D.J.C MacKay IEE Proceedings Communications, Vol. 152, No. 6, December 2005.

Slides:



Advertisements
Similar presentations
Jesper H. Sørensen, Toshiaki Koike-Akino, and Philip Orlik 2012 IEEE International Symposium on Information Theory Proceedings Rateless Feedback Codes.
Advertisements

LT Codes Paper by Michael Luby FOCS ‘02 Presented by Ashish Sabharwal Feb 26, 2003 CSE 590vg.
Digital Fountains: Applications and Related Issues Michael Mitzenmacher.
Company LOGO F OUNTAIN C ODES, LT C ODES AND R APTOR C ODES Susmita Adhikari Eduard Mustafin Gökhan Gül.
Performance analysis of LT codes with different degree distribution
Digital Fountain Codes V. S
José Vieira Information Theory 2010 Information Theory MAP-Tele José Vieira IEETA Departamento de Electrónica, Telecomunicações e Informática Universidade.
Jump to first page A. Patwardhan, CSE Digital Fountains Main Ideas : n Distribution of bulk data n Reliable multicast, broadcast n Ideal digital.
LT-AF Codes: LT Codes with Alternating Feedback Ali Talari and Nazanin Rahnavard Oklahoma State University IEEE ISIT (International Symposium on Information.
David Ripplinger, Aradhana Narula-Tam, Katherine Szeto AIAA 2013 August 21, 2013 Scheduling vs Random Access in Frequency Hopped Airborne.
Cooperative Multiple Input Multiple Output Communication in Wireless Sensor Network: An Error Correcting Code approach using LDPC Code Goutham Kumar Kandukuri.
Data Persistence in Sensor Networks: Towards Optimal Encoding for Data Recovery in Partial Network Failures Abhinav Kamra, Jon Feldman, Vishal Misra and.
Compressive Oversampling for Robust Data Transmission in Sensor Networks Infocom 2010.
1 Data Persistence in Large-scale Sensor Networks with Decentralized Fountain Codes Yunfeng Lin, Ben Liang, Baochun Li INFOCOM 2007.
Wireless Network Design for Distributed Control Liu and Goldsmith - Appeared at CDC 2004 Presented by Vinod Namboodiri.
Threshold Phenomena and Fountain Codes
Erasure Correcting Codes
Fountain Codes Amin Shokrollahi EPFL and Digital Fountain, Inc.
Sliding-Window Digital Fountain Codes for Streaming of Multimedia Contents Matta C.O. Bogino, Pasquale Cataldi, Marco Grangetto, Enrico Magli, Gabriella.
1 NETWORK CODING Anthony Ephremides University of Maryland - A NEW PARADIGM FOR NETWORKING - February 29, 2008 University of Minnesota.
1 Distributed LT Codes Srinath Puducheri, Jörg Kliewer, and Thomas E. Fuja. Department of Electrical Engineering, University of Notre Dame, Notre Dame,
Code and Decoder Design of LDPC Codes for Gbps Systems Jeremy Thorpe Presented to: Microsoft Research
RAPTOR CODES AMIN SHOKROLLAHI DF Digital Fountain Technical Report.
1 rStream: Resilient and Optimal Peer-to-Peer Streaming with Rateless Codes Chuan Wu, Baochun Li Department of Electrical and Computer Engineering University.
DNA Research Group 1 Growth Codes: Maximizing Sensor Network Data Persistence Vishal Misra Joint work with Abhinav Kamra, Jon Feldman (Google) and Dan.
Forward Error Correction. FEC Basic Idea Send redundant data Receiver uses it to detect/correct errors Reduces retransmissions/NAKs Useful when RTT is.
1 Verification Codes Michael Luby, Digital Fountain, Inc. Michael Mitzenmacher Harvard University and Digital Fountain, Inc.
Digital Fountain with Tornado Codes and LT Codes K. C. Yang.
The Role of Specialization in LDPC Codes Jeremy Thorpe Pizza Meeting Talk 2/12/03.
Anya Apavatjrut, Katia Jaffres-Runser, Claire Goursaud and Jean-Marie Gorce Combining LT codes and XOR network coding for reliable and energy efficient.
New Protocols for Remote File Synchronization Based on Erasure Codes Utku Irmak Svilen Mihaylov Torsten Suel Polytechnic University.
Repairable Fountain Codes Megasthenis Asteris, Alexandros G. Dimakis IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, VOL. 32, NO. 5, MAY /5/221.
Rateless Codes with Optimum Intermediate Performance Ali Talari and Nazanin Rahnavard Oklahoma State University, USA IEEE GLOBECOM 2009 & IEEE TRANSACTIONS.
Wireless Mobile Communication and Transmission Lab. Theory and Technology of Error Control Coding Chapter 7 Low Density Parity Check Codes.
Rateless Coding with Feedback Andrew Hagedorn, Sachin Agarwal, David Starobinski, and Ari Trachtenberg Department of ECE, Boston University, MA, USA IEEE.
Shifted Codes Sachin Agarwal Deutsch Telekom A.G., Laboratories Ernst-Reuter-Platz Berlin Germany Joint work with Andrew Hagedorn and Ari Trachtenberg.
An Optimal Partial Decoding Algorithm for Rateless Codes Valerio Bioglio, Rossano Gaeta, Marco Grangetto, and Matteo Sereno Dipartimento di Informatica.
Basic Concepts of Encoding Codes, their efficiency and redundancy 1.
Chih-Ming Chen, Student Member, IEEE, Ying-ping Chen, Member, IEEE, Tzu-Ching Shen, and John K. Zao, Senior Member, IEEE Evolutionary Computation (CEC),
X1X1 X2X2 Encoding : Bits are transmitting over 2 different independent channels.  Rn bits Correlation channel  (1-R)n bits Wireless channel Code Design:
User Cooperation via Rateless Coding Mahyar Shirvanimoghaddam, Yonghui Li, and Branka Vucetic The University of Sydney, Australia IEEE GLOBECOM 2012 &
Threshold Phenomena and Fountain Codes Amin Shokrollahi EPFL Joint work with M. Luby, R. Karp, O. Etesami.
1 Network Coding and its Applications in Communication Networks Alex Sprintson Computer Engineering Group Department of Electrical and Computer Engineering.
Kai-Chao Yang VCLAB, NTHU 1.  Unequal Error Protection Rateless Codes for Scalable Information Delivery in Mobile Networks (INFOCOM 2007)  Rateless.
CprE 545 project proposal Long.  Introduction  Random linear code  LT-code  Application  Future work.
Andrew Liau, Shahram Yousefi, Senior Member, IEEE, and Il-Min Kim Senior Member, IEEE Binary Soliton-Like Rateless Coding for the Y-Network IEEE TRANSACTIONS.
Stochastic Networks Conference, June 19-24, Connections between network coding and stochastic network theory Bruce Hajek Abstract: Randomly generated.
15-853:Algorithms in the Real World
Novel network coding strategy for TDD Use of feedback (ACK) improves delay/energy/ throughput performance, especially for high latency- high errors scenarios.
UEP LT Codes with Intermediate Feedback Jesper H. Sørensen, Petar Popovski, and Jan Østergaard Aalborg University, Denmark IEEE COMMUNICATIONS LETTERS,
FEC coding scheme in Multi-hop Wireless Networks Koh Choi Networked Media Laboratory Dept. of Information & Communications Gwangju Institute of.
1 Raptor codes for reliable multicast object delivery Michael Luby Digital Fountain.
A Robust Luby Transform Encoding Pattern-Aware Symbol Packetization Algorithm for Video Streaming Over Wireless Network Dongju Lee and Hwangjun Song IEEE.
Multi-Edge Framework for Unequal Error Protecting LT Codes H. V. Beltr˜ao Neto, W. Henkel, V. C. da Rocha Jr. Jacobs University Bremen, Germany IEEE ITW(Information.
Computer Science Division
Improving Loss Resilience with Multi- Radio Diversity in Wireless Networks by Allen Miu, Hari Balakrishnan and C.E. Koksal Appeared in ACM MOBICOM 2005,
Pei-Chuan Tsai Chih-Ming Chen Ying-ping Chen WCCI 2012 IEEE World Congress on Computational Intelligence Sparse Degrees Analysis for LT Codes Optimization.
1 Unequal Error Protection Using Fountain Codes With Applications to Video Communication Shakeel Ahmad, Raouf Hamzaoui, Marwan Al-Akaidi Faculty of Technology,
Nour KADI, Khaldoun Al AGHA 21 st Annual IEEE International Symposium on Personal, Indoor and Mobile Radio Communications 1.
Reliable Multi-hop Firmware Upload Protocol for mica2 motes. CSE 534 Advanced Networks Dmitri Lusnikov Fall 2004.
Distributed Rateless Codes with UEP Property Ali Talari, Nazanin Rahnavard 2010 IEEE ISIT(International Symposium on Information Theory) & IEEE TRANSACTIONS.
Raptor Codes Amin Shokrollahi EPFL. BEC(p 1 ) BEC(p 2 ) BEC(p 3 ) BEC(p 4 ) BEC(p 5 ) BEC(p 6 ) Communication on Multiple Unknown Channels.
OPTIMIZATION of GENERALIZED LT CODES for PROGRESSIVE IMAGE TRANSFER Suayb S. Arslan, Pamela C. Cosman and Laurence B. Milstein Department of Electrical.
Hongjie Zhu,Chao Zhang,Jianhua Lu Designing of Fountain Codes with Short Code-Length International Workshop on Signal Design and Its Applications in Communications,
Simulation of Finite Geometry LDPC code on the Packet Erasure channel Wu Yuchun July 2007 Huawei Hisi Company Ltd.
1 Implementation and performance evaluation of LT and Raptor codes for multimedia applications Pasquale Cataldi, Miquel Pedros Shatarski, Marco Grangetto,
Coding for Multipath TCP: Opportunities and Challenges Øyvind Ytrehus University of Bergen and Simula Res. Lab. NNUW-2, August 29, 2014.
An improved LT encoding scheme with extended chain lengths
MinJi Kim, Muriel Médard, João Barros
CRBcast: A Collaborative Rateless Scheme for Reliable and Energy-Efficient Broadcasting in Wireless Sensor/Actuator Networks Nazanin Rahnavard, Badri N.
Presentation transcript:

D.J.C MacKay IEE Proceedings Communications, Vol. 152, No. 6, December 2005

 Introduction  Fountain Codes  Intermission  LT Codes  Raptor Codes 2

 Erasure Channel: ◦ Files are transmitted in multiple small packets. ◦ Each packet is either received without error or loss. ◦ Such as the Internet.  How to deal with packet loss? ◦ Some simple retransmission protocols:  ACKs: for missing packets.  ACKs: for received packets. ◦ Erasure Correcting Codes. 3

 Why Erasure Correcting Codes?  Retransmission are wasteful when erasure is serious:  ACKs: for missing packets.  ACKs would be enormous.  ACKs: for received packets.  Would lead to multiple copies.  Broadcast Channel with erasure: 4 server A D C F E

 Erasure Correcting Codes:  Block Code, such as (N, K) Reed-Solomon Code:  Any K of the N transmitted symbols are received, then the original K source symbols can be recovered.  High Complexity: O( K(N-K) log 2 N)  Estimate the erasure probability f first, then choose the code rate R = K/N before transmission.  Ex. If loss-rate = 50%, then set code rate R = 1/(1-50%) = 1/2 = K/N. ( N = 2K ) 5 loss-rate = 50%

 Erasure Correcting Codes:  Block Code, such as (N, K) Reed-Solomon Code:  If f is larger than expected (decoder receives fewer than K symbols):  Ex. We thought loss-rate is 50%, and set the code rate R = 1/2 ( N = 2K ); however, the actual loss-rate = 66.7%, the proper code rate R should be lower: R = 1/3 ( N = 3K )  We would like a simple way to extend the code on the fly to create a lower-rate (N’, K) code. 6 loss-rate = 66.7%  No way!

 Fountain Codes are rateless:  The number of encoded packets generated can be determined on the fly. 7

 Fountain Codes are rateless:  The number of encoded packets generated can be determined on the fly.  Fountain Codes can also have fantastically small encoding and decoding complexities.  Depends on the careful choice of Degree Distribution.

 Balls–and–Bins Problem: ◦ Imagine that we throw N balls independently at random into K bins, what probability of one bin have no balls in it? 9 … K bins … N throws

 Balls–and–Bins Problem: ◦ After N balls have been thrown, what probability of one bin have no ball in it?  The probability that one particular bin is empty after N balls have been thrown: 10 … K bins

 Balls–and–Bins Problem: ◦ After N balls have been thrown, what probability of one bin have no ball in it?  The probability that one particular bin is empty after N balls have been thrown:  The expected number of empty bins: δ =,which roughly implies: the probability of all bins have a ball is (1- δ) only if: 11 … K bins

 Luby Transform (LT) Codes: ◦ Encoding process:  For the i th encoded packet, select degree d i by carefully chosen Degree Distribution (Robust Soliton Distribution).  Choose d i source data.  Perform XOR on chosen data. ◦ Decoding process:  Decode degree-one encoded packets.  Remove degree-one edges iteratively. 12 Degree 123…k probability μ1μ1 μ2μ2 μ3μ3 …μkμk … x1x1 x2x2 x3x3 x4x4 x5x5 y1y1 y2y2 y3y3 y4y4 y5y5 x1x3x1x3 x2x2 x2x5x2x5 x4x4 x6x6 x3x5x6x3x5x6

 Designing the Degree Distribution: ◦ A few encoded packets must have high degree.  To ensure that every source data are connected to encoded packets. ◦ Many encoded packets must have low degree.  So that decoding process can get started, and keep going.  Also the total number of XOR operations involved in the encoding and decoding is kept small. 13 x1x1 x2x2 x3x3 x4x4 x5x5 y1y1 y2y2 y3y3 y4y4 y5y5

 Some properties of Degree Distribution: ◦ The complexity (both encoding and decoding) are scaled linearly with the number of edges in the graph. ◦ Key factor: The average degree of the encoded packets.  How small (number of edges) can this be? ◦ Recall: Balls–and–Bins Problem.  Balls: linked edges.  Bins: source data. 14 x1x1 x2x2 x3x3 x4x4 x5x5 y1y1 y2y2 y3y3 y4y4 y5y5 How small number of edges can assure that every source data must have at least one edge on it? (all bins have a ball)

 Some properties of encoder: ◦ Encoder throws edges into source data at random.  The number of edges must be at least of order : K lnK.  Balls–and–Bins Problem:  The expected number of empty bins: δ =,which roughly implies: the probability of all bins have a ball is (1- δ) only if: 15

 For decoder: ◦ If decoder received optimal K encoded packets, the average degree of each encoded packet must be at least: lnK  The number of edges must be at least of order: K lnK.  The complexity (both encoding and decoding) of an LT code will definitely be at least: K lnK  Luby showed that this bound on complexity can indeed be achieved by a careful choice of Degree Distribution. 16

 Ideally, to avoid redundancy: ◦ We would like just one check node has degree one at each iteration.  Ideal Soliton Distribution:  The expected average degree under this distribution is roughly: lnK 17

 In practice, this distribution works poorly: ◦ Fluctuations around the expected behavior:  Sometimes in the decoding process there will be no degree-one check node. ◦ A few source data will receive no connections at all.  Some small modification fixes these problems. ◦ Robust Soliton Distribution:  More degree-one check node.  A bit more high-degree check node. 18

 Robust Soliton Distribution:  Two extra parameters: c and δ  The expected number of degree-one check node (through out decoding process) is about:  δ: a bound on the decoding failure probability.  Decoding fails to run to completion after K’ of encoded packets have been received.  c: a free parameters smaller than 1.  Luby’s key result. 19

 Luby defines a positive function:, then adds the Ideal Soliton Distribution ρ to τ and normalize to obtain the Robust Soliton Distribution μ: 20, where ※ Receiver once receives K' = KZ encoded packets ensures that the decoding can run to completion with probability at least 1 - δ.

21 τ( k/S ) ※ High-degree ensures every source data is likely to be connected to a check. ※ Small-degree of τ ensures the decoding process gets started.

22 ※ Histograms of the number of encoded packets N required in order to recover source data K = 10,000

23 ※ Practical performance of LT codes - Three experimental decodings are shown. ※ All codes created with c = 0.03, δ = 0.5 (S= 30, K/S = 337, Z = 1.03), and K = 10,000 an overhead of 10%

 Complexity cost: ◦ LT Codes: O(K lnK), where K: the number of original data.  Average degree of the encoded packets: lnK  Encoding and decoding complexity: lnK per encoded packet ◦ Raptor Codes: Linear time encoding and decoding.  Concatenating a weakened LT Code with an outer code.  Average degree of weakened LT code ≒ 3 24

 Weakened LT Code: ◦ Average degree of encoded packets ≒ 3 ◦ A fraction of source data will receive no connections at all.  What fraction? ◦ Balls–and–Bins Problem: 25 Also the fraction of empty bins ≒ 5%

 Shokrollahi’s trick: ※ Encoder: ◦ Find a outer code can correct erasures if the erasure rate is:, then pre-code K source data into: ◦ Transmit this slightly enlarged data using a weaken LT Code. ※ Decoder: ◦ Once slightly more than K encoded packets been received, can recover of the pre-coded packets (roughly K packets). ◦ Then use the outer code to recover all the original data. 26

27 K = 16 K’ = 20 N = 18 ※ Schematic diagram of a Raptor Code Pre-Coding Weaken LT covered = 17

28 ※ The idea of a weakened LT Code. ※ LT codes created with c = 0.03, δ = 0.5 and truncated at degree 8 (thus average degree = 3)

30  Ideal Soliton Distribution:  The average degree is roughly: lnK

31  Robust Soliton Distribution μ:  The average degree is roughly: lnK, where