802.11ah - Overview
802.11ah - Overview
802.11ah - Overview
11ah
What's new?
Is intended to support extended
range Wi-Fi, and the IoT
consumption of nodes/stations
Don't need to maintain backward
compactibility
IEEE 802.11ah
Use cases
1.Smart sensors and meters
smart grids,
weight),
IEEE 802.11ah
Use cases (cont.)
2. Backhaul aggregation:
802.11ah covers the backhaul connection between 802.15.4g devices and remote servers
(since IEEE 802.15.4g also works in Sub 1GHz band, TGah addresses coexistence with
these devices).
IEEE 802.11ah
PHY / radio interface
802.11ah uses OFDM to provide the modulation scheme for the signal
there are two categories into which the 802.11ah physical layer PHY can be split:
2, 4, 8, and 16 MHz: These modes are based on 20, 40, 80, 160 MHz
bandwidth modes of 802.11ac, being 10 times down-clocked, i.e. symbol
lengths are 10 times of that in 802.11ac. MIMO is also used in these mode (up
to 4 spatial streams).
1MHz channel bandwidth: mainly use for low data rate (150Kbps - 4Mbps for
single spatial stream) & long range (up to 1Km) applications such as M2M,
IoT,... One more Modulation and Coding Scheme - MCS10 - is defined and
used (for 1MHz channel only), it's a mode of MCS0 (see table on next page)
but with a 2x repetition of the data.
There is a variety of Modulation and Coding Scheme - MCS - options available. These
are tabulated on next page:
1MHz
(Mbps)
2MHz
(Mbps)
4MHz
(Mbps)
8MHz
(Mbps)
16MHz
(Mbps)
MCS0
BPSK
1/2
.300
0.65
1.5
3.25
6.5
MCS1
QPSK
1/2
.600
1.3
6.5
13
MCS2
QPSK
3/4
.900
1.95
4.5
9.75
19.5
MCS3
16-QAM
1/2
1.2
2.6
13
26
MCS4
16-QAM
3/4
1.8
3.9
19.5
39
MCS5
64-QAM
2/3
2.4
5.2
12
26
52
MCS6
64-QAM
3/4
2.7
5.85
13.5
29.25
58.5
MCS7
64-QAM
5/6
6.5
15
32.25
65
MCS8
256-QAM
3/4
3.6
7.8
18
39
78
MCS9
256-QAM
5/6
N/A
20
43.33
86.67
MCS10
BPSK
1/2
.150
Modulation
Code Rate
Normal GI (8us)
Short GI (4us)
BPSK
1/2
0.65
0.72
QPSK
1/2
1.3
1.44
QPSK
3/4
1.95
2.17
16-QAM
1/2
2.6
2.89
16-QAM
3/4
3.9
4.33
64-QAM
2/3
5.2
5.78
64-QAM
3/4
5.85
6.5
64-QAM
5/6
6.5
7.22
256-QAM
3/4
7.8
8.67
256-QAM
5/6
N/A *
N/A
* MCS9 is not valid for 802.11ah with a single spatial stream for a 2 MHz channel, the reason is
same with that in 20MHz channel of 802.11ac.
US
902 - 928
Korea
917.5 - 923.5
Europe
863 - 868
China
755 - 787
Japan
916.5 - 927.5
1MHz
2MHz
4MHz
8MHz
16MHz
US
26
13
Korea
Europe
China
32
Japan
11
Robust 900MHz
client device
0dB
-17dB
Tx antenna gain
0dB
-3dB
+8.5dB
+8.5dB
+10dB
+10dB
Flat fading
-4.5dB
-4.5dB
Sub Total
+14dB
-6dB
+3dB
+3dB
2x repetition coding
+3dB
+3dB
Total
+20dB
0dB
IEEE 802.11
MAC Layer
In some cases, the AP and the STAs can use NAV (Network Allocation Vector - be
carried in MAC header) to reserve the medium for some consecutive frames.
RTS/CTS mechanism is one of those cases. The mechanism in that NAV is used also
known as Virtual Carrier Sense mechanism.
There is a Point Coordinator (reside in the AP) that has the right to allocate the medium
to the PCF-based STAs by polling frames.
The Point Coordinator poll the STAs one-by-one to send and receive data, hence there is
not internal contention in a BSS. This period is so called CFP - contention-free Period.
PCF-based STAs, in fact, are the STAs which can reply to the polling frames of the AP.
AP use NAV(sent in beacon) to prevent all of STAs transmit data but which one is polled.
NAV is used to prevent DCF-based STAs (or normal STAs) to access to the
medium.
Due to the priority of PCF over DCF, the CFP is not provided full-time. It alternate with
the period where the standard DCF-based services are provided. In that case, the first
period after Beacon Frames are always CFP. A CFP-end frame will be send to
explicitly terminate the CFP.
The PCF is not widely implemented, it's restricted to infrastructure networks and Wi-Fi
Alliance does not include yet PCF functionality in their interoperability standard.
802.11e-2005 is an approved amendment to the IEEE 802.11 standard that defines a set
of quality of service (QoS) enhancements for wireless LAN applications through
modifications to the MAC layer.
The 802.11e enhances the DCF and the PCF, through a new coordination function: the
hybrid coordination function (HCF).
Within the HCF, there are two methods of channel access, similar to those defined in the
legacy 802.11 MAC:
HCCA - HCF Controlled Channel Access works a lot like PCF and:
EDCA - Enhanced Distributed Channel Access uses classified IFS for different
categories of traffic and TXOP (Transmit Opportunity) to provide contention-free
periods to QSTAs (QoS-STAs STAs in 802.11e) to access the channel in DCF
period.
Both EDCA and HCCA define Traffic Categories (TCs). For example,
Video or Voice over Wireless LAN could be assigned to a high priority class.
HCCA is generally considered the most advanced (and complex) coordination function.
Implementing the HCCA on end stations uses the existing DCF mechanism for channel
access (no change to DCF or EDCA operation is needed). Stations only need to be able to
respond to poll messages!
High-priority traffic has a higher chance of being sent than low-priority traffic because
of it has shorter inter-frame space (IFS) and back-off time.
AP can provide contention-free access period called a Transmit Opportunity (TXOP)
a bounded time interval during which a Q-STA can send multiple frames without
contention (within their BSS).
TXOP is assigned to each Q-STA, specified per TS Traffic Stream in that QSTA.
Each Q-STA may have upto 4 TSs classified as mention above.
A Q-STA can use a TXOP to transmit multiple frames within an access category as
many frames as possible as long as the duration of the transmissions does not
extend beyond the maximum duration specified of a TXOP.
If a frame is too large to be transmitted in a single TXOP, it should be fragmented into
smaller frames and send them in consecutives TXOP.
Beacon
Beacon
Busy
Backo
f
Frame 2
Frame 1
Backo
f
AIFS
AIFS
SIFS
SIFS
Ack
Ack
Polling can be issued in both CFP and CP (in Controlled Access Periods)
HC grants a polled TXOP to one Q-STA, which restricts the duration of the Q-STAs
access to the medium.
UPLINK TXOP
DOWNLINK TXOP
Beacon
Busy
Poll + Data
Ack
Ack
PIFS
SIFS
SIFS
SIFS
Data + Ack
Data
Data
PIFS
SIFS
SIFS
SIFS
Data
Ack
During association, the STA tells the AP that it will use PS-mode, and it's listen
interval (or wake up cycle, this interval may be longer than few beacon
interval). In PS-mode, the STA has to listen periodically Beacon frames.
The AP will buffer incomming frames of the PS-mode STAs, notify them by the
TIM (Traffic Indication Map) associated in every beacon frame whether the
frames for each PS-mode STA.
For Multi-cast or Broad-cast data, AP use TIM to notify whether the frames like
that, and use DTIM (Delivery TIM) for more detail. Every DTIM-interval
(= beacon-interval x n), TIM is periodically replaced by DTIM in beacon frame, AP
also uses TIM to notify STAs when DTIM will be used instead of TIM. After the
beacon frame with DTIM, AP will transmit Multi-cast and Broad-cast data frames
before any individually addressed frames.
If an STA get relevant notice, it should stay awake to retrieve it's data.
In CFP: STAs have to wait for the AP transmit the polling frames (CF-poll) to the
pollable STAs (CPF-based STAs), and then send them their data, one-by-one.
In non-CFP (where DCF is used): STAs transmit a PS-poll frame, the AP either
Two STAs listen TIM and retrieve their uni-cast data in PS-mode (when DCF is used)
Two STAs listen TIM and retrieve their uni-cast data in PS-mode (when PCF is used)
Data
CF-Poll
Data
TIM
CF-Poll
Beacon_
Interval
ACK
PS-poll
STA 2 in
PS mode
ACK
STA 1 in
PS mode
PS-poll
AP
TIM
TIM
Two STAs listen DTIM and retrieve broad-cast data in PS-mode (DCF & PCF)
Beacon_
Interval
TIM
DTIM
Broadcast Data
AP
STA 1 in
PS mode
STA 2 in
PS mode
TIM
Gets beacon, during association, notifies that I shall be in PS-mode, I shall wake-up
every N beacon interval. If the AP accept, STA can switch to sleep mode.
Wakes up prior every N beacon interval (estimate by a local timer), listen to the beacon,
if :
1. Do not heard the beacon: continue to wait for a beacon, sometime the beacons are
delayed due to the busyness of the media.
2. A beacon with DTIM is received, stays awake to receive group data, continues to
awake to get next beacon with TIM, if TIM indicates uni-cast data buffered, send PSpoll to retrieve data, sleeps again when finish or there is not any data.
3. A beacon with TIM is received:
a)Sends PS-poll to retrieve data if TIM indicates uni-cast data buffered.
b)Stays awake to get DTIM (if it is indicated by TIM).
c) Sleeps again when finish or not (a) and (b) aboved.
Note: after retrieves uni-cast data, STA has to stay awake to receive one more beacon to
make sure that the relevant indicator bit is reset.
PS-STAs do not need to listen for beacon (to receive TIM or DTIM).
802.11e MAC
Unscheduled-APSD (cont.)
U-APSD vs legacy power save
802.11e MAC
802.11 MAC
With WNM-Sleep, TIM in Beacon may be as big as 6000/8 = 750bytes (e.g. when
AID1 and AID5998 have the same awake DTIM and buffered frames at the awake
DTIM). It is difficult for an AP to group STAs with nearby AID to wake up at the same
TBTT beacon interval (e.g. AID 101 to AID 200).
A beacon interval can not finish polling all 6000 STAs in one 100ms beacon interval
given the lower PHY rate of a Smart Grid BSS for e.g. (~hundreds Kbps):
(100bytes MPDU + 14bytes ACK) * 8 /200000= 4.56ms.
Actually transmission time is longer since Backoff, PS Poll, PHY overhead and IFS are
needed.
TIM
DTIM
802.11 MAC
The PS-STAs have to get both DTIM and TIM to have buffering information about all
kinds of data (broad-cast, multi-cast and uni-cast).
After retrieves uni-cast data, STA has to receive TIM in next beacon to make sure
that the relevant indicator bit was reset, if next beacon is associated with DTIM, it has
to receive one more beacon.
When there is a large number of PS-STAs in a network, the length of the beacon
frame could become extremely long due to the excessive length of the partial virtual
bitmap in TIM (up to 253 bytes). In addition, if the amount of the buffered traffic is too
heavy to be accommodated within a beacon interval, some PS-STAs inevitably stay
awake in a long time to complete the receptions of their buffered packets.
Maximal value of Idle-time is 18.64 hours, somes time it's not enough, for e.g., some
alarm sensors send a packet once a month only to confirm that it's alive for save its
battery.
In legacy 802.11 standard, TIM's length will be long if the number of associated PSSTAs is large, and beacon trigger the contention of PS-STAs.
The polling mechanism used by PCF of HCF is not effcient with large number of PSSTAs, when most of them have no data to transmit:
it requires sensor stations to be active and to consume energy till they are polled
PSMP can not support to thousands STAs and it can only schedule a transmission
not far than 8 ms from the end of the frame carrying this schedule
Many changes in MAC layer can be introduced due to there is not backward compactible
requirement:
In 802.11, the AP allocates Association IDs - AIDs - to stations that associate with the
AP, the maximum number of IDs that can be allocated in legacy 802.11 is 2007.
In 802.11ah, new AID structure is 13 bits long, leading to a maximum number of
associated stations is 8191 (213-1).
This hierarchical AID is split into four levels: page; block (or group); sub-block; and
station index within the sub-block.
The structure of new AID:
In legacy 802.11 networks, the duration of the doze state (sleeping states) is limited
by Max idle period, which is determin by a 16-bit coefficient, time unit is 1024ms,
thus, maximal value is 18.64 hours.
In 802.11ah, two most-significant bits of the Max idle period field is used as scaling
factor, values 00, 01, 10, 11 represent scaling factors 1, 10, 1000, 10 000
respectively, and so, maximal value of Max idle period is 2500 times greater (upto
more than 5 years).
In legacy 802.11
Max idle period = 16-bit coefficient x 1024ms
Max idle
period value
x x
In 802.11ah
00 scaling by 1
01 scaling by 10
10 scaling by 1000
11 scaling by 10000
Max idle period = 14-bit coefficient x 1024ms x scaling factor
STAs wait a random time (e.g. several Beacon intervals) before attempting
authentication
Each unsuccessful attempt increases window
ACK ID containing some bits from the received frame to reduce probability of
false positive acknowledgment.
Duration field which either has the legacy interpretation or is used by power
saving mechanisms.
More Data field indicating whether the NDP ACK transmitter has data to transmit
to the NDP ACK receiver, etc...
Example: 11ah transmission of 100B frame at lowest rate (1MHz x NSS 1 x MCS10)
takes ~8ms
Legacy ACK: ~1.5ms (20% of the data frame!)
NDP ACK: ~0.5ms (6% of the data frame)
Many other NDP frames were defined, for e.g.: NDP CTS, NDP PS-Poll, NDP BlockAck,
NDP Probe Request, etc...
Sensor STAs or Z-class STAs are the STAs transmit low traffic of short packets, they
are also expected to have limited available power.The number of sensor STAs
associated to an AP may reach 6000.
Offloading stations or H-class stations can transmit intensive traffic. This type includes
laptops, wireless cameras, gadgets offloading data. It is unlikely that more than a
hundred of such devices are associated to an AP.
TIM stations & non-TIM stations: (based on scheduling method to access to media)
TIM STAs: They have to listen to both DTIM and TIM beacons to send or receive data.
Non-TIM STAs: They only have to listen to DTIM beacons to send or receive data.
Unscheduled STAs: These STAs do not need to listen to beacons, they can transmit
data anytime that they wake up.
TIM Segmentation mechanism is used to reduce contention probability of TIMSTAs, in that, each beacon is sent with a part of TIM only, thus the number of STAs
triggered after that beacon is not too large.
Restricted access window (RAW) divides STAs into groups and splits the channel
into slots. Then it assign each group to a slot and STAs can transmit only in their
slots. In 802.11e, it's used to increase throughput, but in 802.11ah, the target is
extend the life-time of PS-sensors (Z-class STAs).
Transmit Opportunity (TXOP) is used to allocate the temporary RAWs to STAs. As
like as RAW, in 802.11ah, this one is used to protect the PS-sensors from
collisions, to reduce the awaken time of STAs.
Target Wake Time (TWT), a new version of APSD, allows the PS-STAs sleep in
long time and wake up on appointment which is made with the AP before.
TIM can be splitted into the slices and be sent in seperated beacons, each slice may
contain infomations for upto 4 blocks (4 groups of STAs) in 4 different pages.
DTIM contain one or several Page Slice information elements (PSIE), unique for each
page period (or DTIM period).
PSIE indicates number of TIM slices, their positions regarding current beacon,
number of blocks in each slice, etc.
PSIE includes a Page Bitmap which indicates the presence of buffered data for
each block in all of slices.
Thus, having received an PSIE (in DTIM beacon), an STA knows whether the AP has
buffered data destined for its block or not,
An example in which the number of STAs is smaller than 2048 (1 page) and only the STAs
#1538 and #1539 have data frames buffered on AP.
When the number of STAs greater than 1 AID page (greater than 2048), two signaling
methods can be used: non-TIM offset or TIM offset:
1. Non-TIM offset: The signaling information of a particular TIM group is transmitted in
the same beacon as many times as the number of network pages (upto 4 pages).
This is the default mode in IEEE 802.11ah.This fact implies that the STAs are forced
to listen the information related to pages that they do not belong to.
TIM offset: In this mode, a 5-bit field in the DTIM beacon that allows each TIM group in
4 different pages to be separately scheduled over their own TIM beacons (up to 32
positions in 4 pages, 8 TIM beacons per page).
The TIM offset mode has lower energy consumption than the non-TIM offset mode.
However, its behavior with respect to the maximum number of stations supported,
packet delivery ratio (PDR), and network efficiency is slightly worse.
The STAs which have buffered data on the AP are spread out into 19 different TIM
groups and in 4 different pages.
TIM offset is used, thus there is only one block (or one TIM group) in each slice, TIM
is segmented into 19 slices
TIM-1.1
RAW
TIM-1.2
RAW
TIM-1.3
RAW
Beacon
RAW-1
TIM
...
RAW-i
...
RAW-n
(1)
where NRAW is the number of slots, Noffset is a parameter to improve fairness and equals
two least significant bytes of the beacon FCS,and x is determined as follows:
If the RAW is restricted to stations with AID bits in the TIM element set to 1, x is the
position index of the station among others. Otherwise, x is the station AID. Note that
rule (1) means that a group of stations may access the channel in a slot.
To access the channel in its RAW slot, a station uses EDCA (new version of DCF from
802.11e), however, backoff functions inside and outside RAW are different, since
contention conditions differ.
TXOP - a concept has been approved in 802.11e, widely used in 802.11n - is used with 2 variants: Bi Directional TXOP BDT and
TXOP Sharing.
TXOP-based Sectorization scheme allows STAs of a BSS and STAs of an overlapping BSS to transmit their data simultaneously.
In TXOP, the AP use virtual carrier sense (NAV or RID) to reserve channel temporarily to transfer few consecutive data frames and
ACK frames between AP and a STA, therefore the sender has to delay a SIFS (16us) instead of DIFS (34us) between frames and
without contention.
Relay without
TXOP sharing
TWT Target Wake Time mechanism is used for very low power STAs, in that, STAs do not need to receive the
beacons. A STA like that can makes appointment with the AP to schedule a series of times where it will wake up,
the AP will reserve a gap of time to wait for it or a small group of STAs which are made appointment in the
same gap of time.
For this purpose, an AP may setup RAW(s) for non-TIM STAs and then indicate to TIM STAs RAW information
during which no TIM STA is allowed to contend. The RAW(s) is used for protecting either TWT(s) scheduled by the
AP or specific interval(s) for non-TIM STAs.
Some modification in MAC layer are used to perform the timer synchronization function (TSF) for these STAs. Direct
Multicast Service DMS may also used to transmit group addressed frames destined to these STAs as individually
addressed frames, therefore, it doesn't need to receive DTIM beacon.
Wake-up
SM
LM
Sleep again
UL
BA
DL
BA
TM
RM
RM
TM
SM
Annex A
Beacon Interval
Timestamp: Timing Synchronization Function (TSF) keeps the timers for all stations - counting in
increments of microseconds with modulus 264 - in the same Basic Service Set (BSS) - are
synchronized. Upon receiving a beacon, a station sets its TSF timer to the timestamp of the
beacon if the value of the timestamp is later than the stations TSF timer. On a commercial level,
the vendors assume the TSF's differences in a BSS will be within 25 microseconds.
Supported rates
Traffic Indication Map (TIM or DTIM Delivery TIM) (if there is any associated PS-STA).
In ad hoc networks, there are no access points. As a result, one of peer stations assumes the
responsibility for sending the beacon. After receiving a beacon frame, each station waits for the
beacon interval and then sends a beacon if no other station does so after a random time delay.
This ensures that at least one station will send a beacon, and the random delay rotates the
responsibility for sending beacons.
Annex B
802.11af: White-Fi
802.11e: QoS it is an approved amendment to the 802.11
802.11s: mesh networking
802.11ax: MIMO-OFDM
802.11u: interworking with external networks
Annex C
802.11e - QoS
802.11e-2005 is an approved amendment to the IEEE 802.11 standard that defines a set of quality
of service (QoS) enhancements for wireless LAN applications through modifications to the MAC
layer.
The 802.11e enhances the DCF and the PCF, through a new coordination function: the hybrid
coordination function (HCF).
Within the HCF, there are two methods of channel access, similar to those defined in the legacy
802.11 MAC:
HCF Controlled Channel Access (HCCA) works a lot like PCF and:
Enhanced Distributed Channel Access (EDCA) uses TXOP (Transmit Opportunity) to provide
contention-free periods to access the channel in DCF period when need.
Both EDCA and HCCA define Traffic Categories (TC). For example,
Video or Voice over Wireless LAN could be assigned to a high priority class.
HCCA is generally considered the most advanced (and complex) coordination function. Implementing
the HCCA on end stations uses the existing DCF mechanism for channel access (no change to DCF or
EDCA operation is needed). Stations only need to be able to respond to poll messages!
Annex D
Block Acknowledgement
Annex E
Resource Allocation
The RA - Resource Allocation frame has been proposed. It contains the scheduling
information of each individual station, through which the station can learn the time
slot during which it is allowed to conduct medium access for uplink or downlink
transmission. RA frame is transmitted at the beginning of each RAW and all the
stations assigned to that RAW have to wake up to receive it.
In the PS-poll frame, a special field called Uplink Data Indication (UDI) is added, it is
used to indicate the existence of the uplink frame of a station, and a station with no
buffered downlink frame, can send the PS-poll with UDI field set to 1, to request the
time slot for its uplink transmission. After receiving both normal PS-poll and UDI PSpoll, the AP then can adaptively determine how to schedule the uplink and downlink
transmissions efficiently, and the scheduling information is contained in the RA
frame.
Annex F
request to AP protection for a TXOP duration after the expiration of a wakeup timer.
Annex F