Cisco and friends chase WiFi's searing speeds with new cable standard Cisco is buddying up with three semiconductor companies to push Cat 5e and Cat 6 connections beyond the gigabit mark, to bring the Ethernet world into line with speeds on the table from future WiFi network standards. The Borg, along with Freescale, Xilinx and Aquantia, has looked at enterprise cabling and found it wanting: …
"It's not as if WiFi ever gets remotely close to the maximum theoretical bit rate"
A computer rag did some speed tests a while back on the then new AC routers, and found unless you were no more than several feet away, you are never going to get the fastest speed it's capable of (and as you said, nowhere the "theoretical" speeds).
Once you go past a couple of walls, and a couple of dozen metres distance that would be typical in a home setup, the speeds drop to just barely over what cat5 could do. And considering Cat5e is "good enough" for ethernet gigabit speeds over short lengths, not only does cable win, it's by far more reliable, stable, and repeatable than wifi.
WiFi is useful for convenicence, and it has its place there, but if you want speed and stability, stop reading the bullshit theoretical numbers and use cable.
This post has been deleted by its author
Essentially an open interface which would not transmit the demodulated data, but have the down mixed signal on the line. Essentially the IQ signal coming out of the first part of the chip set put onto sets of pairs on the Cat5 cable. The degradation of the signal on Cat5 would be accounted for in the same step as the one caused by the air interface.
That way you could easily upgrade your infrastructure for any new standard. Plus you could trivially do things like beam forming or MIMO since you have all your signals at a central point.
If I may elaborate. The IQ signals are only as wide as the bandwidth of the signal and they are near DC, so it's easy to transmit 100 MHz wide signals. The degradation is, while making digital pulses hard to transmit, very insignificant compared to what already happens in the air. Since it's also similar you can get away with only doing it once, instead of dealing with it on the access point, then on the cable again.
So what you are saying is that instead of having the modulator/demodulator next to the RF stages and antennas, you put it at the other end of a Cat5 cable?
And this allows you to more easily upgrade your infrastructure? <Sarc>Why, because you don't have to climb a ladder to where the RF/antennas are to change the modulator/demodulator - you just change them in the rack where they are fitted instead? </Sarc> If you seriously change the modulation scheme, you still have to change the modulator/demodulator unless you built enough flexibility into the original design, and in this case it doesn't matter where it physically is.
Also beam forming invloves adding a phase delay into the RF signal. I'm not sure that you could do it at any stage other than RF. Beam forming is also to direct the signal from a particular set of antennas that are close together. If you have a whole site that was say 300m accross and tried beam forming, the beams would only really be formed at a significant distance from all the antennas (likely beyond the range for WiFi), and would not really work inside the perimeter of the antennas. Meanwhile, beamforming accross the 6 antennas of a single AP works very nicely because you are outside the perimeter of the antennas, and at 2-3m away you are at a significant distance compared to the 10-15cm between the antennas.
Look up "Ground Based Beam Forming" as applies to satellites, such a ViaSat-1. You'll find that they already have the conceptual-similar patents, at least Applied For.
In your example, you'd need N cables, one per IQ signal, and you'll find one of those IQ signals per antenna (2, 3, 4, 6?). You've also apparently neglected, or not mentioned, the transmit side. Sounds like multiple cables replacing one cable.
You're also solving a problem that does not exist. As mentioned, it's not that difficult to fetch a ladder if required. In fact, if you think it through, you're making the upgrade path far worse. Vastly worse. Vendor lock-in. Incompatibilities. Horrible. Much easier to buy a new box and plug it in. Something I've done at least a half-dozen times; cheaply. No ladder needed in my case.
Well the transmit side would work the same.
Essentially what you need to transmit is:
a) a frequency reference (i.e. 10 MHz)
b) the received signal (for example downshifted by an amount derived from a)
c) the signal to send, also downshifted.
You could transmit all of those, including power on one pair of wires. The 2,4 GHz range has around 100 MHz of Bandwidth.Category 6 cables provide 250 MHz of bandwidth, so you can easily stack receive and send on top of each other on a single pair. So you have one pair of wires per antenna. Since you already need 4 pairs for gigabit Ethernet, you can run 4 antenna MIMO on it.
I should stress that that estimation is for a node that covers the _whole_ 2.4 GHz range. For 5 GHz you would however have to limit yourself to slices of the band as it's wider.
"...between their laptop and Netflix."
Netflix tops out at about 6 Mbps, so it's a pathetically thin data stream.
Hey. Does YouTube have a setting to fix it permanently to the highest available video quality? Every time I click on another video, it defaults back to 480p and loads the file instantly. Then I set it to 1080p (again), and it reloads the file again. Waste of time.
"Why do they make Access Points with advertised wireless speeds in the Gigabits and the Etherned connection still at 10/100?"
Opposite here. My wifi routers are certainly not claiming to be Gb on the radio Wifi side, but they certainly are Gb on the wired Ethernet side. Other than some clients, the whole household network is Cat 6 STP and Gb.
That would be useful, since currently cat5e maxes out at gigabit speeds, or at least isn't rated for 10GbaseT but may work on shorter runs. Cat6/6a is rated for 10GbaseT, so the utility of 5 Gbps for it is debatable unless 10GbaseT is significantly more expensive to implement for some reason.
I'm sure its not all that cheap today, but that's always true for next gen ethernet standards at first. If this splits the market between 10GbaseT and "NbaseT" it will only increase costs of both by reducing the production quantities of each. More likely if Cisco is successful in pushing this, before long we'll see 10G/5G/2.5G/1G/100M/10M autonegotiation chipsets so it won't end up mattering much.
Seems like people are no longer putting their trust in the 802.3* working groups, considering I've heard of fiber standards for 40 and 100 Gbps, but proposals to push 25, 50, and 400 as alternative standards as well as these alternative copper standards.
Seems like a way for Cisco to gouge its current customer base.
a) Either put more copper in to the building with potentially more APs
b) APs with LAGG/LACP
c) Upgrade to a nifty 5GBs switch.
Even with 50 people connected to an AP I think it is reasonably hard to saturated the 1Gb link for more than a short peak time.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
