Presto is Overture's innovative signal prediction engine, and is offers the fastest and most accurate single-path predictions available. The speed of the Presto engine means that users do not have to maintain huge caches of signal and prediction files. The accuracy of Presto means more reliable performance estimates and more effective optimization.

Presto extends the following industry-standard prediction models:

• COST231
• Okumura-Hata
• Stanford University Interim (SUI)

To improve accuracy, these empirical models are augmented with terrain diffraction, antenna pattern modeling, and radial interpolation calculations, each of which contributes significantly to prediction accuracy. This can be improved further by tuning to measured data.

Terrain Shadowing

Whilst terrain shadowing and diffraction modeling is a common tool in radio prediction tools, they are often limited by the time taken to calculate diffraction knife-edges and artifacts in the underlying terrain data.

Presto predictions are performed on a radial basis with a fast, one-shot, algorithm for determining knife-edges. Wide area predictions for a multi-sector site are available within a fraction of a second, which makes the planning process much faster and more interactive.

Most terrain data comes in a raster file (a regular grid of height values). This type of data can present a challenge for prediction modeling because it is difficult to understand where there are real edges (such as hill tops) and where there are false edges between fast-changing bin values. These bin-edge artifacts can end up causing noise in the signal map as shown below:

By default, Overture smoothes the terrain data to eliminate bin edges. This eliminates the artifacts in the predicted signal.

Radial Interpolation

The artifacts have been significantly reduced, but clearly some artifacts remain. These are a feature of the sampling process: when Presto calculates a prediction, it does so by sampling points radially from the radiation center. In the example prediction this is being done every 2° around the site and in 30m steps along each radial. Presto can also perform interpolation along and between radials to remove these radial-edge artifacts.

Whilst this could also be done by increasing the number of radials per prediction, it would increase the runtime and storage overhead significantly.

Antenna Pattern Interpolation

Antenna radiation patterns are typically supplied as two or more "slices" (horizontal and vertical) through the actual radiation pattern. This is partly a limitation of the rigs that are used to measure the radiation and partly because the radiation away from these cardinal slices is low.

Presto takes full account of whatever pattern information is available, including multiple slices and electrical tilt variations, and produces highly accurate extrapolations for areas without data.

As pattern information is supplied in increments of 1° (sometimes more), Presto also interpolates between these angular bins to eliminate artifacts as shown below.

Smaller Files

Presto computes signal files on-the-fly, so does not require additional disk storage. The files that are stored on the disk (pathloss and inclination) are stored in a novel compressed format that greatly reduces disk storage.

All data that is stored on the disk is done so with full arithmetic precision, so pathloss and inclination values retain their maximum accuracy.

Tuneable Parameters

All Presto prediction models have the same core set of parameters and particular models add additional parameters where necessary. The table below describes the core and model-specific parameters that can be tuned to improve prediction accuracy. Overture will modify these parameters when automatically tuning the prediction models.

Model	Parameter	Typical Range	Description
Common	Base Offset	30 - 60	The fixed adjustment to apply to all pathlosses in dB units. Typically adjusted to correct for mean errors.
	Relative Height Adjustment	0 - 10	Controls how much the relative height between the transmitter and the reciever modifies pathloss.
	Long Range Diffraction Exponent	0.1 - 1.0	Controls how much diffraction losses between the transmitter and reciever should modulate pathloss.
	Short Range Diffraction Exponent	0.1 - 1.0	When the receiver is below the obstruction height, the signal is assumed to diffract down from the top of the obstruction. This parameter controls the relative size of that loss.
COST 231 & Okumura-Hata	Distance Dependence	44 - 47	Controls how quickly the signal drops with distance.

For a more detailed description of these parameters, see the in-application help.