1、 Rep. ITU-R BT.2137 1 REPORT ITU-R BT.2137 Coverage prediction methods and planning software for digital terrestrial television broadcasting (DTTB) networks (2008) CONTENTS Page Introduction 2 1 Prediction error statistics 3 1.1 United Kingdom (UK) results 3 1.2 Japanese results. 3 1.3 Comparison of
2、 measurements with field-strength predictions, Trondheim Area. 4 1.4 Australian results 4 2 Field-strength prediction method . 6 2.1 The method used in the United Kingdom (the UKPM) 6 2.2 The method used in Japan. 11 2.3 The method used in Canada 13 2.3.1 CRC-PREDICT A VHF and UHF propagation model
3、13 2.3.2 Calculation 14 2.3.3 Ground reflection 16 2.3.4 Tropospheric scatter 16 2.3.5 Location variability. 16 2.3.6 Time availability. 16 2.3.7 Summary. 17 2.4 The method used in Brazil 17 3 Prediction error statistics 17 3.1 United Kingdom results 17 3.2 Japanese results. 19 3.3 Comparison of mea
4、surements with field-strength predictions, Trondheim Area. 19 2 Rep. ITU-R BT.2137 Page 4 Planning software . 20 4.1 Introduction Database centred planning software . 20 4.2 Software developments. 22 4.2.1 Switzerland . 22 4.2.2 Japan . 25 4.2.3 Canada 26 4.2.4 Brazil. 27 4.2.5 LS Telcom 27 5 Additi
5、onal factors impacting coverage. 33 5.1 Introduction. 33 5.2 DVB-T practical reception problems 34 5.3 ATSC 8-VSB practical reception problems . 34 5.3.1 SNR requirement 34 5.3.2 Propagation loss and statistics 34 5.3.3 Receiver/antenna model for coverage planning . 35 6 Discussion of results and me
6、thodologies . 38 Introduction The implementation of DTTB services in parallel with existing analogue services in several countries has created the need to refine some of the traditional computer-based frequency planning techniques to enable a greater degree of accuracy in coverage prediction. Wherea
7、s analogue systems fail rather gracefully, the “cliff-edge” failure characteristics of digital systems can mean that in some situations “holes” in DTTB coverage will result from the various factors that affect signal coverage. These include, but may not be restricted to, propagation characteristics
8、of the bands used for DTTB transmissions, limits imposed on DTTB transmission power in order to protect the existing analogue services, terrain obstruction and man-made clutter. Clearly the identification of geographic areas where such holes might be expected is important for coverage planning as we
9、ll as for the receiver retail trade, where clear advice to potential viewers is essential. It is for these reasons that improved coverage prediction methods have been introduced in a number of countries with considerable success, and that it is considered important that the new methods being develop
10、ed are studied and documented by ITU with a view to achieving an appropriate degree of standardization worldwide. Rep. ITU-R BT.2137 3 This Report provides a brief outline of the results of comparisons between predicted and measured signal levels as reported by some administrations. These results sh
11、ow wide divergences between predicted and measured signal levels in terms of both mean error and standard deviation of errors. While these variations may have been acceptable in analogue television planning, the rapid failure of digital television signals means that a much closer match of prediction
12、s with measurements is required. An approach is discussed for predicting received field strength with particular discussion of profile extraction, radial prediction and the use of clutter data to take into account the effect of buildings and trees. Transmitter and population databases are also discu
13、ssed. It should be noted that in addition to the ongoing systems work described in this Report, Radiocommunication Working Party 3K is in the process of developing a text on a site-specific propagation model for terrestrial services from about 30 MHz to about 5 000 MHz. This deterministic model will
14、 include the effects of terrain features, ground covers and buildings. It will also include location and time variability, and multipath effects. As a first step towards the development of the above text, Working Party 3K is actively evaluating several existing site-specific propagation models. The
15、purpose of developing the improved prediction models is to produce consistent prediction results between related planning organizations while taking advantage of the availability of terrain and clutter data and improvements in computer power. To obtain this consistency the prediction model must spec
16、ify the full sequence of processing steps. Bearing in mind that most new DTTB services will be introduced in parallel with the existing analogue television services, using the existing antenna and down lead, a further point of considerable practical importance is that of providing an accurate model
17、of typical domestic receiver/antenna installations and the impact of losses in this area on the required received field strength. Some initial work on this problem is reported below with the suggestion that typically, the required “implementation margin” may be quite considerable. 1 Prediction error
18、 statistics The following notes provide summaries of work undertaken by some administrations in the comparison of measured and predicted signal levels. 1.1 United Kingdom (UK) results The mean error and standard deviation for the BBC model assuming 500 m profile sampling resolution and the UKPM mode
19、l assuming 500 m and 50 m resolution are presented in Table 3. The better performance of the UKPM model is clearly illustrated by these results. It is also apparent that most of the performance gain is achieved by the inclusion of the clutter loss prediction algorithm which is further improved with
20、the increased resolution. The corresponding excess loss graphs are presented in Figs. 12 and 13. The relatively small scattering of the points in the UKPM model are a clear indication for its superior performance. The validation of the UKPM model has been performed against the mean error, and the st
21、andard deviation of the error. 1.2 Japanese results Predicted field strength was compared with the result of field measurements for about 3 500 paths. Table 4 shows a summary of prediction accuracy statistics, as of 1999. Mean prediction error was 0.7 dB, and 70% of the errors were within 10 dB. 4 R
22、ep. ITU-R BT.2137 1.3 Comparison of measurements with field-strength predictions, Trondheim Area An extensive programme of DVB-T measurements is in progress in Norway and the measurement database functionality of the CHIRplus_BC software of LS Telcom. The measurement data has been provided by the br
23、oadcast operator, NORKRING. The Trondheim area with the transmitter at Mosvik with DVB-T on 722 MHz was one area of consideration. The measurements there range from 5.5 to 86 km distance to the transmitter. Furthermore, the terrain is highly irregular with elevations ranging from sea level to more t
24、han 1 000 m. The majority of measurement points have no direct sight to the transmitter. Initial results indicate that path specific 3-D models, that take into account reflections, have significant advantages in comparison with 2-D-models when dealing with mountainous terrain. Additionally, it is no
25、ted that passive echoes can contribute to the useful signal in a DVB-T system. 1.4 Australian results Signal strength data for surveys undertaken at 34 sites in 4 general locations on the Gold Coast in southern Queensland, Australia, were used as a base for comparison against prediction results. The
26、 Gold Coast region is generally suburban in nature, comprising mainly single and double-storey detached houses. The terrain ranges from open to densely vegetated (with trees taller than 15 m) and steeply undulating. The site and measurement information extracted from the survey reports was compared
27、to predicted field strengths. The propagation models used in the simulations were Recommendation ITU-R P.370-6 + RMD (reflection plus multiple diffraction loss), Recommendation ITU-R P.1546, Longley Rice v1.2.2, Anderson 2D v1.00 and Free Space + RMD. These propagation models used terrain data (appr
28、ox 90 m resolution) by default and clutter data (approximately 500 m resolution) when specified. A summary of the analysis is given in Table 1. Rep. ITU-R BT.2137 5 TABLE 1 Differences between predicted and measured signal levels for the Gold Coast region of Australia Recommen-dation ITU-R P.370+RMD
29、 Longley-Rice Anderson 2D Free-Space+RMD Recommen-dation ITU-R P.1546 Recommen-dation ITU-R P.370+RMD Longley-Rice Anderson 2D Free-Space+RMD No clutter No clutter No clutter No clutter With clutter With clutter With clutter With clutter Max 39.20 39.20 32.80 36.10 30.60 29.20 29.20 27.60 27.70 min
30、2.90 10.20 19.20 5.10 7.10 7.10 20.70 30.70 8.10 median 21.15 14.40 14.65 20.05 14.70 8.10 0.25 4.70 8.10 mean 21.38 12.86 15.01 20.54 13.21 9.39 0.87 4.95 9.47 std dev 8.62 13.34 11.88 8.43 10.60 9.12 13.56 12.84 9.25 NOTE Positive value indicates that predicted level was higher than the measured l
31、evel. 6 Rep. ITU-R BT.2137 2 Field-strength prediction method 2.1 The method used in the United Kingdom (the UKPM) The basis of UKPM is the prediction of received field strength at a location, taking into account the environment in between. This is based on the BBC field-strength prediction method,
32、the principles of which are described by Causebrook 1974 and which has been used and subsequently developed by all United Kingdom planning organizations. An overview of the field-strength prediction process is shown in Fig. 1. Initially a terrain and clutter profile is generated for the path between
33、 transmitter and receiver. Terrain heights are then corrected to take into account the curvature of the Earth. The effective earth radius used in this calculation is modified according to the time percentage required for the prediction. FIGURE 1 Overview of the field-strength prediction method The t
34、errain profile is processed to select the terrain points that would be touched if a string was stretched between the transmitter and receiver (see Fig. 2). These points are termed “running edges”. Adjacent running edges which are close together may be grouped into a single virtual edge. The terrain
35、diffraction algorithm then models the profile as a canonical object, (wedge, multiple knife edges or a cylinder) and computes the diffraction loss associated with these objects. Clutter losses, due to buildings and trees, are then calculated from the profile. Ducting and troposcatter losses are also
36、 taken into account, if the prediction is for a low percentage of time. In the remainder of this section, some of these procedures are presented in more detail. FIGURE 2 Definition of the running edges Rep. ITU-R BT.2137 7 Terrain data The terrain data used for the United Kingdom has a 50 m resoluti
37、on, as supplied by the Ordnance Survey and Ordnance Survey of Northern Ireland. For areas outside the United Kingdom, which are important when predicting interference into the United Kingdom from other countries, the terrain data used is from the GLOBE 30” dataset (approximately 1 km resolution). Cl
38、utter data The main propagation obstacles close to the receiver are likely to be buildings and vegetation. These are identified using a clutter database of the United Kingdom. The most detailed data is derived from aerial photography and provides clutter characterization at a resolution of 25 m. It
39、has 16 clutter classes and provides information on building and tree heights for major cities and towns. A 50 m resolution dataset is used for the remainder of the United Kingdom. This is derived from Land-sat satellite images and provides 10 clutter classifications and covers the whole of the Unite
40、d Kingdom. The two clutter data sets are combined to give the categories listed in Table 2. An example of the clutter map is shown in Fig. 3. TABLE 2 Clutter classification scheme Clutter class Building height (m) 1 Water 0 2 Open 0 3 Open in urban 0 4 Light Wood 0 5 Low Suburban 5 6 Embankment 8 7
41、Suburban 9 8 Wooded Suburban 9 9 Wood 0 10 High Suburban 12 11 High Embankment 15 12 Urban 18 13 Tall Wood 0 14 High Urban 27 15 City 40 16 High City 50 8 Rep. ITU-R BT.2137 FIGURE 3 Building clutter data Profile extraction The objective of the profile extraction algorithm is to compute the shortest
42、 path between the transmitter and the receiver terminals, and retrieve terrain and clutter information along this path. Given that the terrain and clutter data we use are projected via a Transverse Mercator (TM) projection, if the distance between the two terminals is short, the shortest path betwee
43、n them can be approximated by a straight line (on the map). However, when computing the coverage of a high-power broadcasting station, the distance between the receiver and the transmitter can be longer than 100 km, therefore this approximation is no longer valid. In the UKPM we have used an algorit
44、hm, described by Ordnance Survey1, that takes the Earths curvature into account, and therefore accurately traces the correct curve that corresponds to the shortest path. 1Ordnance Survey 1988 The ellipsoid and the Transverse Mercator Projection. Geodetic information paper No 1. version 2.2. Rep. ITU
45、-R BT.2137 9 A common use for the UKPM is for predictions where the transmitter and receiver are in different countries and therefore in different grid systems. This could be dealt with in a number of ways, for example segmenting the profile into sections, each belonging to a single grid system or c
46、omputing the latitudes and longitudes of the profile points along a great circle and then transforming these coordinates to the appropriate grid coordinates. However both these methods are unacceptably slow. For the UKPM we have adopted a simpler approach based on transforming all source terrain and
47、 clutter data to a single grid system. The single grid system used is an extended version of the United Kingdom National Grid System, which uses the Transverse Mercator projection1. The resulting algorithm is almost as fast as the extraction algorithm of the original BBC model, but is able to trace
48、the path profile much more accurately. Edge detection Given a profile, the goal of the edge detection algorithm is to compute the diffracting edges, as shown in Fig. 2. This is achieved in a recursive fashion as illustrated in Fig. 4. The execution time of this algorithm is proportional to the numbe
49、r of points in the profile as well as to the number of edges. By moving to higher resolutions, both these numbers increase, and as a result, the complexity of the edge detection algorithm is proportional to the square of the increase in resolution. FIGURE 4 Description of the edge detection algorithm 10 Rep. ITU-R BT.2137 Clutter loss computation The clutter loss algorithm is designed to take into account the effect of buildings and trees in the area near to the receiver. Clutter loss is calculated separately for buildings and trees using the terrain and clutter data for the
copyright@ 2008-2019 麦多课文库（www.mydoc123.com）网站版权所有
备案/许可证编号：苏ICP备17064731号-1