" The previous article sorted out and introduced the Wi-Fi rate with examples, and compared the difference between the real rate and the declared rate. This article will analyze the factors that actually affect the Wi-Fi transmission rate from a technical point of view. and the corresponding optimization method. ”
The PHY (physical layer) rate is a theoretical limit rate, which may be reached at a certain moment, but it is impossible to maintain this rate continuously, because various factors in actual usage scenarios must be considered, such as the following factors that cause WiFi rate loss, and Technical or usage optimization methods are described.
01
Factors that reduce Wi-Fi speed
Management frame sent
-
The management frame transmission must be sent at the lowest rate: it is necessary to ensure that WiFi devices of all versions of the protocol can receive it. For example, a very early device can also receive and recognize the frame. The minimum rate is:
-
-
2.4GHz band: 1Mbps
-
5GHz band: 6Mbps
-
half duplex
WiFi does not have independent downlink and uplink frequency bands (Ethernet is full-duplex and can receive and send at the same time). All WiFi devices work on a certain channel and must be sent and received in order. In addition, the CSMA/CA (Carrier Sense Multiple Access/Collision Avoidance) mechanism will be used when sending. Before any device sends data, it needs to spend time listening to whether the current spectrum is available.
IFS frame spacing
Before each WiFi transmission, there needs to be a free gap in the spectrum. WiFi has many different IFSs. The purpose of IFS is to regulate the session flow and provide priority for certain types of transmissions.
PHY Signaling
Wireless communications must be synchronized for reliable frame reception. PHY Preamble (header, a string of bit streams) is used for this function. The PLCP Header after Preamble passes the attributes of the next frame to the receiver so that the receiver knows how to process the data. For example, for voice transmission between people, Preamble and PLCP are like the speaker saying: "Hello, I am going to say something. I will finish speaking in 20 seconds, and I will speak very quickly, so please listen alright".
MAC Header
The MAC layer Header is used to negotiate some supported (or unsupported) features and functions. This part is not the data that the user application wants to transmit, so it is part of the transmission overhead. In a data frame, the MAC header is so small (often transmitted at very high data rates) that it barely counts as overhead. In other frames such as 802.11 Beacon frames, the overhead ratio is relatively large.
Guard Interval
Between each 802.11 symbol (modulation), there must be a quiet period called Guard Interval (GI), which allows the previous symbol to "stabilize" on the wireless medium. Without Guard Interval, a symbol may interfere with previous symbols (Inter-synbol Interference). The normal guard interval setting of Wi-Fi is 800ns, and 400ns will also be seen. In 802.11ax, there will be 1600ns and 3200ns, which are related to the coding efficiency during modulation.
Acknowledgment
Wireless communication is inherently unreliable and lossy, and many types of frame transmissions require acknowledgments. Acknowledgments are overhead by themselves (no useful data is transmitted), and they also require an additional interframe space (i.e. SIFS).
Fragmentation
Frames with small payloads are not overhead per se, the overhead lies in its inefficient use of the wireless medium. The design of Frame Fragmentation (frame fragmentation) is allowed in the alliance specification. The purpose is to make the frame smaller in order to try to reduce collisions in a noisy environment (smaller frames are less likely to be disturbed). Frame Fragmentation is currently rarely used, because the harm it adds to overhead outweighs the benefits it brings.
Protection Mechanism (protection frame mechanism)
Guard frames are required between incompatible PHY versions (such as 802.11b and 802.11g) to achieve coexistence. Guard frames are usually implemented via RTS/CTS exchange or CTS-to-self frames before data transmission. Other types of guard frames (for 802.11n) can also be used. In addition to using these frame exchanges for protection, they may also be used at other times (such as when hidden terminals are present) to try to improve the health of the overall network. These protection frames themselves do not transmit effective user data, and excessive use will also cause the user's effective rate to decrease.
Random Backoff
In order to avoid interference, WiFi needs to allow itself to have the opportunity to transmit data. WIFI has designed a random Backoff mechanism to use the spectrum of the channel. It needs to wait for the WIFI channel to meet the random Backoff time and remain idle before trying to send data. . At the same time, the mechanism can also help to achieve the QoS prioritization mechanism. The Random Backoff time represents n "slots", and the wireless channel must be idle during this Backoff time.
Retransmission
Retransmissions may be required when a sent frame is not correctly received (or not acknowledged) by the intended recipient. Transmitting the same frame multiple times is an obvious source of overhead. Other sources of overhead (such as Random Backoff, IFS, etc.) are also repeated when frames are queued for retransmission.
Interference
Interference can come from 802.11 and non-802.11 radio frequency interference. WiFi uses the CCA (Clear Channel Assessment) mechanism to monitor whether the current channel is "busy" or "idle". When it detects that the interference of the current channel is greater than a certain threshold, it will wait without sending data.
terminal hidden
The WiFi terminal is hidden, as shown in the figure below
When the problem of WiFi terminal hiding occurs, both A and B in the above figure send data to the AP. Even though RTS/CTS protection frames are used, A and B cannot hear each other because of the distance, so it will not work. Finally, A and B B will still transmit data at the same time, resulting in spectrum conflict, which will eventually cause the loss and retransmission of A and B transmissions.
02
Methods of Wi-Fi Rate Optimization
Interference (interference problem) solution
Deal with it from the perspective of interference cancellation.
-
Non-802.11 interference: You can only find out the source of interference and turn it off, such as equipment such as microwave ovens
-
WLAN interference: Minimize the conflict of wireless media with other Wi-Fi devices as far as possible, and reduce interference by adjusting AP position and channel, setting transmission power, and turning off unnecessary redundant Wi-Fi devices.
control necessary functions
Most of the overhead cannot be eliminated, such as IFS, Random Backoff, PHY signaling and MAC Header.
But there are ways to reduce the impact of these necessary functions:
-
An extra-long interframe space (EIFS) is generated when a STA receives a frame with a CRC error. Minimizing interference can make EIFS shorter.
-
In addition to using a low PHY rate on the MAC Header and effective Payload, 802.11b also uses a relatively long PHY Signaling (PHY Preamble and Header) in order to be compatible with 802.11b, and its rate can only be maintained at the same rate as 802.11b at a compatible lower rate. The feasible solution here is to turn off 802.11b (usually it can be turned off in the router setting interface).
-
An easy way to reduce MAC (and PHY) overhead is to eliminate unnecessary Beacon traffic. Use dynamic user policy assignment instead of separate ssids to separate WLAN services.
Reduce unnecessary broadcast and multicast applications
Wi-Fi BroadCast, MultiCast (broadcast, multicast) are transmitted in low rate mode (forwarded by routers), and will not retransmit when interference is lost. A large number of such packets in the LAN will affect the overall Wi-Fi air rate. If there is no need for actual use, you can limit the use of unnecessary broadcast and multicast applications. The more common such applications are as follows:
-
Apple: Hello
-
Cisco: CDP(Cisco Discovery Protocol)
-
SSDP: Simple Service Discovery Protocol
Short Guard Interval
The short guard interval of 802.11n is 400ns, which can increase the speed by 10%, but it needs to be avoided in environments with high reflectivity (warehousing, manufacturing, industrial environments, etc.).
Frame Aggregation and Block Acknowledgment (frame aggregation and block confirmation technology)
802.11n takes better advantage of frame aggregation and block acknowledgment. While early WLANs shrunk (fragmented) frames to avoid collisions, the higher physical rates of modern networks allow for larger wireless frames, which greatly improves efficiency. Encapsulating more upper-layer data in each frame is a classic example of reducing overhead. If each aggregated frame was transmitted independently, we would see much higher overhead from IFS, Backoff, PHY signaling and MAC header. When frame aggregation is used, block acknowledgment is also used at the same time. Improved efficiency by using the ack bitmap to represent a unified acknowledgment of successful reception of multiple frames, rather than sending a separate ack for each received frame. Enabling frame aggregation will yield significant rate improvements in most cases. If there is a configuration option, choose A-MPDU instead of A-MSDU.
Protection Mechanism (timely use of protection mechanisms)
Previous technologies require more time to process PHY signals. In addition to their own slow processing speed, traditional WiFi terminals also require more efficient terminals to protect their own data transmission, thus hindering the further improvement of the overall rate. By analyzing the format of protection frames (such as RTS and CTS), it will be found that they are actually very small at the MAC layer. But if you look at the format of the PHY layer, you will see that the actual RTS/CTS exchange takes a long time. This is because of the consumption of the PHY Preamble and PLCP headers. At the same time, one or two SIFSs must be considered, which is why the protection mechanism is enabled. disadvantages. Then the solution is to get rid of the old (especially 802.11b and earlier) WiFi terminals if the business scenario allows. Technology takes priority for faster transfer efficiency. However, this processing method still has disadvantages, and it cannot be said that the protection function should be discarded at any time. For example, there is a lot of interference in the space itself, which has a significant impact on performance, so enable RTS/CTS or CTS-to - self is also effective.
Retrie (retry) as low as possible
In the presence of interference, retries are typically a major source of overhead for the network. Retries are usually caused by reception errors caused by interference, but there are many other reasons as well. For retransmissions, the biggest drawback is that the first (failed) attempt has already taken some time, and the second attempt will consume a longer backoff period, and retransmissions will often cause a rate shift at the same time (from a higher data rate switch to a lower data rate) for improved reliability. After actually deploying the network and verifying performance in the space, the baseline for retrying should be determined. The goal of retry is generally less than 10%, but it is also determined according to the application scenario, such as which business data must be reachable to be retried, which are allowed to be discarded, and the retry rate is determined on the premise of ensuring user experience . After the network is deployed, and the retry target is determined, it is necessary to reduce the retry by improving the signal-to-noise ratio and reducing interference to reduce it to the expected target. The specific way may be to properly adjust the location of the AP, channel settings, antenna and power output settings, and so on.
Data Rate (data rate) try to use the high rate mode
Data rate support is a hot topic in WLAN design. 802.11b was discussed earlier, which will make the network work at a low rate. If you must keep 802.11b WiFi terminals, you can consider disabling the support for 1 Mbps and 2 Mbps. Disabling 1 Mbps and 2 Mbps is a very common practice when the BSS has to use lower data rates that consume a lot of airtime because all terminals have to "receive" these slow frames. If you don't support 802.11b, you can even disable support for 6 Mbps (or 12 Mbps), making 12 Mbps (or 18 Mbps) the lowest rate. Of course, this will be done to the extreme in applications with high-speed requirements, but lower rates are still conducive to improving transmission reliability, so the scenarios are different, and these methods need to be considered comprehensively before use.
For more exciting content, please search and follow the WeChat official account of "Technology Shapes the Future".