There are many things we do within their software to maximize prediction accuracy and optimize the design of networks – 3D modeling, inclined surfaces, body loss modeling, etc – but one of the less talked about yet more powerful ways to maximize prediction results is by using the ‘Prediction Calibration’ feature in the software.

What does it do?

Simply put, with prediction calibration, you can use your survey measurements to calibrate the prediction results of your design.

Why do we have this?

Because no matter how accurately you model your venues with the right materials, you don’t always know what is behind the materials (think steal beams behind drywall for example), or what else is impacting the signal attenuation in a particular space. By using real data to calibrate the prediction algorithm in the software, you will end up with the most realistic prediction results.

Let’s look at the potential difference it can make.

Prediction Results with No Calibration

This floorplan is a basic small/medium sized office space – survey has been done to collect the measurements, and the 3 AP design has been completed.

Looking at the RSSI heatmap, and probing a specific area of the floor you can see the following signal strength predictions for 2.4GHz/5GHz.

	RSSI Prediction Value (dBm)
2.4GHz	-55.72
5 GHz	-69.21

And when you run the compliance report showing the prediction results against the design requirements for each band, you can see that neither 2.4GHz or 5GHz requirements are met.

Prediction Results with Calibration

After setting up the calibration model using the measurement data collected during the site survey, the prediction results change and show the following values:

	RSSI Value (dBm)
2.4GHz	-49.19
5 GHz	-58.00

And re-running the compliance report, you can see that each band now meets it’s KPI design requirements.

Comparing Before and After Calibration

To look at the overall difference prediction calibration can make, let’s compare results of with and without calibration applied to the design:

	Without Calibration (dBm)	With Calibration (dBm)	Delta
2.4GHz	-55.72	-49.19	-6.53
5 GHz	-69.21	-58.00	-11.21

As you can see, the difference can make a meaningful difference, especially when you consider that prior to calibration the design was not meeting their KPI requirements for either the 2.4GHz or 5GHz bands. In this case, with no calibration, the designer may have made unnecessary design adjustments or added more equipment when was not necessary, adding more cost to the design. When using real measurements from the venue to calibrate prediction, you can achieve more realistic prediction results as well.

Wirelessly yours,

Kelly

Forever on a quest to improve the accuracy of network predictions in our software, we recently released a new feature called “Body Loss Modeling.” With Body Loss Modeling, you can now account for the attenuation caused by bodies packed into a tight space together in your design – most useful for high-density venues like stadiums, arenas, or conference centers.

In this blog, I use the design of a basketball arena to examine the impact the body loss modeling feature can have on the prediction results of a network design.

I do that by isolating a small section of the arena seating, placing an Access Point and then looking at the results of both the Signal Strength and SNR heatmaps under two scenarios:

No Body Loss Modeling
With Body Loss Modeling

At the end, I’ll summarize the comparison and discuss the potential impact of the results.

Here is the basketball arena I am using, and the specific seating area looked at in this blog.?

Results: No Body Loss Modeling

Keeping the prediction zone identified as a regular prediction area, I ran the Signal Strength and SNR heatmaps for the 5GHZ band and then used the “Probe” tool to zone in a very specific seating area in the bottom right hand side of the prediction zone (circled).

Here are the results.

Signal Strength Heatmap Results

58.85 dBM {Inclined Surface Area}
58.95 dBM {Horizontal Surface Area}

And zoomed in… ?

SNR Heatmap Results

29.55 dB (Inclined)
29.75 dB (Horizontal)

And zoomed in …?

Results: With Body Loss Modeling

Next, I assigned the same prediction area as a ‘Body Loss Zone’ and then re-ran the Signal Strength and SNR heatmap prediction results. To identify a body loss zone in iBwave Wi-Fi or iBwave Design, you have to first configure the ‘Body Loss Zone’ (unless you just want to use the default), and then assign your prediction area as that particular body loss zone.

Here is the configuration I set up and called ‘Arena Seating’ ?

And here is how I assigned the prediction area as the body loss zone I configured above. ?

With the prediction area now identified as a ‘body loss’ zone, the prediction engine will factor in attenuation caused by tightly packed bodies within that seating area.

Here are the results ?

Signal Strength Heatmap

Incline Surface: 68.20 dBm
Horizontal Surface: 67.88 dBm

SNR Heatmap

Incline Surface: 21.28 dB
Horizontal Surface: 21.44 dB

Comparing Results

To easily compare the prediction results with and without body loss factored in, I put the results into a table.

	No Body Loss	With Body Loss	The Difference
Signal Strength	58.85 dBm	68.20 dBm	-9.35 dB
SNR	29.55 dB	21.28 dB	-8.27 dB

Looking at the table, you can start to see the potential impact that modeling bodies in high-density environments can have on the accuracy of prediction results – and thus on the potential performance of the network post-installation.

In this case, before I modeled body loss into the design, the signal strength is predicted to perform pretty decently with a 58.85 dBM signal strength. With the attenuation due to bodies factored in, the signal strength loses almost 10 dB, which pushes it towards a much less desirable signal strength and could significantly impact the user experience when it comes to critical applications like video streaming or VoWiFi.

Looking at the SNR heatmap, a similar story is supported, even emphasized – before body loss is considered, the SNR sits at a pretty acceptable level of 29.55 dB. After body loss is factored in, the SNR level drops to 21.28 dB – making it even more likely that those critical apps will work as expected for the user.

For the network engineer designing the network, this means she or he needs to factor in that while prediction results without body loss factored in can show acceptable performance results, it could be misleading in high-density venues – which can lead to undesirable and costly consequences later on.

When prediction during the design phase is not accurate, it can lead to more site visits post-install, and possibly re-design work which is all more downtime and cost for the property owner.

How do you factor in body loss into your wireless designs? Let me know in the comments below.