idnits 2.17.1
draft-ietf-fecframe-ldpc-01.txt:
Checking boilerplate required by RFC 5378 and the IETF Trust (see
https://trustee.ietf.org/license-info):
----------------------------------------------------------------------------
No issues found here.
Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt:
----------------------------------------------------------------------------
No issues found here.
Checking nits according to https://www.ietf.org/id-info/checklist :
----------------------------------------------------------------------------
No issues found here.
Miscellaneous warnings:
----------------------------------------------------------------------------
== The copyright year in the IETF Trust and authors Copyright Line does not
match the current year
-- The document date (November 29, 2011) is 4530 days in the past. Is this
intentional?
Checking references for intended status: Proposed Standard
----------------------------------------------------------------------------
(See RFCs 3967 and 4897 for information about using normative references
to lower-maturity documents in RFCs)
-- Looks like a reference, but probably isn't: '0' on line 408
-- Looks like a reference, but probably isn't: '1' on line 410
-- Looks like a reference, but probably isn't: '2' on line 412
-- Looks like a reference, but probably isn't: '3' on line 414
Summary: 0 errors (**), 0 flaws (~~), 1 warning (==), 5 comments (--).
Run idnits with the --verbose option for more detailed information about
the items above.
--------------------------------------------------------------------------------
2 FecFrame V. Roca
3 Internet-Draft INRIA
4 Intended status: Standards Track M. Cunche
5 Expires: June 1, 2012 NICTA
6 J. Lacan
7 ISAE/LAAS-CNRS
8 November 29, 2011
10 Simple LDPC-Staircase Forward Error Correction (FEC) Scheme for FECFRAME
11 draft-ietf-fecframe-ldpc-01
13 Abstract
15 This document describes a fully-specified simple FEC scheme for LDPC-
16 Staircase codes that can be used to protect media streams along the
17 lines defined by the FECFRAME framework. These codes have many
18 interesting properties: they are systematic codes, they perform close
19 to ideal codes in many use-cases and they also feature very high
20 encoding and decoding throughputs. LDPC-Staircase codes are
21 therefore a good solution to protect a single high bitrate source
22 flow, or to protect globally several mid-rate flows within a single
23 FECFRAME instance. They are also a good solution whenever the
24 processing load of a software encoder or decoder must be kept to a
25 minimum.
27 Status of this Memo
29 This Internet-Draft is submitted in full conformance with the
30 provisions of BCP 78 and BCP 79.
32 Internet-Drafts are working documents of the Internet Engineering
33 Task Force (IETF). Note that other groups may also distribute
34 working documents as Internet-Drafts. The list of current Internet-
35 Drafts is at http://datatracker.ietf.org/drafts/current/.
37 Internet-Drafts are draft documents valid for a maximum of six months
38 and may be updated, replaced, or obsoleted by other documents at any
39 time. It is inappropriate to use Internet-Drafts as reference
40 material or to cite them other than as "work in progress."
42 This Internet-Draft will expire on June 1, 2012.
44 Copyright Notice
46 Copyright (c) 2011 IETF Trust and the persons identified as the
47 document authors. All rights reserved.
49 This document is subject to BCP 78 and the IETF Trust's Legal
50 Provisions Relating to IETF Documents
51 (http://trustee.ietf.org/license-info) in effect on the date of
52 publication of this document. Please review these documents
53 carefully, as they describe your rights and restrictions with respect
54 to this document. Code Components extracted from this document must
55 include Simplified BSD License text as described in Section 4.e of
56 the Trust Legal Provisions and are provided without warranty as
57 described in the Simplified BSD License.
59 Table of Contents
61 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3
62 2. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 4
63 3. Definitions Notations and Abbreviations . . . . . . . . . . . 4
64 3.1. Definitions . . . . . . . . . . . . . . . . . . . . . . . 4
65 3.2. Notations . . . . . . . . . . . . . . . . . . . . . . . . 6
66 3.3. Abbreviations . . . . . . . . . . . . . . . . . . . . . . 7
67 4. Common Procedures Related to the ADU Block and Source
68 Block Creation . . . . . . . . . . . . . . . . . . . . . . . . 7
69 4.1. Restrictions . . . . . . . . . . . . . . . . . . . . . . . 7
70 4.2. ADU Block Creation . . . . . . . . . . . . . . . . . . . . 7
71 4.3. Source Block Creation . . . . . . . . . . . . . . . . . . 9
72 5. LDPC-Staircase FEC Scheme for Arbitrary ADU Flows . . . . . . 10
73 5.1. Formats and Codes . . . . . . . . . . . . . . . . . . . . 10
74 5.1.1. FEC Framework Configuration Information . . . . . . . 10
75 5.1.2. Explicit Source FEC Payload ID . . . . . . . . . . . . 12
76 5.1.3. Repair FEC Payload ID . . . . . . . . . . . . . . . . 13
77 5.2. Procedures . . . . . . . . . . . . . . . . . . . . . . . . 14
78 5.3. FEC Code Specification . . . . . . . . . . . . . . . . . . 14
79 6. Security Considerations . . . . . . . . . . . . . . . . . . . 14
80 6.1. Attacks Against the Data Flow . . . . . . . . . . . . . . 14
81 6.1.1. Access to Confidential Content . . . . . . . . . . . . 15
82 6.1.2. Content Corruption . . . . . . . . . . . . . . . . . . 15
83 6.2. Attacks Against the FEC Parameters . . . . . . . . . . . . 15
84 6.3. When Several Source Flows are to be Protected Together . . 16
85 6.4. Baseline Secure FEC Framework Operation . . . . . . . . . 16
86 7. Operations and Management Considerations . . . . . . . . . . . 16
87 7.1. Operational Recommendations . . . . . . . . . . . . . . . 16
88 8. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 18
89 9. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 18
90 10. References . . . . . . . . . . . . . . . . . . . . . . . . . . 18
91 10.1. Normative References . . . . . . . . . . . . . . . . . . . 18
92 10.2. Informative References . . . . . . . . . . . . . . . . . . 19
93 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 20
95 1. Introduction
97 The use of Forward Error Correction (FEC) codes is a classic solution
98 to improve the reliability of unicast, multicast and broadcast
99 Content Delivery Protocols (CDP) and applications [RFC3453]. The
100 [RFC6363] document describes a generic framework to use FEC schemes
101 with media delivery applications, and for instance with real-time
102 streaming media applications based on the RTP real-time protocol.
103 Similarly the [RFC5052] document describes a generic framework to use
104 FEC schemes with with objects (e.g., files) delivery applications
105 based on the ALC [RFC5775] and NORM [RFC5740] reliable multicast
106 transport protocols.
108 More specifically, the [RFC5053] (Raptor) and [RFC5170] (LDPC-
109 Staircase and LDPC-Triangle) FEC schemes introduce erasure codes
110 based on sparse parity check matrices for object delivery protocols
111 like ALC and NORM. Similarly, the [RFC5510] document introduces
112 Reed-Solomon codes based on Vandermonde matrices for the same object
113 delivery protocols. All these codes are systematic codes, meaning
114 that the k source symbols are part of the n encoding symbols.
115 Additionally, the Reed-Solomon FEC codes belong to the class of
116 Maximum Distance Separable (MDS) codes that are optimal in terms of
117 erasure recovery capabilities. It means that a receiver can recover
118 the k source symbols from any set of exactly k encoding symbols out
119 of n. This is not the case with either Raptor or LDPC-Staircase
120 codes, and these codes require a certain number of encoding symbols
121 in excess to k. However, this number is small in practice when an
122 appropriate decoding scheme is used at the receiver [Cunche08].
123 Another key difference is the high encoding/decoding complexity of
124 Reed-Solomon codecs compared to Raptor or LDPC-Staircase codes. A
125 difference of one or more orders of magnitude or more in terms of
126 encoding/decoding speed exists between the Reed-Solomon and LDPC-
127 Staircase software codecs [Cunche08][CunchePHD10]. Finally, Raptor
128 and LDPC-Staircase codes are large block FEC codes, in the sense of
129 [RFC3453], since they can efficiently deal with a large number of
130 source symbols.
132 The present document focuses on LDPC-Staircase codes, that belong to
133 the well-known class of "Low Density Parity Check" codes. Because of
134 their key features, these codes are a good solution in many
135 situations, as detailed in Section 7.
137 This documents inherits from [RFC5170] the specifications of the core
138 LDPC-Staircase codes. Therefore this document specifies only the
139 information specific to the FECFRAME context and refers to [RFC5170]
140 for the core specifications of the codes. To that purpose, the
141 present document introduces:
143 o the Fully-Specified FEC Scheme with FEC Encoding ID XXX that
144 specifies a simple way of using LDPC-Staircase codes in order to
145 protect arbitrary ADU flows.
147 Finally, publicly available reference implementations of these codes
148 are available [LDPC-codec] [LDPC-codec-OpenFEC].
150 2. Terminology
152 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
153 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
154 document are to be interpreted as described in RFC 2119 [RFC2119].
156 3. Definitions Notations and Abbreviations
158 3.1. Definitions
160 This document uses the following terms and definitions. Some of them
161 are FEC scheme specific and are in line with [RFC5052]:
162 Source symbol: unit of data used during the encoding process. In
163 this specification, there is always one source symbol per ADU.
164 Encoding symbol: unit of data generated by the encoding process.
165 With systematic codes, source symbols are part of the encoding
166 symbols.
167 Repair symbol: encoding symbol that is not a source symbol.
168 Code rate: the k/n ratio, i.e., the ratio between the number of
169 source symbols and the number of encoding symbols. By definition,
170 the code rate is such that: 0 < code rate <= 1. A code rate close
171 to 1 indicates that a small number of repair symbols have been
172 produced during the encoding process.
173 Systematic code: FEC code in which the source symbols are part of
174 the encoding symbols. The LDPC-Staircase codes introduced in this
175 document are systematic.
176 Source block: a block of k source symbols that are considered
177 together for the encoding.
178 Packet Erasure Channel: a communication path where packets are
179 either dropped (e.g., by a congested router, or because the number
180 of transmission errors exceeds the correction capabilities of the
181 physical layer codes) or received. When a packet is received, it
182 is assumed that this packet is not corrupted.
184 Some of them are FECFRAME framework specific and are in line with
185 [RFC6363]:
187 Application Data Unit (ADU): The unit of source data provided as
188 payload to the transport layer. Depending on the use-case, an ADU
189 may use an RTP encapsulation.
190 (Source) ADU Flow: A sequence of ADUs associated with a transport-
191 layer flow identifier (such as the standard 5-tuple {Source IP
192 address, source port, destination IP address, destination port,
193 transport protocol}). Depending on the use-case, several ADU
194 flows may be protected together by the FECFRAME framework.
195 ADU Block: a set of ADUs that are considered together by the
196 FECFRAME instance for the purpose of the FEC scheme. Along with
197 the F[], L[], and Pad[] fields, they form the set of source
198 symbols over which FEC encoding will be performed.
199 ADU Information (ADUI): a unit of data constituted by the ADU and
200 the associated Flow ID, Length and Padding fields (Section 4.3).
201 This is the unit of data that is used as source symbol.
202 FEC Framework Configuration Information: Information which controls
203 the operation of the FEC Framework. The FFCI enables the
204 synchronization of the FECFRAME sender and receiver instances.
205 FEC Source Packet: At a sender (respectively, at a receiver) a
206 payload submitted to (respectively, received from) the transport
207 protocol containing an ADU along with an optional Explicit Source
208 FEC Payload ID.
209 FEC Repair Packet: At a sender (respectively, at a receiver) a
210 payload submitted to (respectively, received from) the transport
211 protocol containing one repair symbol along with a Repair FEC
212 Payload ID and possibly an RTP header.
214 The above terminology is illustrated in Figure 1 (sender's point of
215 view):
217 +----------------------+
218 | Application |
219 +----------------------+
220 |
221 | (1) Application Data Units (ADUs)
222 |
223 v
224 +----------------------+ +----------------+
225 | FEC Framework | | |
226 | |-------------------------->| FEC Scheme |
227 |(2) Construct source |(3) Source Block | |
228 | blocks | |(4) FEC Encoding|
229 |(6) Construct FEC |<--------------------------| |
230 | source and repair | | |
231 | packets |(5) Explicit Source FEC | |
232 +----------------------+ Payload IDs +----------------+
233 | Repair FEC Payload IDs
234 | Repair symbols
235 |
236 |(7) FEC source and repair packets
237 v
238 +----------------------+
239 | Transport Layer |
240 | (e.g., UDP) |
241 +----------------------+
243 Figure 1: Terminology used in this document (sender).
245 3.2. Notations
247 This document uses the following notations: Some of them are FEC
248 scheme specific:
249 k denotes the number of source symbols in a source block.
250 max_k denotes the maximum number of source symbols for any source
251 block.
252 n denotes the number of encoding symbols generated for a source
253 block.
254 E denotes the encoding symbol length in bytes.
255 CR denotes the "code rate", i.e., the k/n ratio.
256 N1 denotes the target number of "1s" per column in the left side
257 of the parity check matrix.
258 N1m3 denotes the value N1 - 3.
259 a^^b denotes a raised to the power b.
261 Some of them are FECFRAME framework specific:
263 B denotes the number of ADUs per ADU block.
264 max_B denotes the maximum number of ADUs for any ADU block.
266 3.3. Abbreviations
268 This document uses the following abbreviations:
269 ADU stands for Application Data Unit.
270 ESI stands for Encoding Symbol ID.
271 FEC stands for Forward Error (or Erasure) Correction code.
272 FFCI stands for FEC Framework Configuration Information.
273 FSSI stands for FEC Scheme Specific Information.
274 LDPC stands for Low Density Parity Check.
275 MDS stands for Maximum Distance Separable code.
277 4. Common Procedures Related to the ADU Block and Source Block Creation
279 This section introduces the procedures that are used during the ADU
280 block and the related source block creation, for the FEC scheme
281 considered.
283 4.1. Restrictions
285 This specification has the following restrictions:
286 o there MUST be exactly one source symbol per ADUI, and therefore
287 per ADU;
288 o there MUST be exactly one repair symbol per FEC Repair Packet;
289 o there MUST be exactly one source block per ADU block;
290 o the use of the LDPC-Staircase scheme is such that there MUST be
291 exactly one encoding symbol per group, i.e., G MUST be equal to 1
292 [RFC5170];
294 4.2. ADU Block Creation
296 Two kinds of limitations MUST be considered, that impact the ADU
297 block creation:
298 o at the FEC Scheme level: the FEC Scheme and the FEC codec have
299 limitations that define a maximum source block size;
300 o at the FECFRAME instance level: the target use-case MAY have real-
301 time constraints that MAY define a maximum ADU block size;
302 Note that terminology "maximum source block size" and "maximum ADU
303 block size" depends on the point of view that is adopted (FEC Scheme
304 versus FECFRAME instance). However, in this document, both refer to
305 the same value since Section 4.1 requires there is exactly one source
306 symbol per ADU. We now detail each of these aspects.
308 The maximum source block size in symbols, max_k, depends on several
309 parameters: the code rate (CR), the Encoding Symbol ID (ESI) field
310 length in the Explicit Source/Repair FEC Payload ID (16 bits), as
311 well as possible internal codec limitations. More specifically,
312 max_k cannot be larger than the following values, derived from the
313 ESI field size limitation, for a given code rate:
314 max1_k = 2^^(16 - ceil(Log2(1/CR)))
315 Some common max1_k values are:
316 o CR == 1 (no repair symbol): max1_k = 2^^16 = 65536 symbols
317 o 1/2 <= CR < 1: max1_k = 2^^15 = 32,768 symbols
318 o 1/4 <= CR < 1/2: max1_k = 2^^14 = 16,384 symbols
320 Additionally, a codec MAY impose other limitations on the maximum
321 source block size, for instance, because of a limited working memory
322 size. This decision MUST be clarified at implementation time, when
323 the target use-case is known. This results in a max2_k limitation.
325 Then, max_k is given by:
326 max_k = min(max1_k, max2_k)
327 Note that this calculation is only required at the encoder (sender),
328 since the actual k parameter (k <= max_k) is communicated to the
329 decoder (receiver) through the Explicit Source/Repair FEC Payload ID.
331 The source ADU flows MAY have real-time constraints. In that case
332 the maximum number of ADUs of an ADU block must not exceed a certain
333 threshold since it directly impacts the decoding delay. The larger
334 the ADU block size, the longer a decoder may have to wait until it
335 has received a sufficient number of encoding symbols for decoding to
336 succeed, and therefore the larger the decoding delay. When the
337 target use-case is known, these real-time constraints result in an
338 upper bound to the ADU block size, max_rt.
340 For instance, if the use-case specifies a maximum decoding latency,
341 l, and if each source ADU covers a duration d of a continuous media
342 (we assume here the simple case of a constant bit rate ADU flow),
343 then the ADU block size must not exceed:
344 max_rt = floor(l / d)
345 After encoding, this block will produce a set of at most n = max_rt /
346 CR encoding symbols. These n encoding symbols will have to be sent
347 at a rate of n / l packets per second. For instance, with d = 10 ms,
348 l = 1 s, max_rt = 100 ADUs.
350 If we take into account all these constraints, we find:
351 max_B = min(max_k, max_rt)
352 This max_B parameter is an upper bound to the number of ADUs that can
353 constitute an ADU block.
355 4.3. Source Block Creation
357 In its most general form the FECFRAME framework and the LDPC-
358 Staircase FEC scheme are meant to protect a set of independent flows.
359 Since the flows have no relationship to one another, the ADU size of
360 each flow can potentially vary significantly. Even in the special
361 case of a single flow, the ADU sizes can largely vary (e.g., the
362 various frames of a "Group of Pictures (GOP) of an H.264 flow). This
363 diversity must be addressed since the LDPC-Staircase FEC scheme
364 requires a constant encoding symbol size (E parameter) per source
365 block. Since this specification requires that there is only one
366 source symbol per ADU, E must be large enough to contain all the ADUs
367 of an ADU block along with their prepended 3 bytes (see below).
369 In situations where E is determined per source block (default,
370 specified by the FFCI/FSSI with S = 0, Section 5.1.1.2), E is equal
371 to the size of the largest ADU of this source block plus three (for
372 the prepended 3 bytes, see below). In this case, upon receiving the
373 first FEC Repair Packet for this source block, since this packet MUST
374 contain a single repair symbol (Section 5.1.3), a receiver determines
375 the E parameter used for this source block.
377 In situations where E is fixed (specified by the FFCI/FSSI with S =
378 1, Section 5.1.1.2), then E must be greater or equal to the size of
379 the largest ADU of this source block plus three (for the prepended 3
380 bytes, see below). If this is not the case, an error is returned.
381 How to handle this error is use-case specific (e.g., a larger E
382 parameter may be communicated to the receivers in an updated FFCI
383 message, using an appropriate mechanism) and is not considered by
384 this specification.
386 The ADU block is always encoded as a single source block. There are
387 a total of B <= max_B ADUs in this ADU block. For the ADU i, with 0
388 <= i <= B-1, 3 bytes are prepended (Figure 2):
389 o The first byte, FID[i] (Flow ID), contains the integer identifier
390 associated to the source ADU flow to which this ADU belongs to.
391 It is assumed that a single byte is sufficient, or said
392 differently, that no more than 256 flows will be protected by a
393 single instance of the FECFRAME framework.
394 o The following two bytes, L[i] (Length), contain the length of this
395 ADU, in network byte order (i.e., big endian). This length is for
396 the ADU itself and does not include the FID[i], L[i], or Pad[i]
397 fields.
399 Then zero padding is added to ADU i (if needed) in field Pad[i], for
400 alignment purposes up to a size of exactly E bytes. The data unit
401 resulting from the ADU i and the F[i], L[i] and Pad[i] fields, is
402 called ADU Information (or ADUI). Each ADUI contributes to exactly
403 one source symbol to the source block.
405 Encoding Symbol Length (E)
406 < -------------------------------------------------------------- >
407 +----+----+-----------------------+------------------------------+
408 |F[0]|L[0]| ADU[0] | Pad[0] |
409 +----+----+----------+------------+------------------------------+
410 |F[1]|L[1]| ADU[1] | Pad[1] |
411 +----+----+----------+-------------------------------------------+
412 |F[2]|L[2]| ADU[2] |
413 +----+----+------+-----------------------------------------------+
414 |F[3]|L[3]|ADU[3]| Pad[3] |
415 +----+----+------+-----------------------------------------------+
416 \_______________________________ _______________________________/
417 \/
418 simple FEC encoding
420 +----------------------------------------------------------------+
421 | Repair 4 |
422 +----------------------------------------------------------------+
423 . .
424 . .
425 +----------------------------------------------------------------+
426 | Repair 7 |
427 +----------------------------------------------------------------+
429 Figure 2: Source block creation, for code rate 1/2 (equal number of
430 source and repair symbols, 4 in this example), and S = 0.
432 Note that neither the initial 3 bytes nor the optional padding are
433 sent over the network. However, they are considered during FEC
434 encoding. It means that a receiver who lost a certain FEC source
435 packet (e.g., the UDP datagram containing this FEC source packet)
436 will be able to recover the ADUI if FEC decoding succeeds. Thanks to
437 the initial 3 bytes, this receiver will get rid of the padding (if
438 any) and identify the corresponding ADU flow.
440 5. LDPC-Staircase FEC Scheme for Arbitrary ADU Flows
442 5.1. Formats and Codes
444 5.1.1. FEC Framework Configuration Information
446 The FEC Framework Configuration Information (or FFCI) includes
447 information that MUST be communicated between the sender and
448 receiver(s). More specifically, it enables the synchronization of
449 the FECFRAME sender and receiver instances. It includes both
450 mandatory elements and scheme-specific elements, as detailed below.
452 5.1.1.1. Mandatory Information
454 FEC Encoding ID: the value assigned to this fully-specified FEC
455 scheme MUST be XXX, as assigned by IANA (Section 8).
456 When SDP is used to communicate the FFCI, this FEC Encoding ID is
457 carried in the 'encoding-id' parameter.
459 5.1.1.2. FEC Scheme-Specific Information
461 The FEC Scheme Specific Information (FSSI) includes elements that are
462 specific to the present FEC scheme. More precisely:
463 PRNG seed (seed): a non-negative 32 bit integer used as the seed of
464 the Pseudo Random Number Generator, as defined in [RFC5170].
465 Encoding symbol length (E): a non-negative integer that indicates
466 either the length of each encoding symbol in bytes (strict mode,
467 i.e., if S = 1), or the maximum length of any encoding symbol
468 (i.e., if S = 0).
469 Strict (S) flag: when set to 1 this flag indicates that the E
470 parameter is the actual encoding symbol length value for each
471 block of the session (unless otherwise notified by an updated FFCI
472 if this possibility is considered by the use-case or CDP). When
473 set to 0 this flag indicates that the E parameter is the maximum
474 encoding symbol length value for each block of the session (unless
475 otherwise notified by an updated FFCI if this possibility is
476 considered by the use-case or CDP).
477 N1 minus 3 (n1m3): an integer between 0 (default) and 7, inclusive.
478 The number of "1s" per column in the left side of the parity check
479 matrix, N1, is then equal to N1m3 + 3, as specified in [RFC5170].
480 These elements are required both by the sender (LDPC-Staircase
481 encoder) and the receiver(s) (LDPC-Staircase decoder).
483 When SDP is used to communicate the FFCI, this FEC scheme-specific
484 information is carried in the 'fssi' parameter in textual
485 representation as specified in [RFC6364]. For instance:
487 fssi=seed:1234,E:1400,S:0,n1m3:0
489 If another mechanism requires the FSSI to be carried as an opaque
490 octet string (for instance after a Base64 encoding), the encoding
491 format consists of the following 7 octets:
492 o PRNG seed (seed): 32 bit field.
493 o Encoding symbol length (E): 16 bit field.
494 o Strict (S) flag: 1 bit field.
495 o Reserved: a 4 bit field that MUST be set to zero.
497 o N1m3 parameter (n1m3): 3 bit field.
499 0 1 2
500 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3
501 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
502 | PRNG seed (seed) |
503 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
504 | Encoding Symbol Length (E) |S| resvd | n1m3|
505 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
507 Figure 3: FSSI encoding format.
509 5.1.2. Explicit Source FEC Payload ID
511 A FEC source packet MUST contain an Explicit Source FEC Payload ID
512 that is appended to the end of the packet as illustrated in Figure 4.
514 +--------------------------------+
515 | IP Header |
516 +--------------------------------+
517 | Transport Header |
518 +--------------------------------+
519 | ADU |
520 +--------------------------------+
521 | Explicit Source FEC Payload ID |
522 +--------------------------------+
524 Figure 4: Structure of a FEC Source Packet with the Explicit Source
525 FEC Payload ID.
527 More precisely, the Explicit Source FEC Payload ID is composed of the
528 following fields (Figure 5):
529 Source Block Number (SBN) (16 bit field): this field identifies the
530 source block to which this FEC source packet belongs.
531 Encoding Symbol ID (ESI) (16 bit field): this field identifies the
532 source symbol contained in this FEC source packet. This value is
533 such that 0 <= ESI <= k - 1 for source symbols.
534 Source Block Length (k) (16 bit field): this field provides the
535 number of source symbols for this source block, i.e., the k
536 parameter.
538 0 1 2 3
539 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
540 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
541 | Source Block Number (SBN) | Encoding Symbol ID (ESI) |
542 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
543 | Source Block Length (k) |
544 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
546 Figure 5: Source FEC Payload ID encoding format.
548 5.1.3. Repair FEC Payload ID
550 A FEC repair packet MUST contain a Repair FEC Payload ID that is
551 prepended to the repair symbol(s) as illustrated in Figure 6. There
552 MUST be a single repair symbol per FEC repair packet.
554 +--------------------------------+
555 | IP Header |
556 +--------------------------------+
557 | Transport Header |
558 +--------------------------------+
559 | Repair FEC Payload ID |
560 +--------------------------------+
561 | Repair Symbol |
562 +--------------------------------+
564 Figure 6: Structure of a FEC Repair Packet with the Repair FEC
565 Payload ID.
567 More precisely, the Repair FEC Payload ID is composed of the
568 following fields: (Figure 7):
569 Source Block Number (SBN) (16 bit field): this field identifies the
570 source block to which the FEC repair packet belongs.
571 Encoding Symbol ID (ESI) (16 bit field) this field identifies the
572 repair symbol contained in this FEC repair packet. This value is
573 such that k <= ESI <= n - 1 for repair symbols.
574 Source Block Length (k) (16 bit field): this field provides the
575 number of source symbols for this source block, i.e., the k
576 parameter.
577 Number of Encoding Symbols (n) (16 bit field): this field provides
578 the number of encoding symbols for this source block, i.e., the n
579 parameter.
581 0 1 2 3
582 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
583 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
584 | Source Block Number (SBN) | Encoding Symbol ID (ESI) |
585 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
586 | Source Block Length (k) | Number Encoding Symbols (n) |
587 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
589 Figure 7: Repair FEC Payload ID encoding format.
591 5.2. Procedures
593 The following procedures apply:
594 o The source block creation procedures are specified in Section 4.3.
595 o The SBN value is incremented for each new source block, starting
596 at 0 for the first block of the ADU flow. Wrapping to zero will
597 happen for long sessions, after value 2^^16 - 1.
598 o The ESI of encoding symbols is managed sequentially, starting at 0
599 for the first symbol. The first k values (0 <= ESI <= k - 1)
600 identify source symbols, whereas the last n-k values (k <= ESI <=
601 n - 1) identify repair symbols.
602 o The FEC repair packet creation procedures are specified in
603 Section 5.1.3.
605 5.3. FEC Code Specification
607 The present document inherits from [RFC5170] the specification of the
608 core LDPC-Staircase codes for a packet erasure transmission channel.
610 Because of the requirement to have exactly one encoding symbol per
611 group, i.e., because G MUST be equal to 1 (Section 4.1), several
612 parts of [RFC5170] are useless. In particular, this is the case of
613 Section 5.6. "Identifying the G Symbols of an Encoding Symbol
614 Group".
616 6. Security Considerations
618 The FEC Framework document [RFC6363] provides a comprehensive
619 analysis of security considerations applicable to FEC schemes.
620 Therefore the present section follows the security considerations
621 section of [RFC6363] and only discusses topics that are specific to
622 the use of LDPC-Staircase codes.
624 6.1. Attacks Against the Data Flow
625 6.1.1. Access to Confidential Content
627 The LDPC-Staircase FEC Scheme specified in this document does not
628 change the recommendations of [RFC6363]. To summarize, if
629 confidentiality is a concern, it is RECOMMENDED that one of the
630 solutions mentioned in [RFC6363] is used, with special considerations
631 to the way this solution is applied (e.g., before versus after FEC
632 protection, and within the end-system versus in a middlebox), to the
633 operational constraints (e.g., performing FEC decoding in a protected
634 environment may be complicated or even impossible) and to the threat
635 model.
637 6.1.2. Content Corruption
639 The LDPC-Staircase FEC Scheme specified in this document does not
640 change the recommendations of [RFC6363]. To summarize, it is
641 RECOMMENDED that one of the solutions mentioned in [RFC6363] is used
642 on both the FEC Source and Repair Packets.
644 6.2. Attacks Against the FEC Parameters
646 The FEC Scheme specified in this document defines parameters that can
647 be the basis of several attacks. More specifically, the following
648 parameters of the FFCI may be modified by an attacker
649 (Section 5.1.1.2):
650 o FEC Encoding ID: changing this parameter leads the receiver to
651 consider a different FEC Scheme, which enables an attacker to
652 create a Denial of Service (DoS).
653 o Encoding symbol length (E): setting this E parameter to a value
654 smaller than the valid one enables an attacker to create a DoS
655 since the repair symbols and certain source symbols will be larger
656 than E, which is an incoherency for the receiver. Setting this E
657 parameter to a value larger than the valid one has similar impacts
658 when S=1 since the received repair symbol size will be smaller
659 than expected. On the opposite it will not lead to any
660 incoherency when S=0 since the actual symbol length value for the
661 block is determined by the size of any received repair symbol, as
662 long as this value is smaller than E. However setting this E
663 parameter to a larger value may have impacts on receivers that
664 pre-allocate memory space in advance to store incoming symbols.
665 o Strict (S) flag: flipping this S flag from 0 to 1 (i.e., E is now
666 considered as a strict value) enables an attacker to mislead the
667 receiver if the actual symbol size varies over different source
668 blocks. Flipping this S flag from 1 to 0 has no major
669 consequences unless the receiver requires to have a fixed E value
670 (e.g., because the receiver pre-allocates memory space).
672 o N1 minus 3 (n1m3): changing this parameter leads the receiver to
673 consider a different code, which enables an attacker to create a
674 DoS.
676 It is therefore RECOMMENDED that security measures are taken to
677 guarantee the FFCI integrity, as specified in [RFC6363]. How to
678 achieve this depends on the way the FFCI is communicated from the
679 sender to the receiver, which is not specified in this document.
681 Similarly, attacks are possible against the Explicit Source FEC
682 Payload ID and Repair FEC Payload ID: by modifying the Source Block
683 Number (SBN), or the Encoding Symbol ID (ESI), or the Source Block
684 Length (k), or the Number Encoding Symbols (n), an attacker can
685 easily corrupt the block identified by the SBN. Other consequences,
686 that are use-case and/or CDP dependant, may also happen. It is
687 therefore RECOMMENDED that security measures are taken to guarantee
688 the FEC Source and Repair Packets as stated in [RFC6363].
690 6.3. When Several Source Flows are to be Protected Together
692 The LDPC-Staircase FEC Scheme specified in this document does not
693 change the recommendations of [RFC6363].
695 6.4. Baseline Secure FEC Framework Operation
697 The LDPC-Staircase FEC Scheme specified in this document does not
698 change the recommendations of [RFC6363] concerning the use of the
699 IPsec/ESP security protocol as a mandatory to implement (but not
700 mandatory to use) security scheme. This is well suited to situations
701 where the only insecure domain is the one over which the FEC
702 Framework operates.
704 7. Operations and Management Considerations
706 The FEC Framework document [RFC6363] provides a comprehensive
707 analysis of operations and management considerations applicable to
708 FEC schemes. Therefore the present section only discusses topics
709 that are specific to the use of LDPC-Staircase codes as specified in
710 this document.
712 7.1. Operational Recommendations
714 LDPC-Staircase codes have excellent erasure recovery capabilities
715 with large source blocks, close to ideal MDS codes. For instance,
716 independently of FECFRAME, with source block size k=1024, CR=2/3,
717 N1=5, G=1, with a hybrid ITerative/Maximum Likelihood (IT/ML)
718 decoding approach (see below) and when all symbols are sent in a
719 random order (see below), the average overhead amounts to 0.64%
720 (corresponding to 6.5 symbols in addition to k) and receiving 1046
721 symbols (corresponding to a 2.1% overhead) is sufficient to reduce
722 the decoding failure probability to 5.9*10^^-5. This is why these
723 codes are a good solution to protect a single high bitrate source
724 flow as in [Matsuzono10], or to protect globally several mid-rate
725 source flows within a single FECFRAME instance: in both cases the
726 source block size can be assumed to be equal to a few hundreds (or
727 more) source symbols.
729 LDPC-Staircase codes are also a good solution whenever processing
730 requirements at a software encoder or decoder must be kept to a
731 minimum. This is true when the decoder uses an IT decoding
732 algorithm, or an ML algorithm (we use a Gaussian Elimination as the
733 ML algorithm) when this latter is carefully implemented and the
734 source block size kept reasonable, or a mixture of both techniques
735 which is the recommended solution [Cunche08][CunchePHD10]. For
736 instance an average decoding speed between 1.3 Gbps (corresponding to
737 a very bad channel, close to the theoretical decoding limit and
738 requiring an ML decoding) and 4.3 Gbps (corresponding to a medium
739 quality channel where IT decoding is sufficient) are easily achieved
740 with a source block size composed of k=1024 source symbols, a code
741 rate CR=2/3 (i.e., 512 repair symbols), 1024 byte long symbols, G=1,
742 and N1=5, on an Intel Xeon 5120/1.86GHz workstation running Linux/64
743 bits. Additionally, with a hybrid IT/ML approach, a receiver can
744 decide if and when ML decoding is used, depending on local criteria
745 (e.g., battery or CPU capabilities), independently from other
746 receivers.
748 As the source block size decreases, the erasure recovery capabilities
749 of LDPC codes in general also decrease. In the case of LDPC-
750 Staircase codes, in order to compensate this phenomenon, it is
751 recommended to increase the N1 parameter (e.g., experiments carried
752 out in [Matsuzono10] use N1=7 if k=170 symbols, and N1=5 otherwise)
753 and to use a hybrid IT/ML decoding approach. For instance,
754 independently of FECFRAME, with a small source block size k=256
755 symbols, CR=2/3, N1=7, and G=1, 8he average overhead amounts to 0.71%
756 (corresponding to 1.8 symbols in addition to k), and receiving 271
757 symbols (corresponding to a 5.9% overhead) is sufficient to reduce
758 the decoding failure probability to 5.9*10^^-5. Using N1=9 or 10
759 further improves these results if need be, which also enables to use
760 LDPC-Staircase codes with k=100 symbols for instance.
762 With very small source blocks (e.g., a few tens symbols), using for
763 instance Reed-Solomon codes [SIMPLE_RS] or 2D parity check codes MAY
764 be more appropriate.
766 The way the FEC Repair Packets are transmitted is of high importance.
768 A good strategy, that works well for any kind of channel loss model,
769 consists in sending FEC Repair Packets in random order (rather than
770 in sequence) while FEC Source Packets are sent first and in sequence.
771 Sending all packets in a random order is another possibility, but it
772 requires that all repair symbols for a source block be produced
773 first, which adds some extra delay at a sender.
775 8. IANA Considerations
777 Values of FEC Encoding IDs are subject to IANA registration.
778 [RFC6363] defines general guidelines on IANA considerations. In
779 particular it defines a registry called FEC Framework (FECFRAME) FEC
780 Encoding IDs whose values are granted on an IETF Consensus basis.
782 This document registers one value in the FEC Framework (FECFRAME) FEC
783 Encoding IDs registry as follows:
784 o XXX refers to the Simple LDPC-Staircase [RFC5170] FEC Scheme for
785 Arbitrary Packet Flows.
787 9. Acknowledgments
789 The authors want to thank K. Matsuzono, J. Detchart and H. Asaeda for
790 their contributions in evaluating the use of LDPC-Staircase codes in
791 the context of FECFRAME [Matsuzono10].
793 10. References
795 10.1. Normative References
797 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate
798 Requirement Levels", RFC 2119.
800 [RFC5170] Roca, V., Neumann, C., and D. Furodet, "Low Density Parity
801 Check (LDPC) Forward Error Correction", RFC 5170,
802 June 2008.
804 [RFC6363] Watson, M., Begen, A., and V. Roca, "Forward Error
805 Correction (FEC) Framework", RFC 6363, September 2011.
807 [RFC6364] Begen, A., "Session Description Protocol Elements for the
808 Forward Error Correction (FEC) Framework", RFC 6364,
809 October 2011.
811 10.2. Informative References
813 [RFC3453] Luby, M., Vicisano, L., Gemmell, J., Rizzo, L., Handley,
814 M., and J. Crowcroft, "The Use of Forward Error Correction
815 (FEC) in Reliable Multicast", RFC 3453, December 2002.
817 [RFC5052] Watson, M., Luby, M., and L. Vicisano, "Forward Error
818 Correction (FEC) Building Block", RFC 5052, August 2007.
820 [RFC5510] Lacan, J., Roca, V., Peltotalo, J., and S. Peltotalo,
821 "Reed-Solomon Forward Error Correction (FEC) Schemes",
822 RFC 5510, April 2009.
824 [SIMPLE_RS]
825 Roca, V., Cunche, M., Lacan, J., Bouabdallah, A., and K.
826 Matsuzono, "Simple Reed-Solomon Forward Error Correction
827 (FEC) Scheme for FECFRAME",
828 draft-ietf-fecframe-simple-rs-01 (Work in Progress),
829 September 2011.
831 [RFC5053] Luby, M., Shokrollahi, A., Watson, M., and T. Stockhammer,
832 "Raptor Forward Error Correction Scheme", RFC 5053,
833 June 2007.
835 [RFC5740] Adamson, B., Bormann, C., Handley, M., and J. Macker,
836 "NACK-Oriented Reliable Multicast (NORM) Transport
837 Protocol", RFC 5740, November 2009.
839 [RFC5775] Luby, M., Watson, M., and L. Vicisano, "Asynchronous
840 Layered Coding (ALC) Protocol Instantiation", RFC 5775,
841 April 2010.
843 [Cunche08]
844 Cunche, M. and V. Roca, "Optimizing the Error Recovery
845 Capabilities of LDPC-Staircase Codes Featuring a Gaussian
846 Elimination Decoding Scheme", 10th IEEE International
847 Workshop on Signal Processing for Space Communications
848 (SPSC'08), October 2008.
850 [CunchePHD10]
851 Cunche, M., "High performances AL-FEC codes for the
852 erasure channel : variation around LDPC codes", PhD
853 dissertation (in
854 French) (http://tel.archives-ouvertes.fr/tel-
855 00451336/en/), June 2010.
857 [Matsuzono10]
858 Matsuzono, K., Detchart, J., Cunche, M., Roca, V., and H.
860 Asaeda, "Performance Analysis of a High-Performance Real-
861 Time Application with Several AL-FEC Schemes", 35th Annual
862 IEEE Conference on Local Computer Networks (LCN 2010),
863 October 2010.
865 [LDPC-codec]
866 Cunche, M., Roca, V., Neumann, C., and J. Laboure, "LDPC-
867 Staircase/LDPC-Triangle Codec Reference Implementation",
868 INRIA Rhone-Alpes and STMicroelectronics,
869 .
871 [LDPC-codec-OpenFEC]
872 "The OpenFEC project", .
874 Authors' Addresses
876 Vincent Roca
877 INRIA
878 655, av. de l'Europe
879 Inovallee; Montbonnot
880 ST ISMIER cedex 38334
881 France
883 Email: vincent.roca@inria.fr
884 URI: http://planete.inrialpes.fr/people/roca/
886 Mathieu Cunche
887 NICTA
888 Australia
890 Email: mathieu.cunche@nicta.com.au
891 URI: http://mathieu.cunche.free.fr/
893 Jerome Lacan
894 ISAE/LAAS-CNRS
895 1, place Emile Blouin
896 Toulouse 31056
897 France
899 Email: jerome.lacan@isae.fr
900 URI: http://dmi.ensica.fr/auteur.php3?id_auteur=5