RF models expected to improve with new built environment data

Mobile and fixed wireless broadband networks provide the backbone for communicating voice, text and data, yet accurate RF modelling remains a challenge for the telecommunications industry.

Fixed wireless broadband services are increasing across Australia as an alternative to fixed line services, particularly in rural and urban fringe areas. And, as mobile technologies evolve, higher frequency bands with a shorter range and greater potential to be impacted by environmental conditions are being adopted. 4G and 5G RF signals, while being faster and more efficient, can be absorbed, reflected and scattered.

To accommodate these technology changes and maximise their potential, RF modelling and network planning is becoming more of an exact science. RF models that don’t accurately reflect the built environment can give rise to unexpected network black spots, delay service rollout and require additional field testing and site visits. Access to more granular data that accurately describes potential environmental impacts on RF signal becomes more critical in ensuring connectivity.

PSMA Australia hypothesised that by incorporating built environment data from its Geoscape dataset into RF models it could improve RF model accuracy and support better network design.

Geoscape is a digital representation of Australia’s built environment across the entire continent. It captures every building with a roof area greater than nine square metres (15,243,669 buildings), tree heights and surface cover at two-metre resolution for urban and remote communities with a population greater than 200, and surface cover for the entire continent at 30-metre resolution.

Geoscape data useful for RF modelling includes building location, elevation, maximum roof height, eave height and footprint, and tree heights and surface cover. The use of three-dimensional, vector-based buildings data in RF modelling would previously only have occurred in pockets across Australia, because the data has not been available on a national level until now.

PSMA Australia added three-dimensional building polygons to an RF propagation model, introducing RF shadows, which may assist with more accurate identification of black spots. (See Figures 1 and 2.)

Figure 1: RF model without 3D buildings data

Figure 2: RF model with 3D buildings data

In general, lower resolution RF models with less data layers generate coverage models that inaccurately predict a good signal across a larger area. Adding data layers — building polygons and clutter — to the propagation model had a significant impact on the count of buildings that were considered to have a good signal. This intelligence could support planning decisions at a macro scale before conducting field tests.

All Geoscape buildings are attributed with query-able data. Geoscape has linkages to other standardised, national datasets such as property, cadastre and the Geocoded National Address File (G-NAF), which is available under open data terms at data.gov.au. Those buildings deemed to have poor RF coverage or be in a black spot area can easily be queried for building attributes and address details.

At a micro level, an address point has traditionally been used to determine if a property falls within a fixed wireless network coverage area. However, address points do not accurately reflect building locations on a property. Incorporating Geoscape data attributes, such as the building centroid, tree and roof heights, can help assess coverage, signal strength degradation due to obstructions and possible antenna install options before a site visit is conducted.

Three-dimensional building and high resolution clutter data can also be used within a RF model to enable identification of areas of strong and weak signal across a single building or property. Modelling variations in signal strength across a building or property may support the identification of alternate and optimised install locations for fixed wireless receivers.

The next step for PSMA Australia is to tune and validate the generated RF models against drive test data and subsequently prove the concept.