I understand, that 802.11x protocolls can use different type of modulations. BPSK, QPSK, 16-QAM, 64-QAM... I also understand that the used modulation depends on the quality of the channel, I guess on the SNR. But how does the actuall process look like? How does the transmitter decide what modulation should be used?
\$\begingroup\$
\$\endgroup\$
Add a comment
|
2 Answers
\$\begingroup\$
\$\endgroup\$
8
As usual, it depends.
Let's restrict ourselves to the OFDM-based IEEE802.11 standards (a,g,n,ac,p, and many more).
These have a typical frame structure, in which one symbol, which is always BPSK-modulated, contains the rate setting info (and thus, implicitly, the modulation used).
As an example: Got this graphic from keysight
See the violet symbol.
Notice that in OFDM, you can use different modulations for different subcarriers, or in fact, you can freely pick your modulation for each frame and subcarrier. You don't typically do that in Wifi (the standards restrict your freedom here, to varying degrees, depending, again, on the individual 802.11 standard); in 4G/LTE OFDM systems, you can expect a frame of the downlink signal to be aimed at a whole set of different receivers, which all pick out "their" part in the subcarrier/symbol plane – and thus, you can expect different robustnesses to apply to different ressource blocks. Yay, complexity!
In essence: for wifi-style OFDM 802.11, the transmitter decides on what modulation it uses for the payload data itself. The optimal choice depends on a lot of factors, some of which are receiver capabilities, throughput and power consumption limits, and an estimate for the interference and noise levels at the receiver.
Thus, the algorithms picking an optimal modulation for any specific instant in any specific use case for any specific x in 802.11x are very diverse, and, often, very highly optimized by the chipset/driver vendor, and thus can be very obscure.
answered May 18, 2017 at 13:19
-
\$\begingroup\$ ODFM modulation for WiFi is highly efficient for burst speeds at the expense of minimum SNR but highly sensitive to motion with Rician fading loss and then recalibrates. While the ODFM used in mobile is much more tolerant to doppler effects and phase nulls of Ricean fading loss. Like other serial protocols "autobaud" down on errors is included. \$\endgroup\$– D.A.S.Commented May 18, 2017 at 13:35
-
1\$\begingroup\$ well, but for serial protocols, the channel is typically assumed to be rather slowly changing, if at all. I mean, the whole "recalibration" is what we tend to call "equalization" in wireless comms, isn't it ? :) \$\endgroup\$ Commented May 18, 2017 at 13:36
-
\$\begingroup\$ I found flaws with WiFi recalibration or dynamic EQ and autobaud, such that in a situation where I had poor WiFi , I got MUCH better performance with my methods than the auto EQ methods in the standard. in weak fringe levels of <-70 to -80dBm I get better disabling n,g and enabling only b protoco and force to 11Mbps then adjust stable position of a laptop by 1mm or 1deg increments until flow is maximized. I learned how to do this using the RSSI signal then the flow rates by imagining the Ricean Sensitivity vs path vector and was able to get it to work and often 10x faster thruput. \$\endgroup\$– D.A.S.Commented May 18, 2017 at 13:51
-
\$\begingroup\$ heh, nice :) Point is that I don't think among broadcom/qualcomm/atheros, intel, and the other "big players" anyone really adheres to the standards – there's simply too much heuristic knowledge on how to optimize real-world networks. For example, in theory, occupying the channel for as shortly as possible by using the modulation that maximizes eff. throughput (considering retransmissions lost throughput) is the "right thing", but in practice, you'll often find drivers screw that principle and rather optimize for max inst. through for loads looking like error-corrected file transfer; and f.… \$\endgroup\$ Commented May 18, 2017 at 14:08
-
\$\begingroup\$ … and for robustness in situations where load is low, anyway, or where retransmissions are less acceptable (voip?). \$\endgroup\$ Commented May 18, 2017 at 14:09
\$\begingroup\$
\$\endgroup\$
Modulation and Coding Scheme (MCS) indices determine the data rate in 802.11, based on the SNR observed. Each index uses a pre-defined modulation scheme (upto 256-QAM in VHT modes) and a coding rate (upto 5/6), which in turn influences the data rate.
Note that higher your data rate, higher would be the SNR requirement. This should give you a neat idea.
