WiFi terminology - by an amateur for amateurs

This is not intended as a dictionary. Read it straight through to get some mental knobs to hang stuff on. Revisit in whole or in part as needed.

• 2x2, 3x3, 4x4, 2rx2tx, etc
  Various ways of saying how many antennas there are for transmitting and receiving signals in a MIMO setup. You will also encounter 2x2:2 or 3x3:3 or variations thereof. The ':n'-part says something about the number of spatial streams.
  Specific hardware may handle fewer spatial streams than the number of antennas otherwise may indicate. In this case, the extra antenna is used for diversity. Link
  
  
• 802.11a/b/g/n/ac
  These are the wireless standards, but you knew that already. Note that 802.11a and 11ac are valid in the 5GHz band only. 11b and 11g is 2.4GHz only. You may want to avoid clients with 11a, 11b or 11g on your network, as they spend much more time “on air” than 11n or 11ac for the same amount of data transmitted. Link. A very interesting graphic illustrating how we went from the 54Mbps of 802.11a/g to 600Mbps 802.11n can be found on page 5 of this PDF.
  
  
• 802.11ad
  AKA WiGig. 60Mhz, max transmission rate 7Gbps. Products do exist. Not widely used yet.
  
  
• 802.11r/k
  When a client realises that the link to the AP is gone (or going), it starts looking for another access point with the same SSID. If the client finds one, it initiates a handoff. The client and the new AP now starts what is termed a 4-way handshake.
  With WPA2, WMM, 802.1x etc. added, the amount of initial traffic required to establish a connection goes up a lot, causing actual user traffic to stall. 802.11r is a protocol designed to pre-authenticate to other APs, such that the handoff is reduced to the original 4-way handshake. Thereby reducing the time associated with a handoff from one AP to another.
  802.11k is related, sort of.
  
  
• 802.11e/WMM/WME/QoS
  Provides mechanisms to give preference to some type of traffic or specific clients. Useful for time and/or jitter-sensitive data like real-time voice communication (Voice over WLAN), video streaming and battery-operated clients.
  
  
• Antenna
  WiFi-antennas are most often rather simple devices receiving and radiating the electromagnetic signal. They can be dual-band or single-band (2.4 and 5 Ghz). They have a gain measured in dB, which states how much a transmitted or received signal is strengthened.
  
  A regular wifi antenna transmits a signal which best can be described as a donut shape, with the antenna pointing through the hole in the donut. Higher gain antennas transmits a flatter, wider donut than antennas with less gain. See the third grapic in this link.
  
  Antenna gain applies to both transmit and receive.
  
  
• Beamforming
  Beamforming is the technology used to give the signal a boost in a particular direction. The glossy commercials present beamforming as a way to send a laser-sharp signals straight to the receiving end. The reality is more like an uneven donut. The signal gain is typically 2-3 dB, which may be enough to extend the range a bit, or enough to raise the bitrate one level at the same range.
  
  Beamforming comes in two flavors: explicit and implicit. Explicit beamforming means that the client is aware of beamforming happing and actively contributes in the process of finding the optimal signal generated by the Access Point/Router. The client needs explicit support for this in hardware and the device-driver.
  
  Implicit beamforming means that the client is unaware of beamforming taking place. It works regardless of support in the client hardware or software.
  
  Beamforming is an optional feature of both 802.11n and 11ac. Not all vendors implement it, and worse is that explicit beamforming for 11n most likely requires equipment from the same vendor in both ends, as the standard does not mandate interoperability. For 11ac the standard is tighter, and interoperability should be ok for explicit beamforming, if implemented.
  
  The cost of beamforming is slightly reduced througput, as the process requires some non-user traffic. If the beamforming results in a higher MCS index, this may make up (and some) for the reduced throughput. See these links for further reading. Link1 , Link2
  
  The FCC appears to think that the use of beamforming requires reducing the outputpower equal to the antenna gain. If this is the case, one may wonder if beamforning actually provides any net gain at all, or if its only benefit is to reduce interference in all other directions. I have not made any effort to verify this claim.
  
  
• Device drivers
  Device drivers are part of the operating system kernel on the computer where the WiFi hardware is installed. In this context, the computer may be a client machine or a wireless accesspoint/router. Device drivers may be provided by the hardware manufacturer, either as a compiled binary, as closed source provided under NDA or as open source code. Or the device driver may be written by someone not associated with the manufacturer at all.
  
  Open source code provided by the manufacturer and complying with established kernel interfaces is preferred, as it allows outsiders to compile the driver with any kernel they like, possibly adapt the driver to other operating systems, and generally tinker with the code.
  
  Manufacturers are often wary about providing open source code for a number of reasons, most of which have nothing to do with technology at all.
  
  
• DFS
  Dynamic Frequency Selection ensures that we automatically avoid channels with interference, like weather radars and such. This is a mandatory requirement for equipment in the 5GHz band.
  
  
• Firmware
  Most WiFi hardware require firmware to operate. This is a piece of code loaded on the wifi hardware, and runs on the microprocessor on the actual wifi device (card/dongle/onboard). Firmware occasionally gets updated to fix bugs and/or add features. This is a closed source binary provided by the hardware manufacturer. The firmware ensures low-level protocol compliance, among other things.
  
  To clear up some potential confusion: a wireless router comes with firmware. This would be the complete operating system of the wireless router, including drivers for the wireless interface, the ethernet interfaces and the webserver presenting a GUI to the end user. Then there is firmware on the wireless interface itself, most often provided by the wifi interface manufacturer to the wireless router manufacturer. This may or may not be part of the driver. The point is: firmware may mean more than one thing.
  
  
• Hostapd
  This is the meat of the matter. Hostapd is a user space daemon for access points and authentication servers.
  
  When the WiFi hardware supports the low-level protocols and options you require, when the firmware of said hardware does the same, and when the kernel driver enables full use of the hardware, then hostapd is what implements an actual Access Point in the *WRTs.
  
  
• HT20/HT40/VHT
  ...later..
  
  
• Interface modes/Station roles
  ...later...
  
  
• MIMO
  Multiple Input Multiple Output. Uses multiple antennas for input and output, in order to create a better signal and increase bitrates. Link
  
  
• SU-MIMO
  Plain MIMO. Single-User MIMO.
  
  
• MU-MIMO
  As MIMO, but permitting multiple transmitters to send separate signals and multiple receivers to receive separate signals simultaneously in the same frequency band. Hence, Multi-User MIMO. Link
  
  
• Transmission bitrate vs throughput
  In short: the radio link spends a lot of data ensuring that the traffic the user wants to get through, gets through unmolested. There is a lot of redundancy, there is a lot of handshaking, there is a lot of waiting for the channel to be clear, and so on and so forth. The end result is that the end user never is going to get 1300Mbit throughput from an AC1900 router. Exactly how much throughput one may get depends on many factors. If you really need Gigabit throughput, a wired link (Ethernet) is better. You will get way better latency as well.
  Transmission bitrates for 802.11ac here.
  Transmission bitrates for 802.11n here.
  Note how the actual bitrate is the result of the MCS (Modulation and Coding Scheme) index, number of parallell streams, the channel width and the guard interval between symbols. All these parameters are largely autotuned based on hardware design, signal quality, interference and so on and so forth.
  
  
• Transmission Power and Receive Sensitivity
  The maximum power with which signals are transmitted from a client is defined in the relevant standards. (a/b/g/n/ac). The maximum power level may differ in various geographic areas. The standards most likely defines a maximum EIRP, rather than exact power levels delivered to the antenna. EIRP is the sum of power from the amplifier and the antenna gain. As one can expect, the higher the EIRP, the longer range your signal will have.
  
  The Receive Sensitivity says something about how weak signals you can translate into meaningful data. The various standards sets a minimum sensitivity for the equipment to be compliant. Better hardware got better sensitivity than the minimum defined in the standard. Better sensitivity translates into longer range and/or higher transmission rates.
  
  
• Wave1/Wave2
  The first 802.11ac specification is retroactively dubbed Wave1. Wave2 (or 802.11ac -2013 update) boosts the maximum transmission rate primarily through the use of MU-MIMO and improved beamforming. Devices certified for either wave may not necessarily implement all parts of the specification, as parts are optional. Buyer beware.
  
  
• *WRT
  There are a number of open source distributions for Access Points/Wireless routers. I.e. software that replaces whatever the router manufacturer put on the device. *WRT is my attempt at a common nickname for these distributions. Their lineage may be difficult to follow. This is one angle. Some are more actively developed than others. The concept of stable releases is largely left behind, it appears. The nice thing about these distributions is a common user interface, independent of brand and model. In other words: the user interface stays the same if you go with (for example) DD-WRT on your various wireless routers.
  
  In more and more cases, individual routers do not have an «official» *WRT at all, but a version cooked by an individual developer with the right combination of hardware, interest and skill.
  
  Companies like ASUS, NetGear and Linksys have largely embraced/copied the WRT model, but in a way that largely does not provide any benefit to the *WRTs. They cook their own WRT for their own hardware, but does not provide source for the WiFi-drivers. Hence it is impossible to cook a custom *WRT with a kernel providing other options than what the router manufacturer deemed appropriate. This may limit what functions and features the *WRTs can add to the device. The manufacturers firmware for the router largely works, as the kernel and driver has been explicitly finetuned for the exact hardware model it ships with, and with the options and features the manufacturer consider stable.