My First Design Part 3 – Main floor

UPDATE: When you finish reading this post, please see the addendum post HERE

Welcome to the next part of my first design saga! Now we get to the good stuff…

Here we get to see what every client wants to see, but often doesn’t really know how to interpret – fancy heat maps! There is a lot of potential here to mislead. It is very easy to make every heat map look a nice shade of green over the entire coverage area without any explanation as to what we’re actually looking at. Therefore, let’s start with my legends:

Signal Strength Color Legend (values in dBm):

SigStrLeg

The map is calibrated to show colors for values from the minimum requirement of -67dBm and higher (better signal). Any values below -67dBm and as low as -73dBm are shown in grey.

SNR Color Legend (values in dB):

SNRLeg

The map is calibrated to show colors for values ranging from the minimum requirement of 25dB and higher (better SNR). Any values below 25dB as low as 20dB are shown in grey.

Channel Overlap Color Legend:

CCILeg

The map is calibrated to show green for areas with no channel overlap (that is, only one AP on the channel over -86 dBm). Yellow indicates areas with 2 APs overlapping; and grey is anything more than 2.

This is exactly how it appears in the document delivered to my client, Aperture Science. It’s important that the heat maps give us a visual reference for where we are or are not meeting our design requirements (see part 1 and part 2!), so the legends are set so that grey shades are used anywhere where the model predicts that our requirements are not going to be met.

Let’s look at the main floor. It’s half the size of the two upper floors. First, what the client was looking for overall – AP placements:

dual AP This is an example AP transmitting in both the 2.4 GHz and 5GHz bands. The purple box indicates the 2.4 GHz channel. The orange box indicates the channel and channel width (40 MHz).

5 only AP This in an example AP transmitting in the 5 GHz band only. 2.4 GHz is disabled to avoid creating co-channel interference.

Main Floor

Main Floor - Access Points

Simple enough. Next, a brief summary of how we made out trying to meet our requirements. Color coding to match Ekahau’s health map for later.

Predicted Metrics

Main Floor

1F - Predicted Metrics

Next, a brief summary of areas the predictive design identified as possible problem areas:

Potential Problem Areas

Main Floor

Signal Strength – The areas which have some potential for issues are the north elevator, the walk-in cooler and freezer, and a small space in the radioactive storage room.

SNR – The predictive design for the main floor includes one small location where the minimum requirement is not met in the outside corner of the north elevator.

Number of APs – Minor spots in the cylinder storage room, walk-in cooler, radioactive storage room, and east fume hood room areas.

These two parts were repeated for each floor, at the beginning of the document. The idea is to give the less technical folks the high-level results in the first handful of pages of the design doc without having to wade through the techno-babble.

Further along, I provide the detailed heat maps. Starting with Signal Strength, against the requirement for TWO APs at -67 dBm or higher:

Appendix: Design Details – 5 GHz

Main Floor Details

Signal Strength – 2 Strongest APs

Color Legend (values in dBm):

SigStrLeg

Main Floor - Signal Strength (2 Strongest APs)

The areas which have some potential for issues are the north elevator, the walk-in cooler and freezer, and a small space in the radioactive storage room.

Here we can see the aforementioned potential problem areas. The elevator is in the top left corner and has a white area (-72 dBm or lower), but we agreed the elevators were not critical. The middle of the drawing has a large grey area (-67 to -71 dBm) which is inside the walk-in cooler/freezer. You can imagine that the steel construction is a strong attenuator here. Note that grey means that we’re within 5 dBm of our requirement, even if we haven’t met it. I can tell you that we do meet the requirement with a single AP, so it’s likely that a 7925 handset will be able to make a call without problems in the freezer, but we can’t guarantee a successful roam. Luckily for us, there’s not likely to be a lot of roaming going on whilst in the freezer.

There are also a couple of minor grey areas on the bottom of the drawing, in the edges of the service area, in shelves, concrete stairwells, and rooms that aren’t actually in our coverage area.

Also, I did provide all of the heatmap files in a separate zip archive, and included the heatmap for the single strongest AP; but, since the signal strength for a single AP is not a metric we’re using to meet our requirements, I did not use it in the main document (the document gets long with the rest of the heatmaps as it is).

I also provided the visualization statistics. I haven’t decided if I will continue to use these in future design documents:

SSstats

Less than 5% of the main floor area does not meet the signal strength requirement for the two strongest APs.

Next, SNR:

SNR

Color Legend (values in dB):

SNRLeg

Main Floor - SNR

The predictive design for the main floor includes one small location where the minimum requirement is not met, the outside corner of the north elevator.

Pretty solid I think.

Next, I briefly pointed out the visualization statistics for Data Rate (100% met) and Number or APs (98.2% met). The heat maps for these are less useful, so I didn’t include them in the document, and just pointed out that there was no concern with these metrics.

Then Channel Overlap, or Co-Channel Interference:

Channel Overlap

Color Legend:

CCILeg

Main Floor - Channel Overlap

Minor overlap of two APs in the lower rooms in the image.

Ok, we have a few areas where two APs on the same channel can be heard over our threshold of -86 dBm. This is not the end of the world, but it does mean that we do not have quite the full capacity of 14 APs across 40 MHz channels each; but we do have 96% of the area without overlap, and the experts I’ve heard from seem to agree that anything over 90% is quite good.

Now it’s important to remember here that this is the PREDICTIVE design, and therefore doesnt take into account interference from neighboring Wi-Fi radios that won’t be under Aperture’s control. This is a multi-tenant building; and looking at the floor plan for the main floor, this is about a quarter of the entire floor area in the building.

The good news is that Aperture is the only tenant on all three floors of this building’s East half. The main floor plan we’re looking at here is the North side of the East half, and the South side is also occupied only by Aperture. The building’s exterior facade is heavy brick masonry, and the construction should also do a good job of isolating radio signals from the West half of the building, so if I’m lucky, the CCI won’t be terribly higher than the predictive model. That being said, I am not banking on this assumption and the document is clear:

CCIstats

There is no channel overlap in the 96% of the coverage area. This will certainly change in the validation stage when some signals will be heard from sources outside of Aperture’s control.

Validation is important. We have no way of knowing whether or not we’ve met our design requirement unless we validate, and here I’m trying to imply that validation is not an optional step. This project was sold to the client with validation as an included component.

I also included Ekahau’s Health map as a high-level reference:

Main Floor – Overall Health & Problem Areas

1F - Health

Health Leg.png

Signal Strength – elevator and walk-in cooler, walk-in freezer areas. Minor reductions in signal at the east lab entry zone and radioactive storage room.

Number of APs – Minor spots in the cylinder storage room, walk-in cooler, radioactive storage room, and east fume hood room areas.

SNR – Minor spot inside the walk-in cooler/freezer which is actually within tolerance.

I’m not sure why, but Ekahau calls out SNR (blue tiles) in the cooler/freezer, even though the metric is met on the SNR heatmap. Jerry – if you’re reading this, feel free to comment if you can shed some light on this.

I’ll wrap this post up here, and we’ll look at the other floors next, then perhaps some of the specific obstacles on some of the floors that I’m curious to see how well my predictive results align with the real world behavior.

UPDATE: See the addendum post HERE

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s