Given: Road warriors hoard battery reserves by using power adapters whenever possible. If you don’t believe me, just try to find an open power outlet at an airport. Most road warriors are aware of notebook power-saving features such as the display and hard drive timers. Something that’s not well known is the power-saving features available on Wi-Fi network adapters.
Great, saving power is always a good thing. So why then do equipment vendors disable the feature by default? In order to resolve this mystery, we need to look at the inner-workings of 802.11 power-saving technology.
Physical power states
It’s normal to think that a Wi-Fi client has only two power modes: on and off. Sorry, it’s more complicated than that. The following chart from the Wireless Net DesignLine article, “How to use optional wireless power-save protocols to dramatically reduce power consumption,” does an excellent job of describing the different power-reduction states used by a Wi-Fi client:
Off: The only power consumption is leakage current, but coming out of the off state can take a long time (many milliseconds).
Sleep/Standby: The device may be consuming as little as 175 uW and can wake quickly unless the main crystal is turned off.
Listen: The device is listening for a packet to arrive, so most of the radio receiver must be on. State-of-the-art power numbers for a WLAN device in this mode is 110 mW.
Active Rx: Similar to the listen state, but use of additional circuitry may push power consumption for WLAN (802.11g) devices to 140 mW.
Active Tx: When transmitting, the device’s active components include the RF power amplifier, which often dominates in high-power transmit systems. State-of-the-art power consumption for an 802.11g WLAN device is 450 mW at 15 dBm Tx power.
Now that we understand the five different power modes, let’s look at the legacy power-saving method that’s part of the 802.11 standard.
Power-Save Polling Mode
The axiom “timing is everything” is very relevant in Wi-Fi, especially since 802.11 technology uses CSMA/CA. Timing is also important when using power-save polling mode (PSPM), where the Wi-Fi client goes into Sleep/Standby mode for a predetermined length of time in order to save power. Makes sense, but what happens to the traffic destined for the sleeping Wi-Fi client? PSPM has that covered. The access point acts as a storage buffer for the Wi-Fi client whenever the Wi-Fi client is in reduced power mode. Here are the details:
The Wi-Fi client announces its intention to enter PSPM by transmitting an 802.11 frame with the power-save bit in the packet header toggled on. This frame informs the access point that the Wi-Fi client wants to power down everything except a simple timer. The timer’s main purpose is to control when the Wi-Fi client powers back up to receive traffic from the access point.
Upon receiving the PSPM frame, the access point and Wi-Fi client synchronize their timers and agree upon a beacon interval, which is the allocated time between beacon frames.
The PSPM frame also informs the access point of the need to begin saving any traffic earmarked for that Wi-Fi client. With everything understood the Wi-Fi client powers down and the access point starts buffering traffic destined for the sleeping Wi-Fi client.
At the end of a beacon interval, the Wi-Fi client powers up opening what’s called the Announcement Traffic Indication Message (ATIM) window. The access point then takes advantage of the ATIM window to send a beacon frame.
Because the Wi-Fi client is in PSPM, the access point uses a special beacon frame called Delivery Traffic Indication Message (DTIM). The DTIM beacon frame synchronizes the timers and notifies the Wi-Fi client on whether it has any buffered unicast traffic or not.
Right after notifying the Wi-Fi client via the DTIM beacon frame, the access point will forward any multicast or broadcast traffic that it was holding for the client. Then if no unicast traffic exists, the client immediately goes back into powered-down state.
Transferring unicast traffic
If there’s buffered unicast traffic, the access point will transmit the stored data using one of two methods:
PSP method: The Wi-Fi client will send a Power Save Poll (PSP) message to the access point, telling it to release a set amount of traffic. If the total amount of traffic stored by the access point exceeds the amount requested by the original PSP message, additional PSP handshakes take place until all the data is transferred to the Wi-Fi client. At that time, the access point indicates there’s no more data and the Wi-Fi client powers down.
CAM method: The Wi-Fi client advises the access point that it will enter Continually Aware Mode (CAM) for a certain period of time and is capable of receiving data traffic at the discretion of the access point. This approach eliminates overhead created by having multiple PSP handshakes, but at the cost of using more power.
So what’s wrong
It’s pretty easy to see that PSPM is going to add overhead to Wi-Fi connections — which isn’t a good thing, because Wi-Fi has more than enough management overhead already. The first problem is latency; the Wi-Fi client has to wait the entire DTIM beacon interval to receive traffic. Typically, DTIM beacon intervals are 300 ms, which may not seem like much but when considering Wi-Fi Quality of Service (QoS) it’s significant.
The second problem is deteriorated data throughput. PSPM requires the Wi-Fi client to move in and out of Sleep/Standby mode. That along with the exchange of PSP messages uses precious bandwidth that should be allocated to data traffic.
Interesting test results
I’ve just read the Network World article by C. J. Mathias (Farpoint Group) named “Wireless computing power saving measures may not be worth the effort,” and the test results are quite interesting. The premise of testing was:
“Do Wi-Fi power-conservation techniques, when enabled, actually save a meaningful amount of energy or have any negative impact on throughput?”
Please check out the chart in the article; you’ll see there’s very little elapsed battery time gained by using PSPM. That’s not good. Adding insult to injury, Mr. Mathias also makes mention of some downside effects of PSPM:
“PSM is mostly harmless, but can also have very negative performance impacts. We also noted in the testing of some of the power-save modes on the Intel adapter that test runs would not complete, timing out with an error message, indicating that the notebook was simply not responding fast enough to meet application demands. Users thus need to be cautioned about setting PSM options without some knowledge of the possible consequences.”
Why PSPM doesn’t help
Mr. Mathias also takes a moment to introspect as to why PSPM is not of any significant value:
“The 802.11 standard was initially developed during a time when processor clocks were in the 100MHz to 200MHz range, and initial WLAN designs involved a significant number of power-hungry components. Today, however, Wi-Fi adapters are highly integrated — meaning fewer chips are required to implement a Wi-Fi solution — and designs are more power-efficient. While the notebooks’ other components — most notably the processor (because of higher clock rates) and display and backlighting (due to much higher resolutions) — often consume more energy than in the past.
Notebook designers have compensated with larger batteries and a continual emphasis on power-conservative designs and provisions for a high degree of end-user control over power conservation settings in many cases, but the proportion of energy consumed between the computer and the WLAN adapter has clearly flipped.”
Check for enabled PSPM
That’s some very interesting insight. With that in mind, I thought I should at least give one example of where to look for the Wi-Fi power-saving settings. The following image is the Intel PRO/Wireless client application under Advanced/Adapter settings:
In this example, the Wi-Fi client power configuration is set for maximum performance, so PSPM is disabled. As mentioned earlier, this is usually the default condition, but it wouldn’t hurt to check.
Researching power conservation
There’s a great deal of research going on right now with regards to power conservation. It’s especially important when considering 802.11n, because multiple radios are used. Mr. Mathias understands this and makes mention of two new approaches:
Power Save Multi-Poll: This approach is specified as part of 802.11n and was developed because of concerns that MIMO-based products, using multiple radios and more circuitry regardless, would become power hogs with a significant adverse impact on battery life. An extension to U-APSD and S-APSD, the scheduled version reserves a time slot for a given client station and thus temporarily silences others associated. This technique may be better with relatively heavy traffic loads.
Dynamic MIMO Power Save: This technique allows MIMO-based (802.11n) radios to downshift to less-aggressive radio configurations (for example, from 2×2 to 1×1) when traffic loads are light.
We’ve enough information now to understand why equipment vendors disable PSPM by default. If PSPM is to be a viable power-saving technology, researchers will have to improve the process of how a Wi-Fi client determines when to use PSPM.