1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom...

53
1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson

Transcript of 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom...

Page 1: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

1

Finite-Length Scalingand Error Floors

Abdelaziz AmraouiAndrea MontanariRuediger Urbanke

Tom Richardson

Page 2: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

2

Approach to AsymptoticPDF File

Page 3: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

3

Finite Length Scaling

Page 4: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

4

Finite Length Scaling

Page 5: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

5

Finite Length Scaling

Page 6: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

6

Finite Length Scaling

Page 7: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

7

Analysis (BEC): Covariance evolution

Fraction of check nodes of degreegreater than one and equal to one.

Covariance terms.

As a function of residual graph fractional size.

Page 8: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

8

Covariance evolution

Page 9: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

9

Finite Length Curves

Page 10: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

10

Analysis (BEC)

• Follow Luby et al: single variable at a time with the trajectory converging to a differential equation.

• Covariance of state space variables also follows a d.e.

• Increments have Markov property and regularity.

Page 11: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

11

Results

Page 12: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

12

Finite Threshold Shift

Page 13: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

13

Generalizing from the BEC?

•No obvious incremental form (diff. Eq.)•No state space characterization of failure.•No clear finite dimensional state space.•Not clear what the right coordinates are

for the general case (Capacity?).

Nevertheless, it is useful in practice to havethis interpretation of iterative failure and tohave the basic form of the scaling law.

Page 14: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

14

Empirical Evidence

Page 15: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

15

Error Floors

Page 16: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

16

Page 17: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

17

Page 18: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

18

Error Floors:Review of the BEC

Page 19: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

19

Error floors on the erasure channel: Stopping sets.

Page 20: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

20

Error floors on the erasure channel: average performance.

Page 21: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

21

Error floors on the erasure channel: decomposition

Page 22: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

22

Error floors on the erasure channel: average and typical performance.

Page 23: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

23

Error floors for general channels: Expurgated Ensemble Experiments.

AWGN channel rate 51/64 block lengths 4k

Random

Girth (8) optimized

Neighborhood optimized

Page 24: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

24

Error floors for general channels: Trapping set distribution.

AWGN channel rate 51/64 block lengths 4k

(3,1)

(5,1)

(7,1)

Page 25: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

25

Observations.

•Error floor region dominated by small weight errors.

•Subset on which error occur usually induces a subgraph with only degree 2 and degree 1 check nodes where the number of degree 1 check nodes is relatively small.

•Optimized graphs exhibit concentration of types of errors.

Page 26: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

26

Intuition

In the error floor event, nodes in the trapping sets receive 1’s with some reliability. Other nodes receive typical inputs.

(Reliable 1)

(Definite 0)

After a few iterations ‘exterior’ nodes and messages converge to high reliability 0s. Internally messages are 1s.

(Definite 0)

1

1

11

1

Nevertheless, if internal received are 1s, internal messaging reaches highly reliable 1 and message state gets trapped.

(Definite 0)

(9,3)

Page 27: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

27

A decoder on an input ℇ Y is a sequence of maps:

Dl : ℇ {0,1}n

Defining Failure: Trapping Sets

(Assume the all-0 codeword is the desired decoding.

For the BEC let 1 denote an erasure.)

We say that bit j is eventually correct if there exists L so that l > L implies Dl(ℇ ) = 0.

Assuming failure, the trapping set T is the set of

all bits that are not eventually correct.

Page 28: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

28

Defining Failure for BP: Practice

Decode for 200 iterations. If the decoding is not successful decode an additional 20 iterations and take the union of all bits that do not decode to 0 during this time.

Page 29: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

29

Trapping Sets: Examples

1. Let the decoder be the maximum likelihood decoder in one step. Then the trapping sets are the non-zero codewords.

2. Let the decoder be belief propagation over the BEC. Then the trapping sets are the stopping sets.

3. Let the decoder be serial (strict) flipping over the BSC. T is a trapping set if and only if the in the subgraph induced by T each node has more even then odd degree neighbors, and the same holds for the complement of T.

Page 30: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

30

Analysis with Trapping Sets:Decomposition of failure

FER() = T P(ℇT, )

ℇT: The set of all inputs giving rise to failure on trapping set T.

Error Floors dominated by “small” events.

Page 31: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

31

1. Find (cover) all trapping sets likely to have significant contribution to the error floor region.

T1,T2,T3,….,Tm

2. Evaluate contribution of each set to the error floor.

P(ℇT1, ), P(ℇT2

, ),…

Predicting Error Floors: A two pronged attack.

Strictly speaking, we get a (tight) lower bound

FER() > i P(ℇTi, )

Page 32: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

32

Finding Trapping Sets

Simulation of decoding can be viewed as stochastic process for finding trapping sets.

It is very inefficient, however.

We could use (aided) flipping to get some speed up.

It is still too inefficient.

Page 33: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

33

Finding Trapping Sets (Flipping)

•Trapping sets can be viewed as “local” extrema of certain functions. E.g., number of odd degree induced checks.

•“Local” means, e.g., under single element removal, addition, or swap.

Therefore, we can look for subsets that are “local” extrema.

Page 34: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

34

Finding Trapping Sets (Flipping)Basic idea:

•Build up a connected subset with bias towards minimizing induced odd degree checks.

•Check occasionally for containment of a in-flipping stable set by applying flipping decoding. Eventually such a set is contained.

•Check now for other types of variation:

•Out-flipping stability.

•Single aided flip stability (chains).

•……

Page 35: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

35

Differences: BP and Flipping

Page 36: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

36

Differences: BP and Flipping

r1

r1+r2+r3

r3

r2

Page 37: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

37

Differences: BP and Flipping

Page 38: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

38

Differences: BP and FlippingPDF File

Page 39: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

39

•Find random variable x on which to condition the decoder

input Y that “mostly” determines membership in ℇT . I.e.,

Pr{ℇT | x} is nearly a step function in x.

•Perform in situ simulation of trapping set while varying x to

measure Pr{ℇT | x}.

•Combine with density of x to get Pr{ℇT }.

Evaluating Trapping SetsBasic idea:

Page 40: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

40

Evaluating Trapping Sets

Condition input to trapping set

Otherwise simulate channel

Page 41: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

41

Evaluating Trapping Sets: BEC

X is the number of erasures in T (=S).

Page 42: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

42

Evaluating Trapping Sets: AWGN

X is the mean noise input in T.

Page 43: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

43

Evaluating Trapping Sets: AWGN

Page 44: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

44

Evaluating Trapping Sets: Margulis 12,4

Page 45: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

45

A tougher test case:G

Page 46: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

46

Evaluating Trapping Sets: G

Page 47: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

47

A Curve from a single point

Page 48: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

48

Extrapolating a curve

Page 49: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

49

Variation in Trapping Sets: (10,4) (10,2) (10,0)

Page 50: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

50

Variation in Trapping Sets: (12,4) (12,2) (12,0)

Page 51: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

51

Variation in Trapping Sets

Page 52: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

52

Conclusions

•Error floor performance is predictable with considerable computational effort. (Would be nice to have scaling law for “best” codes.)

•Trade off between error floor and threshold (waterfall) can be optimized even for very deep error floors.

Page 53: 1 Finite-Length Scaling and Error Floors Abdelaziz Amraoui Andrea Montanari Ruediger Urbanke Tom Richardson.

53

Conclusions

“It is interesting to observe that the search for theoretical understanding of turbo codes has transformed coding theorists into experimental scientists.”physicists.”