802.11n Specification and the use of Space-Time Wireless Channels

1
802.11n Specification and the use ofSpace-Time
Wireless Channels

• Shad Nygren
• April 27, 2006
• Del Mar Electronics Show

2
Objectives

• Discuss the history and present state of the
  802.11n specification.
• Discuss MIMO, Space-Time Wireless Channels and
  Space-Time Block Codes which are one of the most
  interesting aspects of the 802.11n specification.
• Understand how the magic of MIMO and Space-Time
  Wireless Channels work.

3
About Me

• Masters Degree in Computer Science from
  University of Nevada, Reno
• 24 years experience with computers, networking
  and wireless communications

4
802.11n History

• 1999, 802.11a/b standards ratified by IEEE
• June 2003, 802.11g ratified by IEEE.
• 802.11g was based on OFDM from 802.11a but using
  the 2.4GHz band and backwards compatible with
  802.11b
• January 2004, IEEE forms new 802.11 Task Group
  (TGn) to investigate higher data rates

5
802.11n History Cont

• Standards Process From many proposals down to
  two
• TGnSync
• WWiSE
• After much debate these two groups created a
  Joint Proposal
• October 2005, the Enhanced Wireless Consortium
  (EWC) was founded by Intel, Broadcom, Marvell,
  Atheros and others

6
802.11n Progress in 2006

• Jan 19, 2006, IEEE 802.11n task group approved
  the Joint Proposals specification based on EWCs
  specification.
• March 2006 IEEE 802.11 Working Group sent the
  802.11n Draft to its first letter ballot.
• Currently working its way thru the IEEE standards
  process. Hopefully a final standard will be in
  place in about a year.

7
802.11n Goals

• Investigate next generation wireless LAN
  technology capable of supporting multimedia
  applications
• Provide higher data rates than 802.11b/g At
  least 100Mbps at MAC layer
• Backwards compatibility with 802.11b/g

8
802.11n Physical Layer

• Operates in 2.4GHz and/or 5GHz unlicensed bands
• Uses OFDM like 802.11a/g
• Backwards compatible and mixed mode interoperable
  with 802.11a/b/g
• High Throughput (HT) and 40MHz modes
• Optionally uses MIMO

9
2.4GHz Unlicensed Band 802.11b Channel Frequency
Map
Channel Lower Freq Center Freq Upper Freq
1 2.401 2.412 2.423
2 2.406 2.417 2.428
3 2.411 2.422 2.433
4 2.416 2.427 2.438
5 2.421 2.432 2.443
6 2.426 2.437 2.448
7 2.431 2.442 2.453
8 2.436 2.447 2.458
9 2.441 2.452 2.463
10 2.446 2.457 2.468
11 2.451 2.462 2.473
10
802.11g OFDM

• 64 point FFT
• 52 OFDM subcarriers
• 48 Data Carriers
• 4 Pilot Carriers
• 12 unused carriers
• Carrier Separation 0.3125MHz (20MHz/64)
• Total Bandwidth 20MHz with occupied bandwidth of
  16.6MHz
• Symbol duration 4us with 0.8us guard interval

11
OFDM Carriers
Source International Engineering
Consortium http//www.iec.org/online/tutorials/ofd
m/topic04.html
12
802.11a/g OFDM Rates250,000 Symbols per Sec
Modulation Coding Rate Data Carriers Data Rate (Mbps)
BPSK 1/2 48 6
BPSK 3/4 48 9
QPSK 1/2 48 12
QPSK 3/4 48 18
16 QAM 1/2 48 24
16 QAM 3/4 48 36
64 QAM 2/3 48 48
64 QAM 3/4 48 54
13
802.11a/g OFDM Physical Layer

• Divided into two elements
• PLCP Physical Layer Convergence Protocol
  prepares frames for transmission and directs the
  PMD to transmit and receive signals, change
  channels etc
• PMD Physical Medium Dependant layer provides
  actual transmission and reception over the
  wireless medium by modulating and demodulating
  the frame transmissions

14
Options for Increasing Data Rate

• Double the Clock Rate From 20MHz (250,000
  Symbols per Second) to 40MHz (500,000 Symbols per
  Second)
• Double the Number of Carriers From 64 to 128,
  not increasing the bandwidth
• Use Higher Order Modulation From 64QAM (6 bits
  / symbol) to 4096QAM (12 bits / symbol)

15
Options for Increasing Data Rate

• OFDM with Bit Loading Different Modulation Per
  Carrier
• Better Code Turbo or Low Density Parity Check
• MIMO Multiple Input Multiple Output antennas
  for multiple data streams

16
Higher Data Rate Considerations
Larger Constellation 54Mbps already uses 64QAM. Can a wireless system support a larger constellation?
Turbo Coding Requires at least 3 or 4 iterations for good performance.
Double Bandwidth Inefficient use of bandwidth.
MIMO Multiple Antennas Cost is the additional antennas and RF electronics, the DSP does not add much complexity to existing receivers.
17
802.11n OFDM

• 20MHz High Throughput Mode
• 56 OFDM subcarriers
• 52 Data Carriers
• 4 Pilot Carriers
• 40MHz High Throughput Mode
• 114 OFDM subcarriers (2 extra subcarriers)
• 108 Data Carriers (4 extra data carriers)
• 6 Pilot Carriers (2 less pilot carriers)

18
802.11n Mandatory Features

• Frame Aggregation
• Block ACK
• N-immediate ACK Block ACK between two HT peers
  using an immediate Block Ack policy
• Long NAV Provides protection for a sequence of
  multiple PPDUs

19
NAVNetwork Allocation Vector

• Counter resident at each station that represents
  the amount of time that the previous station
  needs to send its frame.
• The NAV must be zero before a station can attempt
  to send a frame.
• The transmitting station calculates the amount of
  time necessary to send the frame based on the
  frames length and data rate.

20
NAVNetwork Allocation Vector

• The transmitting station places a value in the
  duration field in the header representing the
  time required to transmit the frame.
• When stations receive a frame, they examine the
  duration field value and use it as the basis for
  setting their corresponding NAV.
• This process reserves the medium for the sending
  stations.

21
802.11n Optional Features

• Advanced Coding Using different coding per OFDM
  carrier
• Green Field mode
• Beamforming
• Short Guard Interval Reduce from 800ns (250,000
  symbols per second) to 400ns and send 277,778
  symbols per second
• Space Time Block Coding

22
802.11n Modes

• Legacy Mode packets are transmitted in the
  legacy 802.11a/g format
• Mixed Mode packets are transmitted with a
  preamble compatible with 802.11a/g so they can be
  decoded by legacy devices while the rest of the
  packet is transmitted in the new mode
• Green Field optional mode where the packets are
  transmitted without the legacy compatibility part

23
802.11n for 20/40MHz operation

• 40MHz comprised of two adjacent 20MHz channels
• One Control Channel
• One Extension Channel
• Beacon is sent in legacy mode on the control
  channel only
• A single BSS may include
• 20MHz-only capable stations
• 20/40MHz capable stations
• Legacy stations
• Clear Channel Assessment will be done on the
  control channel and possibly on the extension
  channel. The results will then be combined.

24
802.11n Modulation and Coding per Spatial Stream
Modulation Code Rate Data Carriers Data Rate Mbps (GI800ns) Data Rate Mbps (GI400ns)
BPSK 1/2 52/108 6.5/13.5 7.22/15
QPSK 1/2 52/108 13/27 14.44/30
QPSK 3/4 52/108 19.5/40.5 21.66/45
16QAM 1/2 52/108 26/54 28.88/60
16QAM 3/4 52/108 39/81 43.33/90
64QAM 2/3 52/108 52/108 57.66/120
64QAM 3/4 52/108 58.5/121.5 65/135
64QAM 5/6 52/108 65/135 72.22/150
25
802.11n Modulation and Coding Two Spatial Streams
Modulation Code Rate Data Carriers Data Rate Mbps (GI800ns) Data Rate Mbps (GI400ns)
BPSK 1/2 52/108 13/27 14.44/30
QPSK 1/2 52/108 26/54 28.88/60
QPSK 3/4 52/108 39/81 43.32/90
16QAM 1/2 52/108 52/108 57.76/120
16QAM 3/4 52/108 78/162 86.66/180
64QAM 2/3 52/108 104/216 115.32/240
64QAM 3/4 52/108 117/243 130/270
64QAM 5/6 52/108 130/270 144.44/300
26
802.11n Modulation and Coding Four Spatial Streams
Modulation Code Rate Data Carriers Data Rate Mbps (GI800ns) Data Rate Mbps (GI400ns)
BPSK 1/2 52/108 26/54 28.88/60
QPSK 1/2 52/108 52/108 57.76/120
QPSK 3/4 52/108 78/162 86.64/180
16QAM 1/2 52/108 104/216 115.52/240
16QAM 3/4 52/108 156/324 173.32/360
64QAM 2/3 52/108 208/432 230.64/480
64QAM 3/4 52/108 234/486 260/540
64QAM 5/6 52/108 260/540 288.88/600
27
MIMO

• Any sufficiently advanced technology is
  indistinguishable from magic.
• Arthur C. Clarke

28
MIMO Magic

• MIMO is not magic but is an advanced RF
  communications technology based on valid
  mathematical and scientific principals
• MIMO does not violate Shannons Law
• Pronounced MyMoe This was standardized by a
  vote at an IEEE meeting.

29
Multiple Antennas

• Well studied topic for the past few years
• OFDM is very well suited for use with multiple
  antennas
• Many existing 802.11 products already have 2
  antennas, using switched diversity
• Additional component required for exploiting full
  diversity is an additional RF front-end
• Recent advances in RF technology will make this
  cost effective in the near future

30
Antenna Diversity

• Space Diversity
• Polarization Diversity
• Pattern Diversity
• Transmit Diversity

31
Temporal Diversity

• Frequency Diversity
• Code Diversity
• Time Diversity

32
Diversity Reception

• Idea from which MIMO arose
• Several methods are possible
• Selection Combining
• Switched Combining
• Equal Gain Combining
• Maximum Ratio Combining

33
Maximum Ratio Combining (MRC)

• A way of combining signals from diversity
  reception
• The signals are weighted according to their
  Signal to Noise ratios and then combined

34
Diversity Gain Definition

• Diversity Transmission - is a method for
  improving reception of a transmitted signal, by
  receiving and processing multiple versions of the
  same transmitted signal
• Diversity Gain - is a value that quantifies the
  performance improvement by a diversity
  transmission scheme in a fading channel

35
Diversity Gain for Multiple Branches

• The performance gain of a system can be quite
  dramatic
• For example, with a system using QPSK requiring a
  maximum BER of 0.01 diversity gain is 13.9dB

Source Space-Time Wireless Channels by Durgin
36
Shannon Capacity for Conventional Systems

• 1948 Claude Shannons Noisy Channel Coding
  Theorem describes maximum efficiency of error
  correcting codes
• Shannon-Hartley Theorem describes what channel
  capacity is for finite bandwidth continuous time
  channel with Gaussian Noise
• With Single Transmit and Single Receive Antenna
• B is Bandwidth
• SINR is Signal to Interference and Noise Ratio
• C can be increased by increasing B or SINR

37
Shannon Capacity forConventional Multi-Antenna
Systems

• SINR ratio can be improved by using multiple
  antennas
• Overall capacity can be improved because the SINR
  is improved
• Multiple Transmit Antennas
• Multiple Receive Antennas
• Combination of multiple Transmit and Receive
  antennas

38
SINR withMultiple Receive Antennas

• N antennas are used at the receiver
• They receive N various faded copies of the signal
• Which can be coherently combined to produce a N2
  increase in power
• There are also N sets of noise/interference that
  add together as well

39
Shannon Channel Capacity with Multiple Receive
Antennas

• With this NSINR the channel capacity of the
  system becomes

40
SINR withMultiple Transmit Antennas

• If M antennas are used at the transmitter the
  total power is divided into the M branches.
• The power per transmitter antenna drops but
  signals may be phased so that they add coherently
• Noise interference is the same as SISO
• The result is a M-fold increase in SINR

41
Shannon Channel Capacity with Multiple Transmit
Antennas

• With this MSINR the channel capacity of the
  system becomes

42
SINR with Multiple Transmit and Multiple Receive
Antennas

• SINR is a combination of the MISO (multiple
  transmit antennas) SIMO (multiple receive
  antennas) cases

43
Shannon Capacity of a Single Channel with
Multiple Transmit and Multiple Receive Antennas

• With this MNSINR the channel capacity of the
  conventional system using multiple antennas
  becomes

44
Conventional Multi-Antenna Transmission

• Conventionally it is not possible to send more
  than one simultaneous signal per frequency
• Seemingly the best approach would be to weight
  the transmitter elements to maximize signal power
  at the receiver.

Source DATACOMMRESEARCH
45
Increasing Shannon capacity by using multiple
spatial channels

• A shift in perspective led to the development of
  truly multiple-input, multiple-output systems
  that have capacity greater than the best
  conventional single channel system.
• Dramatic capacity increases are possible if we
  consider different signals sent thru each
  transmitter antenna.

46
Multi-Channel MIMO

• Different signals are are sent thru each
  transmitter antenna

Source DATACOMMRESEARCH
47
Wont the physical channels interfere with each
other?I dont believe this is possibleShow me
the Math
48
MIMO Channel Matrix Model

• y received vector
• x transmitted vector
• H channel matrix
• t time, t delay

49
Processing the MIMO Signalat the Transmitter

• At the transmitter a linear signal processing
  operation V is performed on the transmitted
  signal vector x and the result is Vx(t)
• V is an M x M unitary matrix with the property
  VV I where I is the identity matrix and the
  operator indicates the conjugate transpose or
  Hermitian operation
• Unitary matrices do not change the geometrical
  length of vectors so no power is added or
  subtracted from the transmitted signal

50
Processing the MIMO signalat the Receiver

• At the receiver a linear processing signal
  processing operation U is performed on the
  received signal vector y
• U is an N x N unitary matrix where UU I
• I is the identity matrix which means that no
  power is being added or subtracted from the
  received signal

51
MIMO Processing Output

• After the channel H operates on the transmitters
  output Vx(t) the result is HVx(t)
• The receiver then processes this signal with
  matrix U and the result is z(t) described by the
  following
• The wireless system has no control over the
  channel H but by controlling U and V so it can
  control D

52
Controlling the Channel

• U and V are chosen such that they diagonalize D
• ?i are positive constants
• Here N gt M so there are M separate channels
• If M gt N then this is limited to N separate
  channels

53
The result is simplifying z(t)

• The result of diagonalizing the matrix is to
  simplify the received and processed vector z(t)
• Mathematically this shows that the MIMO channel
  can be viewed as a set of Min(M,N) separate
  channels

54
Singular Value Decomposition

• These signal processing steps have a distinct
  physical rational
• They rearrange the channel without adding or
  subtracting power so they do not change the
  channel capacity by amplification
• What they have actually done is a Singular Value
  Decomposition on the channel matrix H
• When squared the diagonal elements of D are the
  eigen values of HH for NgtM or HH for MgtN

55
Capacity Increase withSeparate Channels

• If each signal is a different signal then each of
  the individual channels will have a capacity
• C Blog2(1(N/M)SINR)
• Since there are Min(M,N) of these channels the
  total capacity is
• C Min(M,N)Blog2(1(N/M)SINR)
• Observe how this differs from conventional
  multi-antenna channel capacity
• C Blog2(1 MNSINR)
• There is a linear increase in capacity by Min(M,N)

56
Power of logarithms

• Recall basic property of logarithms XlogN(Y)
  logN(YX)
• Therefore MBlog2(1(N/M)SINR) gt
  Blog2(1MNSINR)
• The essential principle is that it is more
  beneficial to transmit data using many different
  low power channels than a single high power
  channel

57
Physical Interpretation of U and V

• U and V are matrices of complex (amplitude and
  phase) values
• At the transmitter, matrix V operates on symbol
  vector x(t) to effectively provide a unique
  antenna radiation pattern for each symbol
• At the receiver, matrix U operates similarly to
  provide unique antenna patterns that effectively
  pick out different symbols arriving from
  different directions because of multipath
  reflections

58
Knowing the Channels

• In order for a system to achieve this
  supercharged capacity it must be able to
  calculate the correct unitary matrices U and V
• Since U and V depend on the channel matrix H it
  is necessary to estimate the channel at both the
  transmitter and receiver
• Presumably the channel matrix information must be
  sent from the receiver to the transmitter.
• But perhaps not. Maybe there is another way.

59
Practical Signal Extraction

• Few wireless systems will perform SVD on the
  channel at both the transmitter and receiver
  because this requires reliable estimates of the
  channel at both transmitter and receiver
• Instead a training sequence is transmitted to the
  receiver so that it has a reliable channel
  estimate
• Then the receiver operating matrix U is set to
  be the inverse of the channel matrix H so U H-1

60
Practical Signal Extraction Cont

• HH-1 I
• I is the identity matrix
• This has the effect of nulling out the distortion
  effects of the wireless channel

61
Subtraction of Interference

• Data could be processed this way but there is an
  interesting opportunity for signal gain if the
  symbols are processed in the following manner

• Subsequent symbols are processed by subtracting
  previously determined symbols giving 2 estimates
  for the 2nd symbol, 3 for the 3rd and 4 for the
  4th
• These multiple estimates can be combined for
  additional diversity gain

62
Foschinis Layered Architecture

• One of the problems with MIMO is its
  vulnerability to unequal power channels
• Because of this the channels cannot be separated
  at the receiver with equal SINR by using a simple
  inversion operation H-1
• Gerard Foschini in his famous 1996 paper on MIMO
  proposed a transmitter architecture that cycles
  the four streams, one cycle per timeslot
• Thus on average each channel has the same SINR
• This paper stimulated a lot of research in MIMO

63
Optionally using diversity for adding redundancy

• Recall the conventional multi-antenna
  transmission scheme
• For simple MISO case with M2, N1 the channel
  matrix is

64
Conventional MISO Transmission

• This is done as follows
• Each column represents a successive timeslot
• Complex weights v11 and v21 are chosen by the
  transmitter but info must be fed from receiver to
  transmitter to make the best choice.
• Successive symbols are represented by xi

65
A better way to transmit symbols

• Instead use the following method
• Each column represents a successive time slot
• Send different symbols from each antenna
• First send the symbol and then follow by sending
  the complex conjugates.

66
What the receiver sees

• For timeslot 1

• For timeslot 2

• Separately these are just useless mangled
  combinations of the two data symbols

67
Combining Samples

• However the samples can be combined in the
  following manner from which the original symbols
  x1 and x2 can be recovered
• Notice the real constant R on the right side

68
Result of Combining

• The constant R is equivalent to the output
  envelope of a two-branch diversity scheme with
  MRC
• The single-antenna receiver has performed MRC on
  the transmitted symbols
• The receiver still had to have reliable channel
  estimates but didnt send them to the transmitter
• It is therefore possible to use a MISO system to
  combat a fading channel without requiring channel
  feedback

69
Alamoutis Code

• First two columns of transmission are a special
  matrix called Alamoutis Code invented in 1998

• Alamoutis Code is a special instance of a code
  called a Space-Time Block Code (STBC)
• Very special because it is the only orthogonal
  STBC that achieves rate-1 and therefore achieves
  full diversity gain without sacrificing data rate.

70
Space Time Codes

• A method used to improve the reliability of data
  transmission by using multiple transmit antennas
• Modulation scheme that provides transmit
  diversity
• Rely on transmitting multiple redundant copies of
  data stream to the receiver in the hope that at
  least some of them make it and allow reliable
  decoding.
• Two Main Types
• Space Time Block Code (STBC)
• Space Time Trellis Code (STTC)

71
STBC Space Time Block Code

• Easiest type because under the assumption of flat
  fading Rayleigh channels they can be decoded
  using simple linear processing at the receiver
• STBCs create an antenna array in time
• Represented as a matrix where
• Each row represents a time-slot
• Each column represents a transmit antenna

72
Observations onSTBC and MISO

• The number of channels and the potential for
  speed improvement is Min(M,N)
• If you only have one receive antenna you can only
  have one channel
• However, with only one receive antenna but
  multiple transmit antennas STBC allow tremendous
  diversity gain
• Diversity gain provides better BER and allows
  protocols to use faster data rates without having
  to fall back to slower data rates.

73
Higher Order STBC

• Higher order STBC are possible and must be used
  for 3 x 3 or 4 x 4 or M x N systems
• However, it has been proven that no code using
  more than two antennas can reach rate-1.
• This is because it is the only way for a code to
  reach orthogonality.

74
STTC Space Time Trellis Code

• Based on trellis codes
• Provide both coding gain and diversity gain
• Have better bit-error rate performance than STBC
• More complex to encode and decode than STBC
• Rely on Viterbi decoder at the receiver
• Require information about Channel to be conveyed
  from the receiver to the transmitter

75
Multipath is Essential for MIMO

• Problems with MIMO
• Without multipath it degenerates into a single
  transmitter and receiver
• Unequal average branch power
• Keyhole problem

Source Space-Time Wireless Channels by Durgin
76
MIMO Pros and Cons

• Advantage
• Linear increase in capacity with the number of
  antennas
• Multiple paths provide resistance to fading
• Disadvantage
• Cost of multiple RF chains
• Higher power consumption

77
How will MIMO effect you?

• MIMO is a radical paradigm shift away from one
  transmitter / one receiver
• It will change the design paradigm for virtually
  all wireless technologies from cell phones to
  broadband
• If you are involved in wireless then MIMO is in
  your future

78
Other Standards using MIMO

• WiMax
• Cellular
• WiBro

79
Conclusion

• Spatial Multiplexing for higher data rates is
  mandatory in 802.11n
• STBC for diversity and redundancy are optional in
  802.11n
• MIMO requires a multipath environment
• MIMO advantages outweigh disadvantages and many
  standards are adopting it

80
References

• Space-Time Wireless Channels by Gregory D. Durgin
• Provides excellent into understand Space-Time
  Wireless Channels
• If you have had college level engineering
  calculus and statistics then this will be
  understandable

81
References Cont

• Gerard J. Foschini 1996 Layered Space-Time
  Architecture for Wireless Communication in a
  Fading Environment When Using Multi-Element
  Antennas
• DATACOMM Research Company White Paper Using
  MIMO-OFDM Technology To Boost Wireless LAN
  Performance Today http//www.datacommresearch.com
  
• Enhanced Wireless Consortium - http//www.enhanced
  wirelessconsortium.org
• EWC_MAC_spec_V124.pdf
• EWC_PHY_spec_V127.pdf

82
Thank You

• Questions?