After reading about Vivato's newly announced Wi-Fi switch, I have some questions.
The first issue is related to the hidden node problem. 802.11 devices use CSMA/CA, or "listen before talking". The hidden node problem most commonly occurs when two STAs are associated to the same AP, but the STAs cannot hear each other, so they both try to talk to the AP at the same time. The RTS/CTS sequence is a bandwidth-reservation mechanism that is included in 802.11 as a way to mitigate this problem. However, RTS/CTS is an optional feature since it reduces throughput. Vivato says there is no required changes to the STAs that wish to talk to their switch, so I'm assuming that means RTS/CTS is not required. If this is the case, I do not see how a switch with "3 steerable Wi-Fi beams" will be able to handle STAs that cannot hear each other very well. The hidden node problem is exacerbated when you extend the range of STAs so they can't hear each other. Ask any wireless ISP about it! Without bandwidth reservation (RTS/CTS) or polling (e.g. PCF), nothing is preventing STAs from all talking to the Vivato switch at the same time. For a normal AP, simultaneously received messages can corrupt each other, because APs use omni-directional antennas. The Vivato switch, which effectively sets up point-to-point links, is less susceptible to interference. If multiple STAs transmit to the switch at the same time, the Vivato switch may be pointing to only one of the STAs, so the message may not be clobbered by the others.
To make things worse, the Vivato switch can communicate simultaneously on the 3 non-overlapping channels in the FCC domain (1, 6, 11). This means, with 3 steerable beams, there is only 1 beam per channel. How is this different from a normal AP, apart from the range increase?
Next issue: why two gigabit and two 10/100 ethernet ports? The Vivato switch has 3 steerable beams, each delivering data at 11 Mbps (max throughput of 6 to 8 Mbps, due to overhead). So, 3 beams x 8 Mbps/beam = 24 Mbps data throughput. Isn't a single 10/100 enough to handle this? The only reason I can think of for 2 of each port is so the Vivato can act as a "real" ethernet switch for existing 10/100 or gigabit drops. Kind of an expensive switch...
That said, the Vivato product appears to be fully featured and enterprise grade. It also appears to be an attractive alternative to complex WLAN installations, and as a way to simplify WLAN maintenance tasks.
I can't wait to see the statistics from one of these Vivato switches in action!
Posted by Patrick at February 18, 2003 04:51 PM | TrackBack