PROPAGATION MODELS FOR URBAN ENVIRONMENT

For urban propagation, three distinct models can be used. These include propagation in macrocells, microcells, and indoor or picocells. In macrocells, the base station is often placed well above an average rooftop, while for microcells the base station is placed well below the average rooftop. In macrocells, the propagation path is dominated by the over the rooftop path, while for microcells reflections and diffraction from buildings and streets often dominate the propagation environment. For such environments, ray tracing-type simulation models are adequate and their use is justifiable. For picocells and indoor propagation, on the other hand, new challenges appear and improved propagation models and simulation tools are required to achieve reliable, accurate, and computationally efficient propagation predictions and to help overcome many of the indoor propagation impairments. Challenges facing the development of picocell simulation tools may include the following:

It is often argued that results from deterministic electromagnetic-based calculation models are not expressed in terms of parameters that can be used in the simulation of wireless communications systems. Parameters such as delay spread, coverage, direction of arrival, and bit error rate (BER) are necessary for system simulations and need to be incorporated as part of the simulation code development.

Four different types of methods are often used in developing propagation models, and the above listed limitations are expected to impact them differently. For example, statistical models provide parameters suitable for system simulations but lack specificity and accuracy. EM-based deterministic models, on the other hand, provide accurate and site specific coverage and delay spread information but are also very computationally inefficient and time consuming. Empirical and measurement-based models are site specific, frequency specific, and hence lack generality. Researchers use a combination of these methods to help improve the accuracy, broaden the generality, and reduce the required computational time. But much more research and development are needed to fully develop accurate, computationally efficient, and experimentally verified propagation models that may be used for broadband and highly mobile communications systems. With the advances in the signal processing methods and the development of communications algorithms, the envisioned propagation models are expected to play a critical role in the accurate accounting for mobility and the dynamic variation in the characteristics of the propagation channels.

With this in mind, the panel members tried to identify and possibly discuss the on-going R&D activities in this area of channel characterization and propagation models as we continued to travel in Europe and Japan. Only at Philips, CSELT, and Ericsson in Europe, and Matsushita Research Institute Tokyo (MRIT), KDD, and YRP in Japan did the panel identify research activities that the host was interested in sharing and discussing. The following provides a summary of these activities and a comparative study of the level of interest and the type of emphasis in each case. This summary will be presented according to the modeling and the characterization technique used in research activities at the visited sites.

Deterministic EM-Based Propagation Models

The WTEC panel identified strong research activities in this area at Philips and Ericsson. At Philips, new deterministic models for indoor propagation are being developed based on modal expansion techniques and using the Finite Difference Time Domain Method (FDTD). In both cases, both the EM field distributions and statistical parameters such as coverage and delay spread were being calculated (Dolmans 1997). Examples of the obtained results using a 2D FDTD code are shown in Fig. 4.1 for two different indoor propagation environments. Results from some of the statistical parameters calculations are shown in Fig. 4.2 where both the delay spread profile for propagating pulses of different widths and the coverage for single and diversity antennas were calculated. Results from these calculations are being experimentally evaluated using experimental set ups such as the one shown in Fig. 4.3. The important observation from this effort is related to the fact that efforts are being made to include calculations of statistical parameters of interest to system simulations and to verify the results experimentally. The use of 2D FDTD calculations emphasize the need for a more computationally efficient procedure to carry out 3D calculations often required in indoor simulations.

The research activities at Ericsson are closely tied with the European Cooperation in the Scientific and Technical Research (COST) in this area.

The COST 295 task force on "Wireless Flexible Personalized Communications" has a working group (Working Group 2) on propagation and antennas. This working group, chaired by Prof. E. Bonek from the Technical University of Vienna, has the following stated objectives:

As part of this activity, Mr. M. Steinbaur, also from the Technical University of Vienna, is leading an effort to provide characterization for wideband and time-varying directional channels and to ensure that models meet requirements posed by system and network simulations. A summary of the progress and discussion of activities of the COST 295 task force, working group 2, is available elsewhere (COST 295/260 Workshop 1999; Landstorfer 1999).

Fig. 4.1. Fluctuations of the electric field in the building that houses the faculty of Electrical Engineering. 2D FDTD code was used in these calculations which include the following: (a) all doors are open with no object present; (b) doors are closed with one object placed in one of the rooms. Green color denotes high electric field values; blue color represents low signal amplitudes.

Fig. 4.2. Coverage distribution in an indoor environment using space diversity receivers (a) 0-100%, (b) 98-100. The diversity combining consists of a switch (dashed-dotted), a selection receiver (short dashes), an equal-gain combiner (dotted), a maximum-ratio combine (long dashes). The coverage for a single antenna receiver is represented by the solid curve.

Fig. 4.3. A photograph and schematic diagram of the adaptive/diversity antenna measurement system (W.M.C. Dolmans, Eindhoven University of Technology/Philips Research Laboratory).

Ray Tracing-Based Propagation Models

The ray tracing method represents the most commonly used approach in the calculation of propagation models for terrestrial and urban environments. Several software packages are available (Bertoni et al. 1994; Liang et al. 1998; Rappaport et al. 1992), and some research efforts are underway to help in the continued improvement of the accuracy and extension of generality and to increase computational efficiency. The conventional ray tracing method is based on a ray launching and bouncing procedure that can be very inefficient if no speed-up algorithm is employed. There have been several schemes to accelerate this procedure including the image method, the bounding box method, and the utilization of the visibility approach (Landstorfer 1999; Liang et al. 1998; Catedra et al. 1998). Although these methods have their own advantages and specific domains of applications, more efficient methods are needed to cope with the complex and often computationally demanding indoor or indoor/outdoor situations while maintaining good accuracy of the propagation prediction results. Such research activities were found at CSELT in Italy and KDD Research and Development Laboratories in Japan. At CSELT results from several models were presented including site specific ray tracing predictions which were time consuming; ray-launching results, which were faster but lacked accuracy; and semi-deterministic methods, which were being used in small cell planning. For indoor propagation, however, empirical models based on experimental measurements were being used. Results from these calculations were presented in terms of field distributions in the propagating environment, delay spread distribution, and Doppler frequency distribution. An example of the presented results for a microcellular structure is shown in Fig. 4.4.

Fig. 4.4. Microcell structure (L.) and ray tracing results (R.) of delay profile in a parking lot (CSELT).

At KDD two tools were developed. These include the CSPLAN tool, which is used for cell site planning and coverage evaluation using low antenna height and 2D building shapes, and the BSPLA tool, which provides propagation predictions for mobile based station planning. For the BSPLA tool, path loss is evaluated based on geographic information of the propagation area. This effort at KDD points to other ongoing research activities in the area of channel characterization and propagation models development, including the use of geographic information and data from the Global Positioning System (GPS) to guide the development and enhance the accuracy and the computational efficiency of new propagation models for future wireless communications systems (Enge 1994; Kaplan 1996). This may provide significant advantages in systems that intend to incorporate dynamic variations in channel characteristics. For mobile multimedia wireless applications, incorporation of such capabilities may be crucial in enhancing the quality of service or making it even possible in the first place.

Propagation Models At Millimeter Waves

Besides the stated objective by the COST 295 working group, the issue of developing propagation models at millimeter wave frequencies was not discussed in any of the visited European sites. Two groups in Japan, however, discussed R&D activities in this area. These include the MRIT and the Stratospheric Wireless Access Network group at the Yokusuka Research Park (YRP). At MRIT new propagation models are being developed at millimeter wave frequencies, and emphasis is placed on the accurate accounting of multi-path analysis. At YRP, on the other hand, propagation models are being developed to support the development of a stratospheric platform. This unmanned High Altitude Platform Station (HAPS) is expected to fly at an altitude of 22 km, and 15 platforms are expected to provide coverage for all of Japan. Frequencies in the range from 2-20 GHz will be used with the continued increase in attainable data rates. Figure 4.5 shows that optical links will be used for inter-platform communications and radio links will be used for subscriber access. Propagation models that cover this entire frequency band are being developed, and research activities expected to improve the accuracy and enhance the computational efficiency will continue for some time.

Fig. 4.5. Schematic of the stratospheric wireless access network in Japan.

Empirical and Measurement-Based Propagation Models

Clearly many of these models are presently available and are being used in a variety of wireless services. Limited aspects of these models were, however, discussed in the visited sites during this study. As mentioned earlier, the COST 295 task force is involved in a comparative study of the effectiveness of empirical and statistical vs. electromagnetic deterministic models in different situations. The CSELT group is also using this type of empirical model for indoor propagation studies and channel predictions. The limited interest in this area may be justified based on several drawbacks including the following:

The main reason for the attractiveness of such approaches is computation speed. Furthermore, results from these measurements and empirical curve fitting efforts are sometimes used to complement ray tracing-type calculations and provide improved accuracy in areas where it is difficult or time consuming to incorporate numerical calculations of diffraction coefficients.

Technology Assessment

In addition to the information learned and collected during site visits, the WTEC panel tried to further assess research activities in this area by searching the INSPEC database for some of the key words related to this topic. Selected key words include propagation models, channel characterization, and indoor-outdoor propagation models. The results from the database search are summarized in tables 4.1a, 4.1b, and 4.1c in terms of the number of papers published or presented on this subject. From these results it may be noted that research on the development of propagation models is strongest in Europe, followed by the United States, while the research activities in the area of indoor-outdoor propagation models are rather limited (only 20 papers cited during the period from 1995-99) with major participation by the European community. These results also emphasize the need for new R&D for modeling micro- and pico-cells.

Table 4.1a
Propagation Models (1998-1999)*

U.S.A.

21

21%

Europe

38

38%

Japan

8

8%

Canada

6

6%

Others

27

27%




*Results of the database search using INSPEC. Total number of papers, 100.

Table 4.1b
Channel Characterization (1997-1999)*

U.S.A.

16

38%

Europe

19

45%

Japan

1

2.4%

Canada

2

4.8%

Others

4

9.5%




*total number of papers 42

Table 4.1c
Indoor-Outdoor Propagation (1995-1999)*

U.S.A.

3

15%

Europe

9

45%

Japan

1

5%

Canada

2

10%

Others

5

25%




*total number of papers 20

For the overall comparative study among Europe, Japan, and United States in the area of channel characterization and modeling, Table 4.2 was prepared to provide a qualitative comparison. From Table 4.2 it may be noted that while the majority of the available models are based on the computationally efficient statistical and empirical models, there is some growing interest in using EM-based deterministic models particularly in the United States followed by Europe. Efforts to integrate statistical parameters in the models are being mostly emphasized in Europe followed by the United States. It is expected that this trend will continue to grow because of the continued demand for improved accuracy and reliability in system designs and network planning. Table 4.2 also shows some growing interest in modeling new wireless communications systems in the millimeter wave frequency range in both Japan and the United States, while much of the European activity is presently focused on addressing the present industrial needs at lower RF frequencies.

Table 4.2
Research Activities in the Channel Characterization and Propagation Models in Europe, Japan and the United States

 

Europe

Japan

USA

Statistical/empirical

***

***

***

EM based deterministic

**

*

***

Integrative models

(Statistical parameters based on deterministic models)

**

-

*

Microwave and millimeter wave

*

**

**




*Qualitative results of a comparative study illustrating the level of activities and the focus of research in Europe, Japan, and the United States in the area of channel characterization and propagation models.


Published: July 2000; WTEC Hyper-Librarian