802.11 protocol Intensive reading 11: Energy Saving mode (apsd,psmp,smps)

Preface

In the 802.11 major releases, a total of four energy-saving models were defined, and in the previous document we introduced the most basic PSM mode and some details on how it works. Since PSM was designed under the initial 802.11 protocol, it was designed with a more conservative design to ensure the most stable operational requirements. In the later 802.11 new releases, there were other energy-saving modes in order to improve the performance of the energy-saving mode. This paper mainly introduces the main three energy-saving modes: APSD,PSMP and SMPs.

PS: In fact, in some 802.11 of the branch version also defined some of the specific energy-saving mode, such as 802.11v wnm-sleep mode, which we do not expand.
apsd (Automatic Power Save Delivery)

The ASPD mode is a direct improvement to the PSM model. The biggest disadvantage of PSM mode is the "ping-pong" mechanism used when requesting data. That is, a polling frame request, a data frame feedback, such a process of alternating back and forth. In the case of more data cache, this design can cause excessive polling frame appearance and reduce the actual network transmission efficiency. Thus ASPD needs to be improved.

• apsd (Automatic Power Save Delivery): ASPD is designed based on 802.11e. Set the concept of a service Period (SP) in 802.11e. We can simply understand that if a node occupies a channel, it formulates a service time, and in that time, the node can transmit multiple frames, in other words, a competition multiple transmission. In ASPD, there are two kinds of specific working modes in S-APSD and U-APSD. (PS: This article on service period and EDCA of the specific working mechanism will not be expanded, its content will be placed in the finishing of 802.11e)

We use, first of all, to describe the basic idea of APSD:

In, we assume that there is a qos-ap, there is a qos-sta ( PS: Since ASPD is based on 802.11e, Qos-ap and Qos-sta are all 802.11e-capable devices .）。 Assume that a transfer process is initiated by Qos-sta, that is, a "Trigger frame" frame is sent first, which is likely to be a QoS data or a QoS null frame, through which the node requests a service time, that is, service Period Start ( PS: This service time can be understood as the txop time in 802.11e, rather than the service time used in the network model in general sense, in the service time, can pass multiple network frames), then Qos-ap will feedback an ACK and a qos-data frame. Next in the service time range, Qos-sta will continue to exchange data with QOS-AP. When the QOS-AP cache is completely sent, the QoS control field in the last Qos-data frame is labeled EOSP (end of service Period), thus ending a service time, that is, services Period end. In contrast to the ps-poll mentioned in our preliminary discussion of the energy-saving model, this "ping-pong" pattern, the former one requests a feedback, while in APSD, one request can exchange multiple frames. Below we explain S-APSD and U-APSD.

• S-APSD (Scheduled apsd): Scheduling-based APSD,S-APSD can be used in both the EDCA (DCF extension) and the Hcca (PCF extension) modes, from the author's personal point of view, S-APSD is more like the extension in the PSM-PCF mode we mentioned earlier. Because the specific working mechanism of S-APSD is not very good to grasp the package to confirm, so can only be understood by the literal meaning of the agreement, if there are errors, please forgive me.
  

As shown, it is a working mode of S-APSD. In S-APSD, the SP (service Period) is pre-scheduled, and Qos-ap sends a triger frame when the scheduling time is coming, and Qos-sta wakes up to accept the frame in advance, opening a service time. The service time is dispatched through the Schedule element field in the management frame, so the node can know in advance when it should wake up. In, Qos-ap sends the Triger is Ps-poll, and then began to exchange data with Qos-sta, that is, after the STA feedback data to the AP, the AP again to the STA feedback data, so alternately switch, until the end of SP time.

Here, Qos-ap and Qos-sta Exchange data or the upper and lower lines are sequentially switched, that is, there are some similar "ping-pong" mechanism, in the following we mentioned in the PSMP in fact, this will be improved again.

• u-apsd (unscheduled apsd): APSD,U-APSD based on non-scheduling can only be used in EDCA mode, u-aspd more like the PSM-DCF pattern extension mentioned earlier.
  

For example, it is U-APSD's working sequence. In this mode, Qos-sta will first send a triger frame (that is, qos-null) to the AP,AP receive the frame, it means a period of time to start the service, and then the AP and the node will exchange data successively. (PS: As for the service time, the next row of data alternately sent in the process of whether there will be Bakcoff, the author currently read the information is not, but for the original description of the agreement, the author has not determined )

In fact, in an SP time working mechanism, S-APSD and U-APSD is the basic mechanism, the difference is mainly in the S-APSD SP time is the implementation of good scheduling, such as the previous time is the use of beacon frame in the implementation of scheduling, and in the U-APSD is the node initiated, That is, the scheduling time can be initiated at any time, but also the main to, the general S-APSD is the AP to initiate, and U-APSD is the node to initiate more, but this is not absolute.

PSMP (power-save multi-poll)

In fact, introduced from the 802.11e Txop,block ack,802.11n introduced in the rifs, frame aggregation, 802.11AC introduced Mu-mimo (PS: The protocol is only the AP downlink can use Mu-mimo, upstream is not included), These techniques will result in an AP's burst (burst) downstream transmission efficiency higher than the upstream access efficiency.

Therefore, from the point of view of energy-saving mode, in the APSD, the first node, then sub-line transmission mode, there is no first sub-line, and then sub-node transmission efficiency is high, then the latter is the basic idea of PSMP design.

• PSMP (power-save multi-poll) : The new mode in 802.11n can be understood to introduce a more stringent scheduling mechanism, which first uses a PSMP frame to dispatch the entire energy-saving mode transmission period, dividing it into PSMP downlink transmission window, PSMP-DTT (PSMP downlink transmission Time) and PSMP Uplink transmission window, Psmp-utt (PSMP uplink transmission times). During the PSMP-DTT process, the AP bursts, where the inter-frame interval may be smaller, such as RIFs. RIFs time will be smaller than SIFs, SIFs time is to include a send state to accept the switching time, and RIFs is only two times between the sending State of the switch. In this process, the node can not use CCA, directly in the scheduled time to wake up and accept data, so that more sleep time can be obtained, thereby saving energy. In Psmp-utt time, the node will only wake up in the scheduled cycle, and feedback the uplink data frame, thus also reduce the more listening process, saving energy. In PSMP, if a broadcast/multicast packet is present in the downstream transmission of the AP, the first data frame is sent in the PSMP-DTT time, which is actually the same as the previous PSM mode transfer multicast or broadcast frame.
  

As described in a psmp scheduling (refer to "Cambridge.next Generation Wireless lans.802.11n")

Then the AP first sends a PSMP frame that contains the specific scheduling schedule for the downstream link and uplink in the following time. Then start to enter the PSMP-DTT time, the first is to send the local cache multicast/broadcast frame (that is, the psmp-dtt1 moment), and then sequentially send the node corresponding to the buffer broadcast frame, here PSMP-DTT time, the inter-frame interval can be used rifs, in order to achieve higher efficiency. For the ACK part, because the author did not do the detailed textual research, so did not describe. When the PSMP-DTT time is over, the node sends a data frame, such as STA1, to send the upstream data in the PSMP-UTT1, in turn, within its corresponding upstream dispatch cycle. When the Psmp-utt time is over, this round of PSMP data exchange is complete.

PS: Some psmp details in the 9th chapter of "Cambridge.next Generation Wireless lans.802.11n", described in more detail, are interested to expand reading.

SMPS (Spatial multiplexing Power Save)

Finally, we describe the SMPs mode described in 802.11n, which is the power-saving mode in multi-antenna mode. In this mode, the node can choose about the extra antenna, so as to achieve energy-saving purposes. In practice, however, the node is not good at abruptly shutting down its antenna, as it cannot determine whether the AP is sending data to it in the form of a multi-spatial stream or as a single-space stream. If the AP uses multiple streams to send downstream data, and the node is only a single antenna mode of operation, then it is not properly received (here and send diversity, if the transmit diversity is also required to send the AP in the diversity mode, rather than the form of space-to-copy transmission, its pre-coding strategy is different). So the node should communicate with the AP about turning off the antenna, and then the mechanism corresponds to the use of the SMPs.

• SMPS (Spatial multiplexing Power Save): In this mode, the node will turn off the extra antennas and work with just one antenna to achieve energy savings. There are two modes in SMPs, accesses than either static operation mode (static SM Power Save), accesses than either dynamic SM Power save.

Represents a static operating mode of the SMPs, where we can see that the node first needs to send an action frame, switch the AP to a single stream to send the downstream frame, so that the node saves some energy. If you need to revert back to the multi-stream operation mode, you need to feedback an action frame that explicitly indicates the static mode disabled, so that the next frame the AP sends will be the multi-stream mode.

Is the SMPs in the dynamic mode of operation, where the AP sends a dynamic mode action frame with a single-stream mode (the downstream frame in the figure is RTS, which can actually be a data frame), followed by the downstream frame of the feedback. When the frame feedback, the AP sends the next frame, it will return to the default sent multi-stream mode, active from the single-stream mode, and in the static mode, this switch requires a display action frame, so here is a static and dynamic SMPs a difference.

PS: Some smps details in the 10th chapter of the cwap:certified Wireless Analysis Professional official Study guide, described in more detail, are interested in extending the reading.

802.11 protocol Intensive reading 11: Energy Saving mode (apsd,psmp,smps)