5G Network Analysis

There are three simulation approaches when analyzing 5G networks.

Fifth generation (5G) network planning has several unique characteristics, such as the use of antenna arrays at the base stations that can direct relatively narrow beams at individual mobile stations. As base stations are changing the directions in which beams are pointing (beam switching) to serve mobile stations (see Figure 1), their coverage and their interference change. This leads to simulation approaches that differ from the ones used for pre-5G simulations. Therefore, 5G air-interface files (.wst) contain a keyword, such as 5G_FDD or 5G_TDD, that triggers a few small but essential changes. The most important changes are related to simulation strategies to deal with antenna-pattern information.
Figure 1. Examples of beam patterns achieved with phased arrays at a base station (top view).

Three simulation approaches exist for 5G network planning:
  1. with isotropic antennas
  2. with envelope antenna patterns that represent a set of possible beams
  3. with individual beams

Simulation Based on Isotropic Base-Station Pattern

One possible approach is to specify omni-directional (isotropic) antennas. When you do this in a 5G network planning project, ProMan enables you to specify additional antenna gains to account for beamforming. This is done during carrier assignment, see Figure 2.
Figure 2. Example of beamforming gains added to an envelope pattern in a 5G project on the Carrier dialog.

For a given mobile station, one base station antenna is the server, and the other ones are interferers (if they use the same carrier frequency). The server points its beam to the mobile station, so the expected beam gain to be added to the isotropic antenna is positive. The interferers are serving their own mobile stations and are likely to be pointing their beams in other directions, at least most of the time. Therefore, the gain to be added to those antennas is likely to be small and may be negative.

Simulation Based on Envelope Antenna Pattern

The second possible approach is to simulate with the envelope antenna pattern (see Figure 3) of all possible beams. This assumes that, if a mobile station is in the cell, the base-station array directs its best beam at that mobile station. For that mobile station, the gain of the envelope pattern equals the gain of the specific beam, so the propagation result is correct.

Figure 3. Envelope antenna pattern for a set of four beams (usually the set would be larger).

Note: An envelope pattern is different from a traditional sector-antenna pattern.
The envelope pattern is the envelope of all possible beams, which are relatively narrow, so it has a higher gain than a typical sector pattern. The integral of the gain over all directions will be higher than that of a regular antenna pattern.

All mobile stations in the same cell receive the best beam for their location, so the envelope pattern provides correct propagation results. Individual mobile stations in the cell may operate on different sub-carriers or be served in different time slots.

For interference from other cells, the gain of the envelope antenna pattern of the interfering cell has to be corrected, because an envelope pattern (in the simulation) will send more power into neighboring cells than one beam (of an actual base station) would. This correction is specified in ProMan during carrier assignment when using an envelope pattern, see Figure 4.

Figure 4. Example of beamforming gains added to an envelope pattern in a 5G project.

To provide values for additional gain in the interfering case is not hard science. There will be occasional time slots where a narrow beam from an interfering cell points right at a mobile station in a serving cell (correction would be 0 dB, worst case). There will also be more time slots where an interfering beam points in a completely different direction (correction would be -20 dB or lower; only minor interference). The default values of -7 dB and -11 dB are seen as a time average of the occasional worst case and the more-frequent cases where the interfering beam points elsewhere.

Simulation Based on Individual Beams

The final possible approach is to simulate with individual beams, see Figure 1. Since this requires knowledge of where the mobile stations are, this is usually combined with a Monte-Carlo analysis. When setting up a Monte-Carlo analysis, the offered traffic, which can be location-dependent and service dependent, is specified. During the Monte-Carlo analysis, ProMan generates snapshots with semi-random distributions of mobile stations to be served and performs the network analysis for each snapshot. Since the beams are known, no additional gains due to beamforming need to be specified (contrary to the approaches with omni-directional patterns and with envelope patterns).

Antenna Masking in Network Planning

In 5G network analysis, an antenna can have different antenna patterns for control and for data. Furthermore, depending on the simulation approach, many different beams are possible. To avoid the need to perform many simulations, the option Apply Antenna Masking for Network Planning (Propagation Results with Omni Pattern) exists on the Edit Project Parameter, Network tab.
Figure 5. The Edit Project Parameter dialog, Network tab.

Select the Apply Antenna Masking for Network Planning (Propagation Results with Omni Pattern) check box:
  • Propagation simulation is performed, and results displayed, with an omni-directional (isotropic) pattern.
  • Network planning is performed by adjusting propagation results based on actual antenna patterns for control and for data. These could be envelope antenna patterns or individual beams.

The antenna masking feature uses the ray matrix to post-process the propagation paths with each one of the individual beams to compute the power result maps for each beam individually. The best serving beam will be selected accordingly.

Clear the Apply Antenna Masking for Network Planning (Propagation Results with Omni Pattern) check box (traditional mode):
  • Propagation simulation is performed with one antenna pattern.
  • Network planning is based on one antenna pattern for both control and for data.

Masking mode is necessary for Monte Carlo simulations with individual beams, and is useful for simulations with envelope patterns when those for control and for data differ significantly. Traditional mode is attractive for simulations with envelope patterns when the patterns for control and for data do not differ significantly. In the latter case, the display of the propagation results are more informative. In both masking mode and traditional mode, only one propagation simulation per antenna is performed.

Beam Switching Using a Predefined Set of Beams

ProMan supports 5G beam switching capability using a predefined set of beams. Users can be served by a dedicated beam, with the possibility to transition to a different beam as they move.

The workflow for using beam switching is the following:
  1. Generate the beams, see Beamforming in 5G.
  2. Select the antenna pattern.
    • Click Project > Edit Project Parameter.
    • On the Edit Project Parameter dialog, click the Sites tab.
    • Add a site and for each site, specify the antenna.
    • On the Cell dialog, under the Antenna Pattern group, click Direction / Sector antenna.
    • Under Antenna Pattern, browse to the antenna pattern to be used.
  3. On the Edit Project Parameter dialog, click the Network tab.
    • Under Network Simulation Mode, select the Apply Antenna Masking for Network Planning (Propagation Results with Omni Pattern) check box.
  4. Define the users.
    • Add a stationary or time-variant user point. A user point can also be time-variant and attach to a defined trajectory.

      If no users are defined, all data beams are active.

      If users are defined, only the best-serving data beams are active.

  5. Switch back to the area/rectangle prediction mode.
  6. Click Computation > Run All Computations.
  7. The data channel results are accessible from the result tree under Results: Network > Data Channels. Similarly, control channel results are accessible from the result tree under Results: Network > Control Channels.
  8. If the user point is time variant, select the time step.
Figure 6. Left: Only the data beams that are serving a user are active (the best serving beam is selected. Centre: If the user moves, it switches to another predefined beam. Right: The serving beam is not necessarily oriented in the direct direction, it can for example, come from a reflection at the building.