MIMO Introduction from Network Perspective

1
MIMO Introduction from Network Perspective

• Aug 30, 2007
• Tae Hyun Kim
• tkim56_at_uiuc.edu

2
Contents

• MIMO Communications
• MIMO Leverages
• Multi-channel vs. MIMO
• Wireless Comm. Theoretical View
• MIMO Spatial Multiplexing System
• Challenges in using MIMO in Wireless Networks
• Current Research
• Summary

3
MIMO Communications

• Signal propagation model
• Path loss, shadowing and scattering
• Large scale fading path loss shadowing
• Small scale fading scattering

path loss
Received Signal Strength (in dB)
shadowing
scattering
Distance (in log)
4
MIMO Communications

• Different paths for different TX-RX pairs
• Large and small scale fading
• Different paths for antennas in the same node
• Assumption rich scattering (small scale fading)
• If multiple radio chains work together,
  interference can be decoded and discarded

Concurrent transmission is not allowed due to
interference
Space-time processing isdone to decode both
signals
5
MIMO Communications

• Multiple-Input and Multiple-Output (MIMO)
• Expanding resource in space dimension
• Key idea different propagation path for each
  signal from different TX antennas

y1h11x1h12x2h13x3
y1
RX
y2h21x1h22x2h23x3
y2
x1
y3
y3h31x1h32x2h33x3
TX
x2
matrix form yHx Space-time
decoding solving linear equations
x3
scattering
6
MIMO Leverages

• Array gain
• by combining multiple signals
• Diversity
• Transmitting same information redundantly
• Usually in a form of space-time (block) coding
• For reliable transmission (such as RTS, CTS, and
  ACK)
• Spatial multiplexing
• Transmitting different info, creating multiple
  spatial streams
• Interference cancellation
• Two methods
• Decoding and discarding signals not destined to
  oneself
• Using spatial streams which are orthogonal to
  each other
• Can be seen as one method of spatial multiplexing

This talk is mainly about spatial multiplexing
7
What can we do with MIMO?
7
8
Multi-channel vs. MIMO

• Multi-channel is about frequency domain
• MIMO introduces a new dimension
• Terms
• Spatial stream for MIMO
• channel for multi-channel communications
• Key difference orthogonality
• Multi-channel sufficient guard band for OFDMA
• MIMO inter-stream interference exists
• Another difference
• MIMO can jointly process inputs(outputs)
• Eventually, MIMO-OFDM, which enables
  space-time-frequency resource management

9
Wireless Comm. Theoretical View

• Single-input and single-output (SISO)
• y hx n
• h channel gain, x transmit symbol, n additive
  noise, y received symbol
• All scalars
• MIMO
• y Hx n
• x, y and n are vectors, and H is matrix

10
Wireless Comm. Theoretical View

• MIMO channel modeling
• How to characterize H?
• Jakes model (Rayleigh fading) is most popular
• Due to mathematical tractability
• Each element of H follows complex Gaussian dist.
• Known Hs eigenvalues and singular values
  distribution
• Ricean fading, Nakagami fading, etc.
• Clustered model, ring model, etc.
• In standards,
• Combination of clustered model, ricean fading,
  and ring model
• Generally, many parameters and different models
  for different environments

11
Wireless Comm. Theoretical View

• Who is the winner in literature?
• Research in MIMO still relies on Rayleigh model
• Turns out that it is not accurate in real world
• Still open area of research
• Possible research topic
• Measurement-based MIMO channel model for
  multi-hop networking

12
MIMO Spatial Multiplexing System

• Why spatial multiplexing?
• Adaptive use achieves reliable, high throughput
  communication
• Has many variants
• Multi-user communications
• Interference cancellation for concurrent
  transmissions
• Space-time coding is useful for reliable
  transmission without channel info
• RTS, CTS, and ACK

13
MIMO Spatial Multiplexing System

• General architecture

Channel
TRANSMITTER
bits
Spatial multiplexer
Modulators
Space-time precoding
feedback
data
optional
RECEIVER
Antennacombining
Demodulators
Spatial multiplexing receiver
De-multiplexer
14
MIMO Spatial Multiplexing System

• MIMO spatial multiplexing (SM) receiver design
• Nonlinear receiver
• Maximum likelihood receiver
• Successive interference cancellation
• Linear receiver
• MMSE receiver
• Zero-forcing receiver
• Receiver post-process for extra RX antennas
• Antenna combining (or equivalently RX
  beamforming)
• Distort channel to obtain either higher SNR or
  SINR

15
MIMO Spatial Multiplexing System

• MIMO transmitter design
• Open-loop spatial multiplexing
• Multi-mode transmission
• Joint adaptation of of spatial streams,
  modulation order, and channel coding rate
• Unitary precoding
• Precoding matrix is chosen in a set of unitary
  matrices
• It has been shown that unitary precoding is
  optimal linear precoding for Rayleigh channel
• Antenna selection
• Only 3dB worse than unitary precoding, but very
  simple

16
MIMO Spatial Multiplexing System

• Joint design of receiver and transmitter
• Singular value decomposition
• H USVH
• U, V unitary matrices, S diagonal matrix
  (gains)
• TX precoding using V RX decoding using UH
• Complete channel info required - unrealistic
• Multi-user MIMO (discussed later)
• Limited feedback system
• How to compress channel info determines TX and RX
  structures

17
Challenges in using MIMO in Wireless Networks

• In sum,
• Interfering MIMO links may provide more capacity
• Protocols for channel info feedback
• Exploitation of multi-user communications in MAC
  or upper layers
• Physical channel aware research

18
Challenges in using MIMO in Wireless Networks

• Interfering MIMO links may provide more capacity
• Methods
• Using antenna selection
• MIMO broadcast or multiple access

19
Challenges in using MIMO in Wireless Networks

• No protocol perspective approach
• How to coordinate antenna selection, MIMO
  broadcast or multiple access?
• Whats the impact on previous research like
  power/CS threshold control, MAC parameter
  tunings?

20
Challenges in using MIMO in Wireless Networks

• Protocol for channel info feedback
• Channel info is essential for MIMO spatial
  multiplexing
• PHY issue how to extract essential info from
  complete channel info
• MAC issue how to efficiently deliver that
• Tradeoff feedback overhead vs. channel info gain

21
Tradeoff example overhead distribution
22
Tradeoff example overhead distribution
23
Challenges in using MIMO in Wireless Networks

• Exploitation of multi-user communications
• MIMO enables multi-user communications
• Paradigm shift from single communication to
  cooperative communication
• Use up spatial streams
• Cooperative protocols
• Network coding

24
Multi-user Comm. Example PHY and MAC performance
of MIMO systems
Single user MIMO cannot use up all transmit
antennas if Nt gt Nr
Multi user MIMO can use up all transmit antennas
even if Nt gt Nr,so channel capacity increases
25
Multi-user Comm. Example PHY and MAC performance
of MIMO systems
Expected gain In ideal case
One hop network with dimensional constraint (Nt
Nr)
26
Challenges in using MIMO in Wireless Networks

• Physical channel aware research
• Mathematical model-based research at PHY limits
  real performance of real MIMO products (maybe
  802.11n?)
• Essential to dynamically grasp channel
  characteristics
• Inter stream relationship
• SISO gain is important
• MIMO correlation of spatial streams is key for
  channel characteristics
• Lack of PHY-MAC-NWK joint simulators
• Methodology how to efficiently simulate this is
  also an issue

27
Current Research

• Multi-user MIMO for ad hoc networks
• To introduce multi-user communications using MIMO
  and its benefits into network community
• Benefits
• More orthogonality between spatial streams
• higher channel capacity
• Virtually more RX antennas for higher capacity
• One-to-multiple point concurrent transmission

28
Current Research
Preliminary PHY performance result

• Ricean model with K40
• TX antennas4, RX antennas2, and 4 users

29
Current Research

• MAC timing diagram
• Borrow from Medium Access Diversity paper

30
Current Research

• MIMO transmission strategy
• Multi-user MIMO broadcast with joint design of
  limited feedback and RX beamforming

31
Current Research

• Contributions
• Introduction of multi-user MIMO
• Useful for a network-wise cooperation
• Space-time frame aggregation
• Cross-layer user scheduling
• Technical difficulty
• Evaluation should be done at both PHY and MAC

32
Summary

• MIMO expands given resource in space dimension
• Brings more freedoms in protocol design
• Due to inter-stream interference, more tricky to
  deal with
• Eventually, MIMO-OFDM system with network
  cooperation (multi-user communications)

33
Thanks!

• Any questions?

34
Backup Slides
35
MIMO Spatial Multiplexing System

• Maximum likelihood receiver
• s arg max Pr s y where s ? X, X set
  of possible symbol vectors
• Reduces to
• s arg min ?y - Hs?2
• Features
• High complexity, high storage requirement

36
MIMO Spatial Multiplexing System

• Successive Interference Cancellation (SIC)
• Streams decoded successively
• At each stage, the contributions of previously
  decoded streams are removed
• V-BLAST (Vertical Bell-lab Layered Space-Time
  architecture)
• Ordered SIC according to SNR

37
MIMO Spatial Multiplexing System

• Linear receiver
• Uses matrix multiplication to equalize channel
• y G(Hx n)
• G is determined by each strategy
• Zero-forcing receiver
• GZF H-1
• Feature
• Noise enhancement
• Lowest performance, but high mathematical
  tractability

38
MIMO Spatial Multiplexing System

• MMSE receiver
• GMMSE arg min E?Gs - y?F2
• The solution is,
• GMMSE (HHH-I)-1HH
• Features
• At low SNR, GMMSE HH matched filter
• At high SNR, GMMSE (HHH)-1HH zero-forcing
• Performance comparison
• ML gt SIC gt MMSE gt ZF