Page 4 - Performance Tests
For the tests, the wireless router was placed in the middle of the main floor of my house. An ASUSTOR AS3202T network attached storage equipped with a single Seagate NAS HDD ST4000VN000 4TB was connected to the router via a CAT5e cable on a Gigabit Ethernet connection. On the client side, a 2015 13" Apple MacBook Pro running Totusoft's LAN Speed Test application was used to transfer 500MB test files to evaluate real-world throughput. In our results, "upload" is defined as data transfer from the client to the server via the wireless router; conversely, "download" is defined as data transfer from the server to the client via the wireless router.
A brief description of the test locations is as follows:
- Location 1: Line of sight to router, approximately 2m distance
- Location 2: Non-line of sight to router, bedroom, one floor up
- Location 3: Non-line of sight to router, driveway in front of attached garage, same level
- Location 4: Non-line of sight to router, near end of backyard, one floor down
- Location 5: Non-line of sight to router, open area, one floor down
Compared Hardware:
- Linksys Velop (AC2200 Mesh)
- Linksys WRT3200ACM (AC3200)
- TP-Link Archer C3150 (AC3150)
Since wireless channels are generally characterized by path loss, large scale fading, and small scale fading, the router was tested in five different locations described above to comprehensively measure its true throughput performance. This includes a combination of line of sight and non-light of sight spots, different distance and positions relative to the router, as well as shadowing caused by objects between the laptop and the router. In order to overcome inconsistencies due to small scale fading, a relatively large 500MB test file was used. Furthermore, movement of people and objects within the vicinity of the devices during testing was eliminated whenever possible. To test the Velop mesh system with all three nodes, one additional node was placed near Location 3 and one between Location 4 and 5 to strategically exploit the advantage of a mesh system.
Due to the Velop's reliance on the 5GHz band as the backhaul, one node cannot be placed too far away from another node. The 5GHz band suffers from considerable attenuation over distance, and it is not as good as going through walls compared to 2.4GHz despite having much more bandwidth. From my experience, a Velop extension node can only be placed about one room away from the nearest Velop node. One more wall or a bit more distance will disqualify the node from meeting minimum performance requirements.
In Location 1, the Linksys Velop with a single node fell quite a bit behind the high-performance routers on the list. This should not come as much of a surprise, considering the Velop nodes are only rated at AC2200 compared to AC3150 for the Archer and AC3200 for the WRT3200ACM in this line-of-sight test. However, what was surprising was when I deployed two additional nodes, the mesh system suddenly matched the performance of its higher performance brethren. I have repeated all the tests multiple times and came out with the same results in the end. In Location 2, there was no appreciable difference between having one node and three nodes, likely due to the fact my computer was still connected to the same node. Although the Velop's download performance was comparable to the performance routers, the upload performance left quite a bit to be desired. The one node Velop had a lot of trouble reaching Location 3 due to power attenuation through a thick garage wall. An additional node placed just on the inside wall separating the house and the garage boosted performance considerably, but it still failed to outperform the WRT3200ACM. I could not place the additional node in the garage itself, because it was not able to communicate with the main node. The single Velop configuration also had a lot of trouble with Location 4 due to more distance and walls. Deploying an extra node one floor immediately below the primary node fixed the performance problems, but again, it was still slightly slower in download speed and only achieved half the upload speed compared to the WRT3200ACM. In Location 5, despite the laptop's close proximity with the newly deployed node, did not considerably outperform the single node Velop system due bottlenecks in the backhaul. Connecting directly to the central node one floor below was not much different than connecting to an additional node located one floor below with one hop.
Based on the graphs above, I was a little disappointed with the Linksys Velop. Even with three nodes, it could not outperform the company's own WRT3200ACM in any of the tests, unless you count the extra 15 Mbps upload in Location 1 as a win -- and even this could very well be flipped around if the same test was run again. I do not think this is a problem with the concept of a mesh wireless system, but rather as a matter of three factors. Firstly, the Velop is designed to be slower than the WRT3200ACM, which is fair. But this does not explain why its performance was so poor when the distance increased and extra walls came in between. This brings us to my second point: The Velop has weak antennas. It just does not have enough power or the sensitivity needed to overcome negative channel conditions. Thirdly, it is the way I designed the test location. The main node or router is in the center of all our test locations. As such, all additional nodes are always one hop away from the main node. While this is technically the correct way to design a wireless network, the Velop aims to solve problems where the main node may be placed on one side of the house, and you need to extend the signal throughout the entire house in one direction, resulting in a multi-hop design to eliminate dead spots. While this is technically not the correct way to design your wireless network, the Velop will outperform a single router in this setting, which was not demonstrated in my tests.
Page Index
1. Introduction, Packaging, Specifications
2. Physical Look - Hardware
3. Configuration and User Interface
4. Performance Tests
5. Conclusion
