Get documents about "arun
Document Sample
Cooperative Diversity Techniques
for Wireless Networks
Arun ‘Nayagam
Wireless Information Networking Group (WING)
Department of Electrical and Computer Engineering
University of Florida
Wireless Information Networking Group
Introduction
Antenna arrays commonly used to achieve receive
diversity
Size of the antenna array must be several times the
wavelength of the RF carrier
Antenna arrays are an unattractive choice to achieve
receive diversity in small handsets/cellular phones
Alternative: Network-Based Approaches:
An antenna array is inherently present in any
wireless network!
DISTRIBUTED ARRAY
Different nodes in the network can act like
elements of an antenna array
Wireless Information Networking Group
Introduction (contd.)
CHALLENGES
Array elements are not physically connected
Traditional combining techniques (MRC, EGC)
require large amount of information to be sent to
the combining node
GOAL
Design scalable schemes for achieving receive
diversity with small amount of information exchange
Wireless Information Networking Group
Preliminaries
Error Correcting Codes
Adds structured redundancy to the information bits:
Exploits temporal diversity!
Example: Repetition code:
Coding
Information bit Coded bits
Other examples: Block codes, Trellis-based codes
Coding
Systematic bits Parity bits
Wireless Information Networking Group
Preliminaries (contd.)
Soft-input Soft-output Decoding
a priori LLR Log-MAP a posteriori LLR
+ Decoder (output)
Received symbols
(input)
LLRs referred to as soft
information
Hard-decision=sign(output LLR)
Reliability = |output LLR|
Reliability is an indication of the
correctness of the hard-decision
Wireless Information Networking Group
User-Cooperation: The early days
Information theory: The Relay Channel
First studied by van der Meulen (1968)
Coding theorems proved by Cover and El Gamal (1979)
Relay
Source Destination
Principle
Intermediate nodes called relays process
information from the source and retransmit
“refinement’’ information to the destination
Wireless Information Networking Group
Information Theory (contd.)
Information theory: The Relay Channel
Cover and El Gamal (1979) :
- - Facilitation -
- Cooperation (limited by rate between source and relay) - -
- Observation
Wireless Information Networking Group
Information Theory (contd.)
Information theory: The Relay Channel
Cover and El Gamal (1979) :
- - Facilitation -
- Cooperation (limited by rate between source and relay) - -
- Observation
Wireless Information Networking Group
Information Theory (contd.)
Information theory: The Relay Channel
Cover and El Gamal (1979) :
- - Facilitation -
- Cooperation (limited by rate between source and relay) - -
- Observation
Wireless Information Networking Group
Information Theory (contd.)
Other results
Sendonaris, Erkip and Aazhang (2003) :
User-cooperation increases sum capacity with
knowledge of channel phase at transmitter
Laneman, Wornell and Tse (2003) :
Impossible to increase sum capacity without
knowledge of channel at the transmitter
Cooperation using “dumb” relays
Decode-and-Forward (does not achieve full diversity)
Amplify-and-Forward (full diversity guaranteed)
Wireless Information Networking Group
Information Theory (contd.)
Decode and Forward
Amplify and Forward
Wireless Information Networking Group
Information Theory (contd.)
Drawbacks
Based on repetition coding High overhead
Not scalable to large cooperating groups.
Wireless Information Networking Group
From Theory to Practice
Coded Cooperative Diversity Schemes
Hunter and Nosratinia (2002) :
Cooperation using RCPCs
Coding
Decode and Forward
Wireless Information Networking Group
From Theory to Practice (contd.)
Coded Cooperative Diversity Schemes
Zhao and Valenti (2003) :
Cooperation using Turbo Codes
Decode and Forward
Wireless Information Networking Group
Coded Cooperation (contd.)
Drawbacks
Rely on full decoding at the relay
cannot achieve full diversity!
Not scalable to large cooperating groups.
Wireless Information Networking Group
Objective (Revisited)
Design cooperative schemes that do not depend on
full decoding at any of the relay
achieve full diversity
Cooperation overhead should be small
The scheme should easily scale to large groups of
cooperating nodes
Wireless Information Networking Group
System Model
Distant Transmitter Cluster of Receiving Nodes
COLLABORATIVE DECODING
Nodes iterate between a process of information
exchange and decoding
SCENARIOS
Base station communicating with a group of small
mobile units
Battleship broadcasting a message to a
platoon of soldiers
Wireless Information Networking Group
Cooperative Diversity thro’
Reliability Exchange
- ‘Nayagam, Shea, Wong, Li (WCNC 2003)
IDEA
Bits with low reliabilities are more likely to be
incorrect and hence need information (from other
nodes) to correct them
Bits with high reliabilities are likely to be correct
and hence information about these bits can be
shared with other nodes
Wireless Information Networking Group
Reliability Exchange (contd.)
Least Reliable Bit (LRB) Schemes
Each node identifies the set of least reliable bits and requests
for information about these bits from other nodes
Other nodes reply with
their estimate of the APP
LLR (soft output) for
those bits
Requester and the other
nodes use the received
information as a priori
LLRs
For the nodes other than
the requester, information
is obtained for a set of bits
3 iterations of 5% LRB exchange with random reliabilities
Wireless Information Networking Group
Reliability Exchange (contd.)
Most Reliable Bit (MRB) Schemes
Each node identifies the set of most reliable bit and broadcasts
soft output for these bits to other nodes
Other nodes use the
received information as
a priori LLRs
LLR APPs are broadcast
for the set of MRBs
about which information
was not sent by any node
in the previous iteration
In each iteration a new
set of bits get a priori
3 iterations of 10% MRB exchange information
Wireless Information Networking Group
Overhead Comparisons
Number LRB-2 MRB
of Nodes
2 22.5 % 45 %
5 45.0 % 45 %
10 82.5 % 45 %
20 157.5 % 45 %
Overhead per Receiver
(w.r.t MRC)
Wireless Information Networking Group
Reliability Exchange (contd.)
MRB and LRB schemes lie in the realm of decode-and-forward;
Relay transmission consists of soft-information
Does not require correct decoding of entire block; Even if few
bits decode incorrectly, useful information about other bits can be
extracted
Advantages:
Scales easily to multiple relays
Low overhead
Close to MRC performance on AWGN channels
Disadvantage:
Poor performance on block-fading channels
Wireless Information Networking Group
Design Guidelines
In order to obtain full diversity it is necessary to
exchange information closest to the RF front
end i.e., the received symbol values
(soft demodulator outputs).
More information needs to be combined for
unreliable trellis sections whereas more reliable
sections need less information
Nodes with good channels should share more
information than nodes with bad channels.
Wireless Information Networking Group
Water-filling in the Reliability
Domain
- ‘Nayagam, Shea, Wong (Allerton 2003)
The cooperation process be controlled by a
genie with knowledge of the reliabilities of the
information bits at all relays
Genie selects bits from various nodes for
combining based on water-filling in the reliability
domain : Reliability Filling
An idealized technique similar to MRC
Number of coded symbols combined per
- trellis section is reduced based on the
- reliability
Wireless Information Networking Group
Reliability Filling
3 node MRC example
8 7 13
15 6 6 13 9 11
Wireless Information Networking Group
Reliability Filling (contd.)
3 node reliability filling example (T=10)
8 7 13
15 6 6 13 9 11
Wireless Information Networking Group
Reliability Filling (contd.)
Si is the set of all combinations of nodes such that
- the sum of reliabilities of bit i at those nodes
- exceeds a threshold T
Ni is the minimum number of nodes such that the
sum of reliabilities of bit i at those nodes exceeds T.
When Si = , coded symbols are combined from all
nodes
When Si ≠ , coded symbols are combined from the
smallest number of nodes such that the sum of
reliabilities from those nodes is maximized for bit i.
For different trellis sections, information is combined
from a different set of nodes
Wireless Information Networking Group
Simulation Results
Example of reliability filling with eight cooperating nodes
Non-systematic, non-
recursive convolutional
codes with generator
polynomials 1+D2 and
1+D+D2
Block size =900 bits
BPSK modulation
Block fading channel
Wireless Information Networking Group
Simulation Results
Performance of reliability filling with eight cooperating nodes
Wireless Information Networking Group
Work completed
Developed Proportional Transmission :
A practical iterative technique that
mimics the principles of reliability filling
Developed a mathematically tractable
- expression for the density function of soft
- information to be used in the analysis of
- reliability filling
Analysis of two node reliability filling
Next Step
Analysis of generalized reliability filling ?
Space-time overlays for collaborative decoding ?
Wireless Information Networking Group
Simulation Results
Performance of proportional transmission with eight
cooperating nodes
Wireless Information Networking Group
Numerical Results
Wireless Information Networking Group
