When I benchmarked five 802.11n Draft 2.0 wireless routers for Maximum PC’s November 2007 issue, I attributed the performance shortcomings of many of the routers to the harsh environment in which I tested them: a mid-rise office building in which numerous other wireless networks were also operating. I finished building my home in rural Northern California shortly after that, which I’ll now be using as a test lab for wireless networking products, media streamers, speakers, and similar products (I'm calling it Max PC Lab North). I recently retested each of those routers in that far quieter environment to see what would happen.
My home is located on 10 acres of what was once a dairy farm, so I don’t think my neighbors’ wireless networks—if they even had them—would interfere with mine. I expected to see a big leap in performance in this relatively RF-sterile environment. To my surprise, the numbers weren’t tremendously different.
I won’t make apples-to-apples comparisons to my in-the-office testing, because the environments are extremely different. The walls in the office environment contain metal studs and both the floor and the ceiling are concrete. The walls in my home contain wooden studs, the floor is concrete, but the ceiling was drywall. Radio signals don’t pass through concrete very effectively, so they could have bounced between the floor and the ceiling in the office environment. They might also have bounced off the metal studs to increase the signal’s range. And since the MIMO technology in the 802.11n standard takes advantage of reflected signals, this could have increased the router’s range. In order to make the comparison as meaningful as I could, I used the same benchmark utility in both tests; namely, Ixia’s QCheck.
At close range in my home-environment tests (five feet between the notebook and the router), Netgear’s WNR854T beat the stuffing out of the rest of the field, averaging 91.5Mb/sec. This didn’t surprise me, because it also performed well in my office test. Buffalo’s WZR2-G300N came in second place in the close-range test, delivering 66.7Mb/sec. But if your notebook is this close to the router, you might as well plug it in. The second test scenario is more meaningful, with the notebook in an adjacent room 20 feet from the router, and Netgear’s WNR854T was again the winner, delivering TCP throughput of 55.6Mb/sec.
But before you crown Netgear’s router as the new king of hill, take a look at its performance when I moved the notebook into my media room: TCP throughput fell off a cliff, averaging an unusable 1.9Mb/sec. But at least I was able to maintain a connection with the Netgear; Belkin’s N1 and Buffalo’s WZR2-G300N couldn’t reach the notebook at all while it was in that room. Linksys’ WRT350N managed to squeeze a few bits through, but at an equally pathetic rate of just 1.9Mb/sec.
D-Link’s DIR-655, which was the top-performing router in the office tests, was the only router that managed to connect to the notebook in all five of my real-world testing scenarios. It lagged the rest of the field at close range, but it came in second place when the notebook was in my kitchen (Environment 2, below); it delivered throughput of nearly 22Mb/sec when the notebook was in my media room. It also delivered throughput that was faster than any of the other routers when I placed to notebook outside my house, and it was the only router that maintained a connection to the notebook (albeit, just barely) in both outdoor tests.
D-Link's DIR-655 delivered the best throughput in both our test scenarios.
Based on these tests, I’ll be using D-Link’s DIR-655 for at home, but I still don’t think any of the current generation of 802.11n equipment is worthy of a Maximum PC’s Kick Ass award.
| ENVIRON 1
|| ENVIRON 2
|| ENVIRON 3
|| ENVIRON 4
Speed measured in megabits per second. N/C = No Connection. Best scores in each category are bolded.
I didn’t test all the configurations in my home testing that I did for the feature. I set each router to its highest performance setting (802.11n only, with channel bonding enabled if the router in question supported such a configuration). The house I used for testing is a single-story, ranch-style home built on a concrete slab using 2x6 framing with fiber-cement clapboard siding. The interior partition walls are constructed of 0.5-inch drywall over 2x4 framing, with the exception of the media room (Environment 3 in the charts).
The media room consists of 2x6 interior framing with 0.5-inch drywall on one side. A second 2x4 interior framing wall is constructed six inches apart from the first and covered by two sheets of 0.5-inch drywall. The gap in between the two walls is filled by two layers of R19 insulation. The ceiling in this room is also covered by a double thickness of drywall. Since this is a room within a room, I installed two solid-core Masonite doors (one opens out into the foyer and the other opens in to the room) to further deaden the room and prevent sound from leaking out.
Each router was placed on top of a hutch-style desk (5.5 feet off the floor). I installed an 802.11n Draft 2.0 adapter card in a notebook PC configured with Windows XP and benchmarked each router in five different environments, which you can see here:
The router is placed on top of the hutch and the notebook PC is on the desk, about five feet from the router.
The notebook is located on a granite countertop, about 20 feet from the router, with one wall and a set of plywood cabinets separating the PC from the router.
The notebook is on a coffee table inside this double-walled, double-doored media room, about 35 feet from the router. There are five walls (three of which are insulated) separating the notebook from the router. (The doors are an exterior type in order to reduce sound leakage.)
The notebook is outside the house, 90 feet from and at on oblique angle to the router. There are two interior and one exterior walls (all of which are insulated) separating the notebook from the router.
The notebook is located outside the house 85 feet and at a right angle to the router. There is a set of plywood cabinets, four interior walls, and one exterior wall (all of which are insulated) separating the notebook from the router.
The closest house is quite a distance away, so there shouldn't be any router conflicts that would cause problems with channel bonding.