Hybrid Packet FEC and Retransmissionbased Erasure Recovery Mechanisms HARQ for Lossy Networks: Analy

1
Hybrid Packet FEC and Retransmission-basedErasure
Recovery Mechanisms (HARQ) for LossyNetworks
Analysis and Design

• Vijay Subramanian1, Shiv Kalyanaraman1 and K. K.
  Ramakrishnan2
• 1-(Rensselaer Polytechnic Institute) ,
• 2-(ATT Labs Research)

We gratefully acknowledge support from Air
Force/ESC Hanscom and MIT Lincoln Laboratory,
Letter No. 14-S-06-0206
2
Multi-Tier NLOS MANETs Meshes Challenging
Conditions for TCP/Link Layers
Bursty Losses, Disruptions Protocols need to be
loss tolerant and provide reliability
3
Diversity Techniques HARQ

• Efficient schemes can dynamically use all the
  available channel diversity modes, and degrees of
  freedom to provide an attractive rate-reliability
  tradeoff
• Hybrid ARQ/FEC is a time-diversity technique.
• Error coding (PHY/MAC) Bits flipped, but
  destination does not know which ones flipped
• Erasure coding (Link/Transport)
  packets/fragments erased, the destination does
  not know what was their content THIS PAPERS
  FOCUS

4
Block Erasure Coding Reed-Solomon FEC RS(N,K)
RS(N,K)
FEC (N-K)
Block Size (N)
Questions In dynamic loss conditions How to
set N ? How to choose K? How to use this
inventory of FEC effectively?
Data K
Recovery possible if we receive at least K
packets out of N
5
Short Term Per-Block Loss Binomial Distribution
6
Recall Binomials for different loss rates, N 20

• As Npq gtgt 1, better approximated by normal
  distribution (esp) near the mean
• symmetric, sharp peak at mean, exponential-square
  (e-x2) decay of tails
• (pmf concentrated near mean)

10 PER
30 PER
Npq 4.2
Npq 1.8
N 20 for all cases
50 PER
Npq 5
7
Hybrid ARQ/FEC Scheme Structure (Link/Transport)
8
Adaptive Segmentation and PFEC Efficiency
Addition of PFEC. Total data PFEC
transmission window
Segmentation
Wasted FEC Reduces goodput
Recovered DATA Bytes
9
Adaptive Segmentation, PFEC and RFEC Efficiency
Segmentation
Addition of PFEC
DATA Bytes Partially Recovered
Wasted FEC Reduces goodput
Send RFEC
Recovered DATA Bytes
10
Hybrid ARQ/FEC Modeling Notation Tradeoffs
11
Bursty Loss ON/OFF Model Bimodal-Binomial

• This leads to a bimodal-binomial distribution of
  losses in a block

Special case q r p (binomial distribution
of losses/block). We use p 50 in our
examples
12
How Much Granulation (N)?Key Factor P(all units
lost)
? O(sqrt(N))
When all units/block are lost, the HARQ will fail
(lead to timeouts etc). This probability is
non-trivial for N 5 Small-N Binomial effect.
N gt 10 is good enough.
3.175 blocks irrecoverably lost, i.e. all units
lost and no feedback (eg timeout)
13
Tradeoff with larger N gt Per-Packet Overhead
Set N to 20 at link layer and N to 10 at TCP layer

• TCP layer we cannot increase N (minimum
  granulation) gt constrained by packetization
  overhead
• Link layer Overheads are smaller and so we can
  afford a larger value of N

14
Hybrid ARQ/FEC Scheme Structure (Link/Transport)
15
How much Adaptive PFEC? ? or ? ?
16
How much PFEC? (? - ?) or (? - 2.5?)
17
How much Adaptive PFEC? Summary

• PFEC very efficient lt (? - k?) for small k,
• but it increases the burden on FEC in round 1.
• PFEC very inefficient gt (? k?)
• but it reduces the burden on FECs in round 1.
• Feasible PFEC Choices ? or ??.
• We pick ?? for bursty loss robustness

18
Hybrid ARQ/FEC Scheme Structure (Link/Transport)
19
How Much Adaptive RFEC? Residual Units (X)
Distribution Chopped Binomial!
Chop!
20
Reactive FEC Wastage Modeling
RFECs lost (S) RFECs sent (Y)
RFECs received (Q) Prob. of S RFECs lost P(S
(Y-Q))
21
How much adaptive RFEC? Issues

• Like PFEC, send more RFEC than expected number of
  losses to reduce dependence on future rounds
• Problem Many blocks require only a small number
  of units (X 1 to 5 units).
• Need to send gtgt X units when X is very small to
  counter the small-N binomial effect (and avoid
  timeouts).
• A high relative proportion of RFEC wasted vs RFEC
  sent.
• However, the absolute RFEC waste is low when PFEC
  gt ?
• Total FEC waste still dominated by PFEC waste!

RFEC should be large enough to avoid small-N
binomial effect Some RFEC over-provisioning is
ok even for larger X, to avoid steep timeout
penalties. Absolute overhead matters more than
relative overhead. For TCP, we have to do this
in a partially blind manner (X not known), and
be in line with TCP self-clocking constraints etc
22
RFEC Provisioning Effects
Model RFEC in Round 2 (Y) (X 3?)/(1-p) Add
3? and scale up by (1-p), and round-off to
nearest integer
Despite this aggressive overprovisioning,
absolute RFEC waste is small compared to PFEC
because it is highly targeted, and provides
insurance against bursty loss
23
Model Insights Tradeoffs
Analysis Numbers (p 50) Goodput 3.61 Mbps
vs 5 Mbps (max) PFEC waste 1.0 Mbps 10 RFEC
waste 0.39 Mbps 3.9 Residual Loss
0.0 Weighted Avg Rounds 1.13
24
Model Validation Link/Transport Layer,
Uniform/Bursty 10 50 PER
Remarkably good match, even at the transport
layer (despite having abstracted several
features)
25
Hybrid ARQ/FEC Modeling Summary
How to choose RFEC (Y)? Units Needed (X),
Rfec Wasted, Residual Blocks Unrecovered
26
Thanks!
Thanks also for the support from AFOSR ESC
Hanscom and MIT Lincoln Laboratory, Letter No.
14-S-06-0206

• Researchers
• Vijay Subramanian
• subrav_at_rpi.edu (Rensselaer Polytechnic Institute)
  
• Shivkumar Kalyanaraman
• shivkuma_at_ecse.rpi.edu (Rensselaer Polytechnic
  Institute)
• K.K. Ramakrishnan,
• kkrama_at_research.att.com (ATT Labs Research)

27
Comparison of RFEC Response from Analysis and
Simulation (LT-TCP)

• We consider both Uniform and ON/OFF loss models
  with PER 50.

NO RFEC Needed but some sent in practice. This
can be considered as equivalent to additional
PFEC.
28
Symbols Used and Formulae
29
Block Size (N) Choice Adaptive MSS/Granulation