« Getting Cultured | Main | Dancing with the Small Cell Stars »

February 09, 2013

TRUTH OR DARE? Let's Clear the Air.

Guy-with-long-nose Don't get us wrong. We love Pinocchio.  

But when he rears his ugly nose into our business we get, well....all up IN it.  

Sometimes things can get a little complex, technical and confusing. We're big on simplifying things and being brutally honest about technlogy and it's real value - whether it's ours or not. So here goes.

Cisco's new whitepaper "All_Beamforming Solutions are not Equal" is at best misleading and at worst mythology.

In this paper, Cisco effectively got two things right:
1) they spelled the word "beamforming" properly and
2) they used the appropriate Cisco corporate logo.


FUDDING UP

Cisco is clearly going out of their way to give the impression that Ruckus-patented BeamFlex and digital signal processing (DSP)-based beamforming (often called chip-based transmit beamforming or TxBF) techniques are categorically similar. This is patentedly false.

BeamFlex and TxBF are radically different in function, benefit, and usefulness. What’s more important is that product vendors with only DSP-based TxBF would like to suggest that adaptive antenna switching somehow prevents supporting chip-based TxBF as well. 

Ruckus supports both. The benefits of BeamFlex sit on top of those brought by chip-based beamforming. The gain of these techniques is cumulative. Why pretend that customers must pick one or the other when they can get both? 

For IT managers, administrators, and engineers alike, 802.11n’s MIMO techniques and all this other WI-Fi technical mumbo jumbo can be both complex and confusing – making vendor marketing (admittedly) an opaque fog.  To that end here's some absoute truth (facts that are true for all people, at all times, in all places) that should help. Cisco-Response-Cover

Customer have LOTS of choices and can be easily deceived by supplier marketing.  Yet at the end of the day, customers just want the truth. 

SAFETY TIP: If you want to avoide reading further, just download, read and memorize this document.

FOR THE TECHNICALLY MINDED 
(you know who you are)

Cisco’s ClientLink uses a type of chip-based beamforming with implicit client feedback. In theory, implicit transmit beamforming (TxBF) such as ClientLink allows an AP to create multiple downlink beams that constructively add together at the client to improve the overall signal. To do this successfully, the AP relies on what it hears from the client (the “implicit” feedback, which is really no feedback at all) to determine how best to steer the beams. 

  1. ClientLink is a chip-based MIMO technique that cannot be used on the same antenna at the same time as other chip-based MIMO techniques like Spatial Multiplexing (SM), Space-Time Block Coding (STBC), and Cyclic Delay Diversity (CDD), so TxBF comes with a tradeoff from the start. 
  2. ClientLink forms its beams based on how the AP “hears” and not based on how the client “hears.” 
  3. Implicit beamforming uses received signals to determine the relative signal offset for each downlink beam. Theoretically, this works fine for fixed clients, but for mobile clients, every downlink frame will be optimized for the client’s previous location, and not for its current location. 
  4. TxBF such as ClientLink requires incredibly precise phase and amplitude weights for constructive gain to occur exactly at the client location. When the weights are not accurate, destructive interference occurs and can actually reduce signal quality instead of improving it. 

Implicit.jpg

Perhaps the actions of the IEEE’s 802.11 working group— removing implicit beamforming entirely—speak the loudest. For more evidence on this topic, researchers from leading universities readily acknowledge the problem of non-reciprocity and the faults in implicit beamforming.

For beamforming to work successfully, transmitted signals from the AP must constructively combine at the client’s antenna(s). CalibrationWhat happens if a client device has more than one antenna? In such a case, it is not possible for the AP to determine which client antenna(s) the AP heard from. Given that almost all notebooks and tablets as well as most mobile phones (even ones with a single Tx/Rx chain) have multiple antennas, ClientLink is becoming applicable to an ever-dwindling minority of clients— SISO devices without antenna diversity—such as the old iPhone4 with the well-documented “antennagate” problem.

Also, it's important to note that the throughputs in Cisco's plot are north of 200Mbps. This means that the client is at least a three-stream MIMO 802.11n client. But ClientLink can't work with multi-antenna clients because it needs the channel state information (CSI) between EACH antenna combination. And with multi-antenna clients it's simply NOT physically possible to get the per antenna CSI without client cooperation (and client cooperation requires standardization).

 Finally, Cisco attempts to dismiss the gains of PD-MRC, but this is simply because their brand of beamforming can't support it. IEEE channel models measure PD-MRC gains in the 7-15 dB range for LOS conditions and in the 3-5 dB range for NLOS conditions

Ultimately marketing is relatively simple. But building RF features that truly impact real clients in everyday networks isn't. End-users need to review third-party testing that includes both Cisco and Ruckus products. Better yet, test both products in your own environment, with your expected client devices and applications.

At the end of the day, the worst thing about being lied to is knowing you're not worth the truth.

Comments

Cecil Coe

Love your sense of humor in this write-up, coupled with explanation that is easy to understand.

Post a comment

Comments are moderated, and will not appear on this weblog until the author has approved them.

If you have a TypeKey or TypePad account, please Sign In.