Space Division Multiple Access (SDMA) for Wireless Local Area Network (LAN)

1
Space Division Multiple Access (SDMA) for
Wireless Local Area Network (LAN)
S-88.4221 Postgraduate Seminar on Signal
Processing 1 (6 cp)DSP SYSTEM DESIGN FOR
WIDEBAND WIRELESS COMMUNICATIONS

• Fernando Gregorio
• Signal Processing Laboratory
• HUT

This presentation is based on -P.Vandenameele,
Space Division Multiple Access for Wireless Local
Area Networks, Kluwer Academic Publisher, 2001.
2
Outline

• Introduction
• Spectrally Efficient WLAN
• Bandwidth reuse.
• Pico-cellular WLAN vs. Intra-cell bandwidth
  reuse.
• SDMA
• SDMA-OFDM
• Model
• Receiver structures
• Advanced Receiver structures
• Single-Carrier SDMA
• Advantages over OFDM
• Practical SDMA system
• Conclusions
• References

3
1-Introduction

• The internet traffic double every 100 days.
• Digital mobile phones must be cheap, small and
  power efficient.
• Laptops has become widely available.
• Future users will expect universal wireless
  internet access from their laptops in order to
  obtain a wide range of services and multimedia
  contents.

Spectrally efficient WLAN
4
2-Spectrally efficient WLAN

• In the second generation WLAN (IEEE 802.11a) high
  spectral efficiency can be obtained using
  high-order constellation size.
• The implementation of high-order constellation is
  reduced to good quality channels or small
  coverage area.
• Is it possible to increase spectral efficiency?
• By reusing the bandwidth in adjacent cells
  (Pico-cellularization)
• By reusing the bandwidth within one cell by array
  processing (SDMA)

5
2-Spectrally efficient WLAN

• Pico-cellular WLAN
• Each cell is partitioned in multiple smaller
  cells and to reuse the same frequency bands in
  some of these smaller cells.
• Millimeter wave length carrier frequency is
  needed.
• Low penetration through obstacles.
• High path loss.

6
2-Spectrally efficient WLAN

• Pico-cellular WLAN
• Disadvantages
• Cell size vs. network reinstallation cost
• Cell size vs. cell planning effort
• Cell size vs. total system capacity
• Cell size vs. hand-over and routing

7
2-Spectrally efficient WLAN

• SDMA WLAN
• In this structure the bandwidth can be reused
  within each cell.
• The base station is equipped with an antenna
  array and with digital signal processing that
  allows to separate the signals from multiple
  users sharing the same frequency band and time
  slot.
• The users have only a single antenna giving a
  reduced impact over the system cost.
• Spatial diversity exploitation is preferred over
  beamforming because of the strong multipath
  propagation.

8
2-Spectrally efficient WLAN

• SDMA WLAN
• L different users
• User-specific spatial signature
• The signal signature generated by the channel
  over the transmitted signal acts like spreading
  code in a CDMA system.
• Multiuser detection techniques known from CDMA
  can be applied in SDMA-OFDM

9
2-Spectrally efficient WLAN

• SDMA WLAN vs. Pico-cell

10
3-SDMA-OFDM

• Model

-L users -Base Station with P antennas
11
3-SDMA-OFDM

• Model
• per-carrier basis

Discrete Fourier Transformed Channel User 1,
Antenna 2
1
User 1
2
User 2
P
User L
Base Station
12
4-Receiver structures

• Multiuser detection
• Cellular telephony , satellite communication,
  high-speed data transmission lines, digital TV,
  fixed wireless local loops are subject to
  multi-access interference
• Several transmitters share a common channel.
• The receiver obtains a superposition of the
  signals sent by active transmitters.
• Multiuser detection exploits the considerable
  structure of the multiuser interference in order
  to increase the efficiency with which channel
  resources are employed.

13
4-Receiver structures

• Linear receivers
• Different users transmitted signals are estimated
  with the aid of a linear combiner.
• The residual interference caused by remaining
  users is neglected.

14
4-Receiver structures

• Linear receivers
• L-user
• Statistical characterization

AWGN
Desired user
Interferening users
15
4-Receiver structures

• Linear Receivers
• Least Squares
• LS Combiner combiner attempts to recover the
  signals transmitted by the different users
  regardless of the signal quality quantified in
  terms of the signal to noise ratio (SNR) at the
  reception antennas.
• The linear combiner for user l is designed to
  fully suppress the contribution of all users
  other than user l.

16
4-Receiver structures

• Linear Receivers
• Minimum Mean Squared Error
• Exploits the available statistical knowledge
  concerning the signals transmitted.
• MMSE combiner is designed to minimize the
  expected variance of the error on the combined
  signal, reducing the noise amplification.
• Balance between the recovery signals transmitted
  and the suppression of the AWGN.

17
4-Receiver structures

• Linear Receivers Implementation Complexity

R

• There are two alternatives to solve this
  equation
• LU Factorization
• It converts a general linear system into two
  subsystems know as the LU factorization,
• where L is unit lower triangular and U is upper
  triangular.
• LDLH Factorization
• The correlation matrix R is hermitian and
  positive definite.
• In this factorization L is a unit lower
  triangular and D a diagonal matrix.

18
4-Receiver structures

• Linear Receivers Implementation Complexity

4 Receive antennas and 4 Users Case
19
5-Advanced receiver structures

• Nonlinear receivers
• Linear detector assumes that the different users
  associated linear combiner output are corrupted
  only by AWGN
• Linear combiner output signal contain residual
  interference which is not Gaussian distributed.
• LS and MMSE have sequential structure.
• The operation of classification can be included
  into the linear combination process.
• The residual multi-user interference observed at
  the classifiers input is reduced
• Successive Interference Cancellation (SIC)
• Parallel Interference Cancellation (PIC)

20
5-Advanced receiver structures

• Successive Interference Cancellation (SIC)
• If a decision has been made about an interfering
  users bit, then that interfering signal can be
  recreated at the receiver and subtracted from the
  receiver waveform.
• This will cancel the interfering signal provided
  that the decision is correct, otherwise it will
  double the contribution of the interferer.
• Users with high received power will be
  demodulated in first order best
  performance

21
5-Advanced receiver structures

• Successive Interference Cancellation (SIC)
• Only the specific user having the highest SINR
  (or SIR or SNR) in each iteration at the output
  of the LS or MMSE combiner is detected.
• Having detected this users signal, the
  corresponding
• demodulated signal is subtracted from the
  composite signal received by the different
  antenna elements.
• With this reduced set of received signal a new
  iteration is executed.

22
5-Advanced receiver structures

• Successive Interference Cancellation (SIC)
• Initialization
• Detection
• Calculation of remaining users weight matrix
• Selection of the most dominant user
• Detection of the most dominant user
• Demodulation of the most dominant user.
• Removing of the most important user contribution.
• New iteration

23
5-Advanced receiver structures

• Parallel Interference Cancellation (PIC)
• The order in which users are canceled affects the
  performance of SIC receivers.
• The basic idea of PIC is to estimate the
  transmitted symbols of each user using a
  conventional MMSE method in the first stage. In
  the second stage, the interfering signals can be
  reproduced and removed from the received signal.
• Assuming that the symbols had been estimated
  correctly, a new symbol estimation is carried out
  using the free of interference signal. This
  process can be repeated several times to obtain a
  satisfactory result.

24
5-Advanced receiver structures

• Parallel Interference Cancellation (PIC)

25
5-Advanced receiver structures

• Parallel Interference Cancellation (PIC)

26
5-Advanced receiver structures

• Maximum Likelihood detection
• Optimum from a statistical point of view.
• Potentially excessive computational complexity.
• Join detection of the L different users.
• McL possible combinations of symbols transmitted
  by the L different users are considered by
  evaluating their Euclidean distance from the
  received signal, upon taking into account the
  effects of the channel.

27
5-Advanced receiver structures

• ML estimation
• The estimation procedure can be expressed as
• The estimation of a ML symbol requires comparing
  the Euclidean distance between the vector x of
  the received signals by the different antenna
  elements for all the different vector of symbol
  combinations.

28
Simulation Results
PIC
MMSE
ML
LS
29

• Implementation Complexity

LP4 QPSK
30
OFDM-SDMA Case of study
31
Implementation Complexity
32
6-Single-Carrier SDMA

• Multicarrier systems requires a more linear power
  amplifier and more accurate carrier frequency
  oscillator than single carrier systems.
• Low cost terminals are required in commercial
  applications.
• Single-carrier with cyclic prefix (SC-CP)

Avoids high Peak To Average Power Ratio (PAPR)
and carrier offset sensitivity.
Robust to multipath distortion
Low cost terminals
33
6-Single-Carrier SDMA
34
6-Single-Carrier SDMA

• Linear multiuser detection receivers can be
  applied in the same way than OFDM-SDMA.
• Per-carrier implementation of Non-linear
  detection can not be implemented.
• SIC and PIC can be implemented in the time
  domain.
• Requires back and forward Fourier transform
  during each iteration step.
• High latency.
• Non-linear detection is not a promising technique
  for SC-SDMA

35
7-Practical SDMA implementations

• Real world problems in WLAN implementations
• Channel estimation
• Symbol timing
• Carrier frequency synchronization
• Power control (imbalance in the received power
  from different users)
• Implementation complexity
• Low cost terminals

36
6-Single-Carrier SDMA

• An implementation case of study
• Functional specification

Performance requirements BER10-3 Eb/No15
dB
37
6-Single-Carrier SDMA

• An implementation case of study

U- number of users A Number of receive antennas
w
38
6-Single-Carrier SDMA

• An implementation case of study
• Complexity Results

RU Reused
39

• The total area of the SC-SDMA is amounts to 5 mm2
  
• A SDMA-OFDM modem with similar characteristics
  requires a chip area of 16 mm2

40
8-Conclusions

• SDMA-OFDM is a good alternative for WLAN systems.
• SC-SDMA shows interesting properties in order to
  be considered as a candidate for future WLAN
  implementations.
• Nonlinear detection techniques are not suitable
  for SC-SDMA
• Multiuser channel estimation and frequency
  synchronization are open topics to be considered
  in SDMA-OFDM .
• Nonlinear detection structures provide diversity
  gain.
• Diversity gain vs. Implementation complexity.

41
9-References

•  
• 1)  Sergio Verdu ,  Multiuser Detection, 1998
•  2)  L.Hanzo, M. Munster, B.J. Choi, and
  T.Keller, OFDM and MC-CDMA for broadband
  Multi-User Communication , WLANs   and
  Broadcasting, John Wiley Sons, 2003
• 3) P. Vandenameele, et. al,   A combined
  OFDM-SDMA approach, IEEE Journal on  Selected
  Area on Communications , Nov. 2000

42
Homework

• LS detector reaches similar performance than MMSE
  in high SNR levels.
• Explain
• The order in which users are canceled affects the
  performance of SIC receivers.
• Explain