When I was introduced to the YFI multi-radio 802.11a/g AP STA unit technology, I was at first skeptical about YFI’s claims that it’s technology would perform much better than single-radio 802.11n/ac AP STA unit technology in high density environments. As it was my job, I did a comparative analysis of both technologies and, indeed, YFI’s claims were true. Over time, data about real-world high-density installations confirmed the analysis that 802.11/n/ac performed as predicted. 

Cable companies and, apparently, enterprises have tested YFI's multi-radio 802.11a/g AP STA unit using both the 2.4 and 5.0 GHz bands. The testing was/is done because deploymens of 802.11n and 802.11ac have failed to deliver the high performance expected from their appealing characteristics (e.g., 20, 40, 80 & 160 MHz variable channel widths.
 
Thus, the testing of multi-radio 802.11a/g 20 MHz AP STAs comprising extended service set WLANs was/is to prove that the system performs substantially better than WLANs composed of 802.11ac AP STAs for similar high-density areas and use cases.
 
Notes:

  1. the 802.11n-2009 Supplement had been released, product was available, and data re its real-world use was being reported;
  2. the 802.11ac-2018 Supplement had not been released but drafts of it were available, of course product was not yet available, and of course there was no real-world use reported;
  3. the 802.11ax Supplement was practicaly unheard of.
User image
802.11n and 802.11ac MAC Throughput rates assuming 65% of maximum information rates.
 
Notes re the figure above:
 
  • 802.11ac Wave1: supports up to 1.3 Gbps data rates on 3 spatial streams with 80 MHz channel bonding.
 
  • 802.11ac Wave2: supports up to 3.47 Gbps data rates on 4 spatial streams with 160 MHz channel bonding. These numbers are only the theoretical numbers from the standard, differences will apply depending on the specific AP datasheet.
 
  • 802.11ac is not directly defined in data rates speed, but is rather a combination of 10 modulation encoding scheme (MCS 0 to MCS 9), a channel width ranging from 20mhz (1 channel) to 160Mhz (8 channels), a number of spatial streams (typically 1 to 4). The short or long Guard Interval (GI) will also add around a 10% modification to this.
 
Importantly, though, the 802.11n/ac analysis results were disturbing because it showed that the effective information rates of both single-radio 802.11n and single-radio 802.11ac WLANs in a range of general areas and demanding use cases would be nowhere near what they would achieve in greenfield areas with favorable use cases. Yet, it was those incredibly high information rates that were being advertised and fed though the trade press. Indeed, it became clear that some of the appealing characteristics of 802.11n and especially for 802.11ac where apparently increasingly impractical. E.g.,
 
  • Variable Channel Width expansion protocol from 20 MHz to 40, 80 and 100 promised increased effective information rates by 2, 4 and 8 times, but in general areas and use cases actually dramatically reduces the effective information rate;
 
  • Quadrature amplitude modulation (QAM) symbol numbers 4-QAM, 16-QAM and 64-QAM each results in a reduction of the maximum range of a single spatial stream to about 10 yards, but that of 256-QAM is  down to a few yards;
 
  • Multiple-input and multiple-output (MIMO) 2, 3, 4, … 8 spatial stream protocol promised performance increases of 2, 3, 4, … 8 times but regulatory transmit power limits and practical transmit power limits of radio frequency (RF) power amplifiers reduced the range for each additional spatial stream, and the typical transmit antenna separations and receive antenna separations further reduced to ranges;
 
  • beamforming becomes ineffective due to increasing co-channel interference in higher-density areas;
 
  • Multi-user MIMO (MU-MIMO) is dependent on MIMO efficiency, and so is similarly affected;
 
  • Frame aggregation (a feature that increases throughput by sending two or more data frames into a single MAC service data unit (MSDU) or MAC protocol data unit (MPDU) frame into a single frame) can give higher average throughput due to a reduction in MAC and PHY protocol  overheads but the overheads for assembly of the fames for transmission, longer serial transmission/reception times,  and individual receiver disassembly and throwing of data meant for other receivers make it ineffective;
 
  • Etc.
 
Furthermore, with the information available about 802.11ax, its two additional attractive characteristics similarly provide increases in the maximum information rates in greenfield areas with favorable use cases but in a range of general areas and demanding use cases the effective increases in the maximum information rates will far less:
 
  • Orthogonal frequency-division multiple access (OFDMA) allows two additional users per 20 MHz of channel width (at the cost of 20% reduction in information rate due a practical increase in each symbols guard interval (GI) to account for path delays from different user radio STAs), multiplying the multi-user capability for 40, 80 and 160 MHz wide channel, but of course the Variable Channel Width but in general areas and use cases actually dramatically reduces probability of 160, 80, 40 and even 20 MHz wide channels, thus practically eliminating the additional multi-user capability;
 
  • 1024-QAM symbol numbers further reduces the maximum effective range down to a few feet.
 
It is inconceivable that some, if not most, members of the 802.11n/ac/ax Amendment working groups were aware not aware that the desirable new characteristics would not provide anywhere near the greenfield area and favorable use case information rates in most general areas and demanding use cases. Indeed, it is reasonable to conclude that these appealing characteristics where deliberately included in the 802.11n/ac/ax Revisions to promote Wi-Fi chip (set), AP STA unit, smart phone client, etc. sales. Although, there may have been no overt collusion planning such a strategy, it is very apparent that the strategy worked and is continuing to work. 
 
In a following next post, I will demonstrate how the 802.11n/ac/ax MIMO Spatial Streams appealing characteristic can provide high performance (six times) in a greenfield but provide no better than a single spatial stream over much shorter distances in a range of general areas and demanding use cases. Combined with the other appealing characteristics, 802.11n/ac/ax APs provide practically no better performance than 802.11a/g APs and certainly less performance than a multi-radio 802.11a/g AP.