Single page Print

Hardware configuration
With dozens of possible configuration combinations to choose from, the prospect of documenting everything I tried before running my final tests is matched only in tediousness by the thought of what might happen if I forced someone to try and read them. Suffice it to say that multiple physical locations for both the router and extender were investigated using Wi-Fi Analyzer from [gasp!] the Windows Store before I chose the final placement.

Additionally, I checked all manner of firmware settings on both the router and extender using Network Performance Test, a UWP implementation of iPerf3. Then I picked the ones that offered the best overall performance and finalized my testing parameters. It turns out that I needed to enable beamforming and use Asus' ExpressWay feature to maintain the fastest, most reliable connection between the router, the extender, and their clients.

ExpressWay dedicates either the 2.4GHz or 5GHz radios on the router and extender to inter-device communication. This leaves the other frequency open for clients to connect to without the chance of interfering with the link between the router and extender. It turned out that the typically shorter-range 5GHz connection was significantly faster than the further reaching 2.4GHz one even at ~200' (61m) apart, since I was able to set up nearly a direct line-of-sight between both ends of the link. That result was exactly what I was hoping for, because it meant that the all-important internet connection would be broadcast further on a frequency better at penetrating walls and other obstacles (and out by the lake, there isn't much worry about interference). By the way, the typical link speed reported between the RT-AC88U and RP-AC68U in this configuration was about 175 Mbps, or 3-4x the reported 2.4GHz link speed at the same distance.

The other Asus-specific feature that turned out to be quite important was Roaming Assistant. Roaming Assistant is a method for accomplishing what folks used to professional networking solutions may take for granted: handing off clients from access point to access point. Consumer networking gear doesn't normally concern itself with that scenario, as there is typically only one device serving up the connection. Enabling Roaming Assistant defaults to -70 dBm as the threshold for disconnecting a client and forcing the client to switch to the better signal, and it proved to be the best value to use. Without this feature enabled, my laptop would desperately cling to a poor signal even though a much better one was readily available. The one downside to Roaming Assistant that I found was that it appeared to still disconnect clients even if they didn't have a better signal to switch to. That wasn't really a practical problem for the location, though.

Internet speed test results
The conditionally formatted matrixes below are populated with data from running internet speed tests. I used Speakeasy's speed test from its Chicago server, since it's my go-to test for general-purpose use. The ISP for the connection is Charter Spectrum, and my service tier is 60Mbps down and 4Mbps up. There are some local benchmarks coming, as well, but keep in mind that the goal of this installation is not to maximize speed, but instead to ensure coverage and reliability for the TR BBQ. All tests were performed multiple times, and the fastest repeatable result was recorded to give the benefit of the doubt to uncontrollable internet and environmental conditions.

Here's where understanding the layout of the testing area will come in handy and hopefully help you apply what you learn from this review to your own real-world Wi-Fi implementations. The left-to-right dimension on the chart represents west-to-east data, while reading top to bottom corresponds to north-to-south data. Just think of it how any normal map would display on your screen. Keep in mind that the center of the grid is where I am standing in the 360-degree photo.


Average download speed: 13.95 Mbps

This first set of results is from the router that Asus' duo ousted from duty. For the curious, it's a Netgear WNR1000v3. Right off the bat, you can tell there's a problem because even highest value recorded can't quite reach half of the internet connection speed. I should mention that I tested the old router in the location it normally resides in, not the new location that the RT-AC88U was placed in. That only puts them about 20' (6m) apart, but line of sight through widows and the number of walls the signal is passing through is slightly different.

Now let's see how the new hotness performs. We'll start with tests performed with only a connection to the router.


Average download speed: 47.43 Mbps

That's a lot better. Of course, that was obviously going to be the case going against our seven-year-old competition. The old router was just a point of reference, not anything to draw real comparisons to. It's worth noting how big of an impact the location of the buildings in the yard have on the performance of the router, though. Even though it's probably safe to assume that the RT-AC88U is bottlenecked by the internet connection at many of the locations measured, it's obvious that that dropping a building in the path of the signal can have a pretty significant impact on performance.

The group of roughly 40-Mbps tests on the upper right of the download grid show just how much the bunkhouse reduces the signal strength compared to the clear line of sight at same distance in the middle of the grid. Similarly, the garage reduces the download speed to 25 Mbps on the bottom of the grid, but the line of sight opens up again another 45' (13.7m) to the west, allowing the link to achieve more than twice the speed of the neighboring grid segment.

The most important thing to note is that down the middle of the grid on the left side, where the front cottage stands in the way, there is effectively a connectivity dead zone. It's an unfortunate spot for a coverage drop. The deck situated there overlooks the big lake, and it's a popular place for gerbils to congregate. Hopefully the RP-AC68U can help out here. These tests were performed with only a connection to the extender.


Average download speed: 31.14 Mbps

Well, at least there's no dead zone on the west deck anymore. Performance overall is not as good as the RT-AC88U mothership, but we can see that the RP-AC68U is capable of delivering the full speed of my internet connection so long as the client is in close proximity. Even though it doesn't fill up the bars quite as far, it is interesting to note that there are no completely dead zones in the test area while I was connected to the extender.

As with the router, you can see the impact of the smaller buildings on the property, but from the opposite direction. The 24Mbps download speed in the center of the bottom of the grid shows the garage coming into play again before a better signal appears 45' (13.7m) further east. The worst measurement on the upload grid is once again on the popular eastern deck. That just won't do for our needs.

Let's see what happens when the wonder twins unite.


Average download speed: 47.04 Mbps

Interesting. What's immediately apparent is that the upload grid has no dead zones or anything remotely approaching a slowdown. That's good news for the two of you reading this that will be watching the BBQ live stream. Upon closer inspection, you can see that some of the grid sections aren't displaying the fastest measurements we've taken for those locations. In most cases that's likely due to the the Roaming Assistant feature and the -70 dBm threshold we're running with. It makes sense that if the signal from one device or the other is still strong enough the client will stay connected to it instead of switching to a technically better option. That behavior is dependent on the adapter, though, and investigating that client-to-client difference is beyond the scope of this review.

It's a little stranger that there are a couple cases where one device or the other running solo outperforms what we see when they are working together. For example, in the top row, second column, the roughly 18-Mbps result is significantly lower than the 34 Mbps the router delivers alone. It also trails the 39-Mbps rate the extender delivers by itself. In another result, the opposite case is true. Take a look at the second column of the second row. When I connected to the extender alone, the transfer rate is measured at 55 Mbps. The same cell on the individual grids measures ~37 Mbps.

I'm afraid I don't have an explanation for why those results are what they are, but as I mentioned up front, this is real-world testing. I've tried to eliminate all the variables that I can, but I'm not shocked that there are a couple inconsistences. Wi-Fi is notoriously tricky to bench in the best of conditions. In any case, I consider a couple performance drops a fair trade for seamlessly eliminating dead zones.