouse or cursor keys does not enable the user to edit fields already entered. The data required should in most cases be self-explanatory, and the following comments are aimed at clarification. If at any time the entered data is outside acceptable limits, an explanatory message will appear at the top of the new prediction window, above.
The site name can have a maximum of 20 alphabetic characters. This name is used as a directory name in the file system storing the results. Section 4.5 describes the structure of this file system.
After the National grid reference has been input in the form ab123456, the system will return the ground elevation at that point in metres as interpolated by RFDESIGNER. Hit return if this is acceptable. If the user has more accurate knowledge of the local terrain then the value suggested by the system should be deleted and the known value entered. Errors in the height used for base ground height can lead to substantial errors in the predictions.
The name given for this set of predictions should be chosen to be as meaningful as possible within the constraint of eight alphanumeric characters, bearing in mind that you will, no doubt, wish to investigate other designs using this base station.
The next field requires the height of the antenna above the base of the base station mast in metres. The type of antenna can be chosen from a repertoire of types held in the system and are chosen by clicking on the words displayed in antenna type. The system defaults to an omnidirectional antenna unless antenna typeis specifically selected. If other than an omnidirectional antenna is chosen then it is appropriate to give the bearing (direction of maximum radiation) of the antenna. This is given in degrees due east of grid north.
The power is the effective radiated power after feeder losses and any antenna gain has been allowed for. The minimum and maximum limits are 0.01 to 100 W.
The frequency can range from 30 to 999 MHz.
A default mobile antenna height of 1.5 metres is supplied by the system, and is the value that is recommended for general use although this value can be deleted and another value entered if required.
The purpose of giving a choice of sampling interval is to allow the user to perform faster predictions when a rapid comparison between a number of sites is sought. By doubling or quadrupling the sampling distance the time required to compute the predictions reduces by a factor of two and four respectively. The sampling distance also determines the maximum radius for the area of predictions. These are shown in the table on Page 10. For detailed designs the 0.5 km sampling distance should be used.
When this information has been keyed in and the data has been accepted, a further choice is added to the new predictions window.
This is a choice between:
1. Cellular. This invokes the values appropriate to the JRC Band cellular plan whereby, on specifying the identity of the service cell (Base Cell ID) in response to a prompt which replaces the cellular / free format buttons, all grid references and cell radii are automatically entered and computation started. This is shown in the left hand path in Figure 7 below. The main map window is still active and it may be convenient to turn on the cell pattern overlay to ascertain or confirm the cell identity.
2. Free Format. Selecting this choice results in being prompted for the number of areas for which predictions are required. This is shown in the right hand path in Figure 7. The user can then define the position and size of up to ten areas.
Figure 7 The choice between JRC Band Cellular Areas and Free Format
The maximum size of the individual areas to be surveyed depends on the sampling interval, which has been chosen, as shown in the table below
|
Sampling Interval |
Maximum radius |
|
0.5 km |
30 km |
|
1.0 km |
60 km |
|
2.0 km |
90 km |
The maximum distance between the base station site and the centre of a remote area is 125 km.
When the details are completed, the final carriage return starts the computation of the predictions.
Figure 8 - Prediction Monitor - which appears after completion of input.
The prediction monitor window gives the grid reference of the base station, the grid reference of the area being calculated, and a slide bar display showing progress through the calculations for each area in turn. The three main activities of calculating losses, converting to rectangular co-ordinates and writing to disk are described on the lower line. The status area at the top of the screen also indicates what is currently being done by the system. If it is awaiting user input this area displays Ready.
When all the calculations have been completed, they are presented on the screen as displayed at that time. If the main map window is displaying an inappropriate area, then nothing will appear in that window. However the Overview window will show both the prediction areas and the area selected for display in the map window (the area in the map window is identified by having two diagonal lines across a square). Re prediction areas are bounded by circles. The results are presented initially as a coarse representation, that is, they are presented at one quarter of the resolution at which they were computed (a 0.5 km sampling distance being shown with symbols at 2 km spacing).
The display of a set of predictions following the selection of the cellularoption is shown in Figure 9. Using the free formatoption, the coverage over a wide area around the base site can be investigated. By choosing a sampling interval of 2 km the signal levels in an area of up to 90 km radius can be predicted, and a display for 90 km radius around the same site as in Figure 9 is shown in Figure 10.
The control and monitoring for the main window area at the top of the screen has new controls and information added to it, appropriate to the additional functions now possible.
Figure 9. Predictions for Hampton site in cell 16E selected via the cellular option.
Figure 10 Predictions for Hampton site but with 90 km radius selected via the Free Format option.
4.4. Interacting with the results
4.4.1. Alteration of the presentation
The functions available in the control and monitor area of the main map window, as visible in the upper part of Figure 9, allow geographic information to be added in stages so that the user can make his own decisions on what is a reasonable amount of detail to have on the screen. The cyclic arrow selection controls have two ways of being used, determined by whether the left or right mouse button is clicked. If a control is pointed to and the left button clicked then it will step that choice to add the next category to that already displayed or will toggle to the alternative. If the right mouse button is depressed then a menu will be displayed and by sliding the pointer down the menu items to a choice and then releasing the right mouse button activates that menu item.
The amount of information that needs to be added to a display will depend on the scale of the area displayed. When an area of several 100 km squares is displayed only the larger towns and motorways may be appropriate. When a small area is being inspected then only small towns may be needed.
The area to be displayed can be selected and altered through both the overview and the map windows using the same procedure. With the pointer positioned in either of these windows, its representation changes to a cross. (Figure 11A) Position this cross at one corner of the area that you wish to have displayed in the main window, and click on the left mouse button.
|
(A)
|
(B)
|
Figure 11 Selection of an area to be displayed in the main map window.
Then move the pointer, now an arrow forming a rectangle behind it, to the opposite corner of the required area and click again (Figure 11B). The map window will be erased and rewritten with the new area. While zooming in can be achieved using either window, the overview window is necessary for zooming out.
Figure 12 Display resulting from selection in Figure 11, and the representation of this area in the overview window.
The speed of changing the view is significantly affected by the amount of detail that is selected for display and by the size of the default area as shown in the overview window. It is therefore advisable to minimise the additional detail displayed when changing views.
4.4.2. Spot values of grid reference, ground height and signal level, and path profiles.
The middle mouse button gives the facility, when used in the main map window, to display the national grid reference, the ground height, and the received signal level at a mobile located at the tip of the pointer.
When this function is used in a particular area for the first time it will take longer than subsequent calls, due to the loading of the relevant information.
It is also possible to examine the path profile between the base station and the point selected by the middle mouse button, by turning the profiles control to on. With profiles turned on, the operation of the middle button will, in addition to the presentation of the grid reference and height of the selected point, generate the terrain profile between the base station and the selected point and display this in the lower half of the main window, as shown in Figure 13. This terrain profile can be generated for paths up to 150 km from the base station. Both the base and mobile antennas are displayed by a scaled thick vertical line. The loss to the specified point is also calculated and displayed irrespective of whether the point is in the area of previously displayed predictions or not.
Figure 13: Display of path profile between the base station and a selected point.
4.4.3. Histograms of signal level distributions.
Histograms showing the distribution of signal levels within the prediction areas can be selected either individually or for all areas. To select the histogram for an individual area, point with the mouse to the outer part of the area (within 3 mm of the bounding circle) and click the right mouse button. The histogram will appear in the Statistics window at the bottom right of the screen. When the results are initiallydisplayed, the statistics window shows the histogram relevant to the first area specified in the input list.
Figure 14 Distribution of signal levels within a selected prediction area.
The horizontal scale is of the received signal level at the mobile with the design value of eirp at the base, and is in dBW. The bar at the top of the display gives the percentage falling into each of the three bands, this being the percentage of the area within the circle that have signals falling into the three categories.
Figure 15 - The distributions for all areas as selected by the statistics button.
When the service and all neighbouring co-channel cells are displayed in the main map window, the distributions for all the areas can be viewed by clicking on the statistics button above the main window. The histograms are drawn over the areas to which they relate, as shown in Figure 14.
4.4.4. Modifying the categorisation levels.
The design limits for JRC Band are such that a serviceable signal is greater than ‑128 dBW and less than ‑147 dBW for acceptable interference.
These values are displayed as three bands:
(a) greater than ‑128 dBW,
(b) between ‑128 and 147 dBW,
(c) less than 147 dBW,
These are the initial settings of the system. By selection of modifythresholdsa panel is presented, as in figure 15, which allows the user to adjust both levels by means of a slide bar, selected and moved with the mouse with the left mouse button depressed.
Figure 16 Control Panel that appears after selecting modify thresholds
When the levels have been adjusted to the required values the donebutton is hit and the results are re-displayed with the symbols having their new representation. The new values are displayed in the main map control and monitoring area and the settings will remain with their new values until deliberately changed. When the program is restarted from UNIX the levels will revert to ‑128 and 147 dBW.
4.5. Modifications to predictions.
While providing the ability to modify proposed designs of radio system is a principal objective of this software, the variables involved in radio propagation result in two levels of modification. Re task of modification is usefully approached from a description of the underlying propagation prediction.
4.5.1. The underlying structure of transmission loss and level prediction. The level of received signal is determined by:-
- transmission loss
- radiated power and antenna characteristics.
RFDESIGNER considers these issues separately. It is the transmission loss calculations which are computationally intensive, so that changes to design parameters are logically divided into those which affect transmission loss and those which dont, the latter being computed much more rapidly.
From the point of view of dealing with modification of designs and the structuring of files of results, the parameters affecting transmission loss fall into two groups.
- The first group consists of the grid reference and the height of the ground at a base station site. The site name is used to uniquely identify a grid reference and height, thus fixing its position in three dimensions.
- The second group are base and mobile antenna height above local ground, and frequency.
Thus a site name has attached to it just one grid reference and height, and the site name is used as the directory name for the results of all calculations for that site name. The second group of transmission loss factors are treated as being a prediction belonging to that site, and are stored in the site name directory, as a file called by the prediction name.
Parameters that do not affect the transmission loss (the base antenna type and orientation and radiated power) do not require that the transmission loss should be re-calculated when they are modified. Variations of power and antenna details are thus treated as trials dependent on the prediction to which they relate, and the effective file structure is shown in Figure 17.
In practice the prediction and trial files are held at the same level under the site name directory, the linkage being achieved with auxiliary files.
The areas specified to be covered and the resolution of the prediction relates to the calculation of transmission losses and hence is linked to the prediction name.
Figure 17 The effective file structure for prediction and trial results.
4.5.2. The use of these facilities.
When changes in the parameters determining transmission loss are to be made thenthe starting point is the selection of perform predictions and a new predictions panel will appear on the screen. When a site name is entered for which there are already predictions, this will be recognised and the details for the fi