Design Article
Meeting the requirements for video-grade Wi-Fi Access Points
Nir Shapira, Celeno Communications
9/4/2011 8:53 AM EDT
Link Budget enhancements – Channel Aware Beamforming and Antenna Diversity
Using channel awareness, a multi-antenna AP can significantly enhance link budget by utilizing digital beamforming techniques. With beamforming, the transmitted signal is shaped such that it optimally matches the instantaneous channel condition, and the emitted energy is focused on the target client device.
With OFDM modulation, each OFDM tone, from each transmit antenna, undergoes a different phase shift in the wireless channel, according to the instantaneous channel conditions. With a typical channel-blind transmission scheme, as is still used by most 802.11n APs, the signals coming from the multiple AP antennas sum at the client device location in random phases can lead to a complete signal nulling.
With coherent beamforming, using channel awareness, the phase of each transmitted OFDM tone from each AP antenna is set to the exact opposite of the phase of the corresponding estimated channel coefficient of the same tone. This results in a coherent addition of voltage at the client device location, as opposed to a statistical addition of power, as in the case of a non-beamformed transmission.
Channel aware beamforming is, in fact, the theoretically optimal transmission scheme in a MIMO channel and it provides for optimal antenna diversity. With transmit beamforming, not only the signal phase is matched to the channel condition, but also the signal amplitude, such that more transmit energy is allocated to antennas that have better link budget, so energy is not wasted on antennas that are in a momentary null state.
Assuming NTX transmit antennas, each transmitting at an equal voltage A (with the actual transmit power from each antenna is PTX=A2), and with L being the equal attenuation from each AP antenna to the client device location, the overall energy that reaches the client is NTXA2L2 in the case of a non-beamformed transmission, and (NTXA)2L2 in the case of beamformed transmission. The overall energy is therefore NTX times larger in the case of a beamformed transmission, compared with even an advanced non-beamformed transmission scheme like STBC.
An AP having three transmit chains can triple the energy reaching the client device by using digital beamforming, regardless of client location, without using hardwired directional antennas. This huge gain of 4.7 dBs (10log(3)), is achieved with close to zero additional cost, and is effective with any number of client device antennas, in particular with a small number of only one or two antennas as is typical in most portable devices. These 4.7dBs can account for no less than a 35% range increase in a typical indoor propagation scenario.
The 802.11n standard supports the procedure of explicit beamforming. Here the client device returns channel state matrix information to the AP as a response to a sounding frame that was sent by the AP from which the transmit beamforming vectors can be extracted. As an added bonus, bundled together with the feedback of the channel matrices, the client device returns an SNR metric that applies for the case of beamforming. This SNR metric can be used to track channel conditions for channel-aware rate selection, as will be detailed in the next section.
In cases where mobile devices do not support beamforming feedback, which is an advanced optional mode in 802.11n (though it is expected to be endorsed by the Wi-Fi Alliance as part of the new certification program for the next generation 802.11ac standard that will supersede 802.11n), a video-grade AP can still use beamforming. This can be accomplished by implementing a one-sided client agnostic implicit transmit beamforming. Instead of getting the feedback from the client, the AP estimates the downlink channel from measurements made on uplink frame receptions. This does not require the portable device to implement explicit beamforming feedback, and achieves a similar gain as in explicit beamforming.
The following figure shows measurement results comparing an explicit and implicit beamforming operation to the same device. Also shown are reference measurements in non-beamforming mode.
Figure 2 Explicit and Implicit Beamforming Performance (Click on figure to download larger image)
The principle behind implicit beamforming is the assumption of wireless channel reciprocity, i.e., the fact that the downlink and uplink channel are identical. Therefore an estimate of the uplink channel (from uplink frame receptions) can be used to design beamforming signals for the downlink. This assumption is true for the wireless channel itself, but the assumption is not true for the RF front end of the AP, since the RF receive path gain and phase is different from that of the RF transmit path. In order to perform implicit beamforming, the difference needs to be compensated by estimating or calibrating the RF front end. 802.11n has defined a complex calibration scheme to enable implicit beamforming operation.
In this scheme, a client device helps the AP to estimate the RF front-end difference. It turns out that with modern RF front-ends, this same calibration can be done without the help of the client device, in a simple one-sided manner by the AP, usually by simple measurements done at production time. With one-sided calibration, implicit beamforming is completely client agnostic, and can be used effectively with any portable client device.
Another channel-aware technique for increasing link budget in a fading channel is using transmit antenna selection. In rich multipath, the RSSI (Receive Signal Strength Indicator) span between different antenna elements can reach up to 10dBs. Therefore, all antennas are rarely equally effective. With antenna switching the AP has more transmit antennas than actual transmit RF chains, and uses instantaneous channel knowledge to select the best antennas for each transmission.
Obviously the optimal antenna set is unique for each serviced client device, and an efficient antenna switching technique needs to be employed. A channel-aware AP can periodically estimate the best antennas to be used with each client device. The optimal antenna set changes dynamically with channel conditions. The AP switches the antennas in use upon each transmission according to the end-point client. This transmit antenna selection scheme does not require any cooperation from the client device, and usually the best antenna sets can be selected using simple metrics such as RSSI. Using transmit antenna selection is a low- cost solution for increasing diversity, since it reduces the dimensions of the actual RF front end in use.
A cost-effective antenna selection scheme enables each RF chain to select one out of two antennas. In this way a simple low cost DPDT (Double Pole Double Throw) switch can be used to select between each antenna pair.
To get an idea of the expected performance gain, let’s assume each antenna has a 20% independent chance of being in fade. It can be shown that the probability for two out of four antennas being in fade in a 4x4 device (having four RF chains) that does not employ antenna switching is 15%. This effectively reduces this system to a two antenna device. In comparison, a 3x3 device (having three RF chains) that uses a simple antenna selection scheme that can select the best three out of six antennas has only an 11% chance of having one antenna in fade and being reduced to an effective two antenna device. We see that by using smart antenna switching, a 3x3 system can outperform a 4x4 system.
Using channel awareness, a multi-antenna AP can significantly enhance link budget by utilizing digital beamforming techniques. With beamforming, the transmitted signal is shaped such that it optimally matches the instantaneous channel condition, and the emitted energy is focused on the target client device.
With OFDM modulation, each OFDM tone, from each transmit antenna, undergoes a different phase shift in the wireless channel, according to the instantaneous channel conditions. With a typical channel-blind transmission scheme, as is still used by most 802.11n APs, the signals coming from the multiple AP antennas sum at the client device location in random phases can lead to a complete signal nulling.
With coherent beamforming, using channel awareness, the phase of each transmitted OFDM tone from each AP antenna is set to the exact opposite of the phase of the corresponding estimated channel coefficient of the same tone. This results in a coherent addition of voltage at the client device location, as opposed to a statistical addition of power, as in the case of a non-beamformed transmission.
Channel aware beamforming is, in fact, the theoretically optimal transmission scheme in a MIMO channel and it provides for optimal antenna diversity. With transmit beamforming, not only the signal phase is matched to the channel condition, but also the signal amplitude, such that more transmit energy is allocated to antennas that have better link budget, so energy is not wasted on antennas that are in a momentary null state.
Assuming NTX transmit antennas, each transmitting at an equal voltage A (with the actual transmit power from each antenna is PTX=A2), and with L being the equal attenuation from each AP antenna to the client device location, the overall energy that reaches the client is NTXA2L2 in the case of a non-beamformed transmission, and (NTXA)2L2 in the case of beamformed transmission. The overall energy is therefore NTX times larger in the case of a beamformed transmission, compared with even an advanced non-beamformed transmission scheme like STBC.
An AP having three transmit chains can triple the energy reaching the client device by using digital beamforming, regardless of client location, without using hardwired directional antennas. This huge gain of 4.7 dBs (10log(3)), is achieved with close to zero additional cost, and is effective with any number of client device antennas, in particular with a small number of only one or two antennas as is typical in most portable devices. These 4.7dBs can account for no less than a 35% range increase in a typical indoor propagation scenario.
The 802.11n standard supports the procedure of explicit beamforming. Here the client device returns channel state matrix information to the AP as a response to a sounding frame that was sent by the AP from which the transmit beamforming vectors can be extracted. As an added bonus, bundled together with the feedback of the channel matrices, the client device returns an SNR metric that applies for the case of beamforming. This SNR metric can be used to track channel conditions for channel-aware rate selection, as will be detailed in the next section.
In cases where mobile devices do not support beamforming feedback, which is an advanced optional mode in 802.11n (though it is expected to be endorsed by the Wi-Fi Alliance as part of the new certification program for the next generation 802.11ac standard that will supersede 802.11n), a video-grade AP can still use beamforming. This can be accomplished by implementing a one-sided client agnostic implicit transmit beamforming. Instead of getting the feedback from the client, the AP estimates the downlink channel from measurements made on uplink frame receptions. This does not require the portable device to implement explicit beamforming feedback, and achieves a similar gain as in explicit beamforming.
The following figure shows measurement results comparing an explicit and implicit beamforming operation to the same device. Also shown are reference measurements in non-beamforming mode.
Figure 2 Explicit and Implicit Beamforming Performance (Click on figure to download larger image)
The principle behind implicit beamforming is the assumption of wireless channel reciprocity, i.e., the fact that the downlink and uplink channel are identical. Therefore an estimate of the uplink channel (from uplink frame receptions) can be used to design beamforming signals for the downlink. This assumption is true for the wireless channel itself, but the assumption is not true for the RF front end of the AP, since the RF receive path gain and phase is different from that of the RF transmit path. In order to perform implicit beamforming, the difference needs to be compensated by estimating or calibrating the RF front end. 802.11n has defined a complex calibration scheme to enable implicit beamforming operation.
In this scheme, a client device helps the AP to estimate the RF front-end difference. It turns out that with modern RF front-ends, this same calibration can be done without the help of the client device, in a simple one-sided manner by the AP, usually by simple measurements done at production time. With one-sided calibration, implicit beamforming is completely client agnostic, and can be used effectively with any portable client device.
Another channel-aware technique for increasing link budget in a fading channel is using transmit antenna selection. In rich multipath, the RSSI (Receive Signal Strength Indicator) span between different antenna elements can reach up to 10dBs. Therefore, all antennas are rarely equally effective. With antenna switching the AP has more transmit antennas than actual transmit RF chains, and uses instantaneous channel knowledge to select the best antennas for each transmission.
Obviously the optimal antenna set is unique for each serviced client device, and an efficient antenna switching technique needs to be employed. A channel-aware AP can periodically estimate the best antennas to be used with each client device. The optimal antenna set changes dynamically with channel conditions. The AP switches the antennas in use upon each transmission according to the end-point client. This transmit antenna selection scheme does not require any cooperation from the client device, and usually the best antenna sets can be selected using simple metrics such as RSSI. Using transmit antenna selection is a low- cost solution for increasing diversity, since it reduces the dimensions of the actual RF front end in use.
A cost-effective antenna selection scheme enables each RF chain to select one out of two antennas. In this way a simple low cost DPDT (Double Pole Double Throw) switch can be used to select between each antenna pair.
To get an idea of the expected performance gain, let’s assume each antenna has a 20% independent chance of being in fade. It can be shown that the probability for two out of four antennas being in fade in a 4x4 device (having four RF chains) that does not employ antenna switching is 15%. This effectively reduces this system to a two antenna device. In comparison, a 3x3 device (having three RF chains) that uses a simple antenna selection scheme that can select the best three out of six antennas has only an 11% chance of having one antenna in fade and being reduced to an effective two antenna device. We see that by using smart antenna switching, a 3x3 system can outperform a 4x4 system.
|
|
|
|
|
Navigate to related information
Frank Eory
9/7/2011 2:40 PM EDT
Excellent article. Thanks for explaining the various ways in which transmit parameter optimization and channel awareness can improve multi-HD stream delivery using the existing 802.11n air interface.
Sign in to Reply
WKetel
9/7/2011 3:22 PM EDT
This article does do a good job in explaining how it is possible to send quality video through a wi-Fi linkup. The process is quite demanding of resources, it appears. Also, it costs a bit.
One thing that was never mentioned is the degree of signal security, which evidently there is no privacy here on WIFI. So the big question is why give up any thought of privacy, and possibly some signal quality, and spend more money, just to avoid needing to route some coaxial cable? Are that many folks really that lazy, or incompetent, that they are not able to run a bit of cable?
Sign in to Reply
kinnar
9/9/2011 3:01 PM EDT
Too good explanation of the cutting edge technology, very great work done by the author.
Sign in to Reply