Internet Draft Adam Li draft-ietf-avt-ulp-13.txt Editor November 28, 2005 Expires: May 28, 2006 An RTP Payload Format for Generic FEC STATUS OF THIS MEMO By submitting this Internet-Draft, each author represents that any applicable patent or other IPR claims of which he or she is aware have been or will be disclosed, and any of which he or she becomes aware will be disclosed, in accordance with Section 6 of BCP 79. Internet-Drafts are working documents of the Internet Engineering Task Force (IETF), its areas, and its working groups. Note that other groups may also distribute working documents as Internet- Drafts. Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at anytime. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as work in progress. The list of current Internet-Drafts can be accessed at http://www.ietf.org/ietf/1id-abstracts.txt The list of Internet-Draft Shadow Directories can be accessed at http://www.ietf.org/shadow.html. This document is a submission of the IETF AVT WG. Comments should be directed to the AVT WG mailing list, avt@ietf.org. ABSTRACT This document specifies a payload format for generic Forward Error Correction (FEC) for media data encapsulated in RTP. It is based on the exclusive-or (parity) operation, and it is a generalized algorithms that includes Uneven Level Protection (ULP). The payload format described in this draft allows end systems to apply protection using arbitrary protection lengths and levels, in addition to using arbitrary protection group sizes. It enables complete recovery or partial recovery of the critical payload and RTP header fields depending on the packet loss situation. This scheme is completely backward compatible with non-FEC capable hosts. Those receivers that do not know Adam H. Li [Page 1] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 about FEC can simply ignore the protection data. This specification obsoletes RFC 2733 and RFC 3009. Adam H. Li [Page 2] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 Table of Contents 1. Introduction .................................................. 4 2. Terminology ................................................... 6 3. Basic Operation ............................................... 6 4. Parity Codes .................................................. 7 5. Uneven Level Protection ....................................... 8 6. RTP Media Packet Structure .................................... 9 7. FEC Packet Structure .......................................... 9 7.1. Baseline Mode FEC ........................................... 9 7.2. Extended Mode FEC .......................................... 10 7.3. RTP Header of FEC Packets .................................. 10 7.4. FEC Header of FEC packets .................................. 11 7.5. ULP Level Header of FEC Packets ............................ 12 8. Protection Operation ......................................... 15 8.1. Protection Operation in the Baseline Mode .................. 15 8.2. Protection Operation in the Extended Mode .................. 16 8.2.1. Extended Mode - Protection Level 0 ....................... 16 8.2.2. Extended Mode - Protection Level 1 and Higher ............ 16 9. Recovery Procedure ........................................... 16 9.1. Reconstruction in the Baseline Mode ........................ 17 9.2. Reconstruction in the Extended Mode ........................ 18 9.2.1. Extended Mode - Reconstruction of Level 0 ................ 18 9.2.2. Extended Mode - Reconstruction of Level 1 and Higher ..... 18 10. Examples .................................................... 19 10.1. A Baseline Mode Example ................................... 19 10.2. An Extended Mode Example With one Protection Level ........ 21 10.3. An Extended Mode Example With Two Protection Levels ....... 23 10.4. An Extended Mode Example With FEC As Redundant Coding ..... 26 11. Security .................................................... 29 12. Congestion Considerations ................................... 30 13. MIME Registrations .......................................... 31 13.1. Registration of audio/ulpfec .............................. 31 13.2. Registration of video/ulpfec .............................. 32 13.3. Registration of text/ulpfec ............................... 33 13.4. Registration of application/ulpfec ........................ 34 14. Multiplexing of FEC ......................................... 35 14.1. FEC as a Separate Stream .................................. 35 14.2. FEC as Redundant Encoding ................................. 36 15. Indication FEC Usage in SDP ................................. 36 15.1. FEC as a Separate Stream .................................. 36 15.2. FEC as Redundant Encoding ................................. 37 15.3. Offer / Answer Consideration .............................. 38 16. Application Statement ....................................... 39 17. Acknowledgements ............................................ 40 18. Bibliography ................................................ 40 18.1. Normative References ...................................... 40 18.2. Informative References .................................... 41 19. Author's Address ............................................ 42 Copyright Statement ............................................. 43 Disclaimer of Validity .......................................... 43 Adam H. Li [Page 3] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 1. Introduction Because of the real-time nature of many applications, they usually have more stringent delay requirements than normal data transmissions. As a result, retransmission of the lost packets is generally not a valid option for such applications. In these cases, a better method to attempt recovery of information from packet loss is through Forward Error Correction (FEC). FEC is one of the main methods used to protect against packet loss over packet switched networks [9, 10]. In particular, the use of traditional error correcting codes, such as parity, Reed-Solomon, and Hamming codes, has attracted attention. To apply these mechanisms, protocol support is required. RFC 2733 [9] and RFC 3009 [11] defined one of such FEC protocols. However, in those two RFCs a few fields (the P, X, and CC fields) in the RTP header are specified in ways which are not consistent as they are designed in RTP [1]. This prevents the payload-independent validity check of the RTP packets. This document extends the FEC defined in RFC 2733 and RFC 3009 to include unequal error protection on the payload data. It specifies a more general algorithm, which includes the algorithm of the two previous RFCs as its special cases. This specification also fix the above-mentioned inconsistency with RFC 2733 and RFC 3009, and will obsolete those two previous RFCs. This document defines a payload format for RTP [1] which allows for generic forward error correction of real time media. In this context, generic means that the FEC protocol is (1) independent of the nature of the media being protected, be it audio, video, or otherwise, (2) flexible enough to support a wide variety of FEC configurations, (3) designed for adaptivity so that the FEC technique can be modified easily without out of band signaling, and (4) supportive of a number of different mechanisms for transporting the FEC packets. Furthermore, in many cases the bandwidth of the network connections is a very limited resource. On the other hand, most of traditional FEC schemes are not designed for optimal utilization of the limited bandwidth resource. A more efficient way to utilize the limited bandwidth would be to use unequal error protection to provide different levels of protection for different parts of the data stream which vary in importance. The unequal error protection schemes can usually make more efficient use of the bandwidth to provide better overall protection of the data stream against the loss. Proper protocol support is essential for realizing these unequal error protection mechanisms. However, the application of most of the unequal error protection schemes requires the knowledge of the importance for different parts of the data stream. Most of such schemes are designed for a particular type of media according to the structure of the media protected, and as a result, are not generic. Adam H. Li [Page 4] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 This document defines an extended mode FEC algorithm and protocol that allows for generic forward error correction with unequal error protection for real-time media. It is called the Uneven Level Protection (ULP). The payload data are protected by one or more protection levels. Lower protection levels provide greater protection by using smaller group sizes (compared to higher protection levels) for generating the FEC packet. As we will discuss below, audio/video applications would generally benefit from an unequal error protection scheme that gives more protection to the beginning part of each packet. So in the ULP algorithm, the data that are closer to the beginning of the packet are protected by lower protection levels because these data are in general more important, and they tend to carry more information than the data further behind in the packet. In many multimedia streams, the more important parts of the data are always at the beginning of the data packet. This is the common practice for most codecs since the beginning of the packet is closer to the re-synchronization marker at the header and thus is more likely to be correctly decoded. In additional, almost all media formats have the frame headers at the beginning of the packet, which is the most vital part of the packet. For video streams, most modern formats have optional data partitioning modes to improve error resilience in which the video macroblock header data, the motion vector data, and DCT coefficient data are separated into their individual partitions. In ITU-T H.263 version 3, there is the optional data partitioned syntax of Annex V. In MPEG-4 Visual Simple Profile, there is the optional data partitioning mode. When these modes are enabled, the video macroblock (MB) header and motion vector partitions (which are much more important to the quality of the video reconstruction) are transmitted in the partition(s) at the beginning of the video packet while residue DCT coefficient partitions (which are less important) are transmitted in the partition close to the end of the packet. Because the data is arranged in descending order of importance, it would be beneficial to provide more protection to the beginning part of the packet in transmission. For audio streams, the bitstreams generated by many of the new audio codecs also contain data with different classes of importance. These different classes are then transmitted in order of descending importance. Thus, applying more protection to the beginning of the packet would also be beneficial in these cases. Even for uniform- significance audio streams, special stretching techniques can be applied to the partially recovered audio data packets. In cases where audio redundancy coding is used, more protection should be applied to the original data located in the first half of the packet. The rest of the packet containing the redundant copies of the data does not need the same level of protection. Adam H. Li [Page 5] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 It is clear that audio/video applications would generally benefit from the FEC algorithms specified in this document. And with the extended mode ULP FEC, more efficient protection of the media payload can be potentially achieved. This document specifies the protocol and algorithm for applying the generic FEC to the RTP media payloads. 2. Terminology The following terms are used throughout this document: Media Payload: The raw, un-protected user data that are transmitted from the sender. The media payload is placed inside of an RTP packet. Media Header: The RTP header for the packet containing the media payload. Media Packet: The combination of a media payload and media header is called a media packet. FEC Packet: The FEC algorithms at the transmitter take the media packets as an input. They output both the media packets that they are passed, and newly generated packets called FEC packets, which contain redundant media data used for error correction. The FEC packets are formatted according to the rules specified in this document. FEC Header: The header information contained in an FEC packet. FEC Payload: The payload of an FEC packet. Associated: A FEC packet is said to be "associated" with one or more media packets (or vice versa) when those media packets are used to generate the FEC packet (by use of the exclusive-or operation). In case of extended mode FEC, this refers to only those packets used to generate the Level 0 FEC payload, if not explicitly stated otherwise. The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in RFC 2119 [2]. 3. Basic Operation The payload format described here is used whenever the sender in an RTP session would like to protect the media stream it is sending with generic parity FEC. The FEC supported by this format is based on the simple exclusive-or (XOR) parities operation. The sender takes the packets from the media stream requiring protection and Adam H. Li [Page 6] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 determines the operation needed, and in case of the extended mode, also the protection levels for these packets and the protection length for each level. The data are grouped as described below in Section 7. An XOR operation is applied across the payload to generate the FEC information. The result based on the procedures defined here is an RTP packet containing FEC information. This packet can be used at the receiver to recover the packets or parts of the packets used to generate the FEC packet. By using unequal error protection provided by the extended mode, this scheme can make more efficient use of the channel bandwidth, and provide more efficient error resilience for transmission over error prone channels. The payload format contains information that allows the sender to tell the receiver exactly which media packets are protected by the FEC packet, and the protection levels and lengths for each of the levels. Specifically, each FEC packet contains an offset mask m(k) for each protection level k. If the bit i in the mask m(k) is set to 1, then media packet number N + i is protected by this FEC packet at level k. N is called the sequence number base, and is sent in the FEC packet as well. When the extended mode is used, the amount of data that are protected at level k is indicated by L(k), which is also sent in the FEC packet. The protection length, offset mask and payload type are sufficient to signal the forward error correction schemes based on arbitrarily defined parity protection with little overhead. A set of rules is described in Section 7.5 that defines how the mask should be set for different protection levels, with examples in Section 10. This document also describes procedures on transmitting all the protection operation parameters in-band. This allows the sender great flexibility; the sender can adapt the protection to current network conditions and be certain the receivers can still make use of the ULP FEC for recovery. At the receiver, the FEC and original media are received. If no media packets are lost, the FEC packets can be ignored. In the event of a loss, the FEC packets can be combined with other received media to recover all or part of the missing media packets. 4. Parity Codes For brevity, we define the function f(x,y,..) to be the XOR (parity) operator applied to the packets x,y,... The output of this function is another packet, called the parity packet. For simplicity, we assume here that the parity packet is computed as the bitwise XOR of the input packets. The exact procedure is specified in section 8. Recovery of data packets using parity codes is accomplished by generating one or more parity packets over a group of data packets. To be effective, the parity packets must be generated by linearly Adam H. Li [Page 7] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 independent combinations of data packets. The particular combination is called a parity code. One class of codes takes a group of k data packets, and generates n-k parity packets. There are a large number of possible parity codes for a given n,k. The payload format does not mandate a particular code. For example, consider a parity code which generates a single parity packet over two data packets. If the original media packets are a,b,c,d, the packets generated by the sender are: a b c d <-- media stream f(a,b) f(c,d) <-- FEC stream where time increases to the right. In this example, the error correction scheme (we use the terms scheme and code interchangeably) introduces a 50% overhead. But if b is lost, a and f(a,b) can be used to recover b. 5. Uneven Level Protection As we can see from the simple example above, the protection on the data depends on the size of the group. In the above example, the group size is 2. So if any one of the three packets (two payload packets and one FEC packet) is lost, the original payload data can still be recovered. In general, the FEC protection operation is a trade off between the bandwidth and the protection strength. A smaller group size will generate stronger protection, and hence have a better chance to recover the protected payload when loss occurs. But on the other hand, it will generate FEC data in a higher frequency, and hence uses more channel bandwidth. As is the common case in most of the media payload, not all the parts of the packets are of the same importance. Using this property, one can potentially achieve more efficient use of the channel bandwidth using unequal error protection, i.e., applying different protection for different parts of the packet. More bandwidth is spent on protecting the more important parts, while less bandwidth on the less important parts. A method to apply unequal error protection with the above-described parity code is to separate the packet into multiple levels and apply parity of different group size to each level. This algorithm is called uneven level protection, or ULP. As we have discussed in the introduction, more of the media streams have the more important parts at the beginning of the packet, so in the ULP it is most useful to have the stronger protection in the levels close to the beginning of the packet, and weaker protection in the levels further back. This is achieved by using different Adam H. Li [Page 8] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 group sizes for different levels, particularly, a lower level (one that is closer to be beginning) will always have a smaller protection size than the ones further back. This is not only because the beginning of the packet has more importance, but it is also to avoid the scenario that a earlier section of a packet is unrecoverable while a later section can be recovered. In such scenarios, the recovered later section is useless and the channel capacity is not fully utilized. The selection of the protection scheme of ULP MUST follow the protection rules as described in Section 7.5. 6. RTP Media Packet Structure The formatting of the media packets is unaffected by FEC. If the FEC is sent as a separate stream, the media packets are sent as if there was no FEC. This scheme leads to a very efficient encoding. When little or no FEC is used, the transmitted stream contains mostly media packets. The overhead for using the FEC scheme is only present in FEC packets, and can be easily monitored and adjusted by tracking the amount of FEC in use. 7. FEC Packet Structure The FEC packets has two modes: Baseline Mode and Extended Mode. The baseline mode protects each media packet to its full length with equal weight to all the parts of the packet. The extended mode provides additional flexibility by allowing different protection to be applied to different parts of the packets. The packets of the two modes of FEC are distinguished by the Extension bit carried in-bound in the FEC header. External signal is not needed. Mode can be changed on the fly in the same stream during a session. 7.1. Baseline Mode FEC In the baseline mode FEC, an FEC packet is constructed by placing an FEC header and FEC payload in the RTP payload, as shown in Figure 1: Adam H. Li [Page 9] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | RTP Header | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | FEC Header (12 or 16 octets) | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | FEC Payload | | | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Figure 1: Baseline mode FEC Packet Structure Please note that baseline mode FEC packets contain only one payload section, and do not have any level headers. 7.2. Extended Mode FEC In the extended mode FEC (which is also called ULP FEC), a FEC packet is constructed by placing an FEC header and one or more levels of FEC header and payload into the RTP payload, as shown in Figure 2: +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | RTP Header | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | FEC Header (12 or 16 octets) | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | FEC Level 0 Header (2 octets) | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | FEC Level 0 Payload | | | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | FEC Level 1 Header (4 or 8 octets) | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | FEC Level 1 Payload | | | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Cont. | | | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Figure 2: Extended mode FEC Packet Structure 7.3. RTP Header of FEC Packets The RTP header of FEC packets are used when the FEC are sent in a separate stream from the protected payload stream (as defined in Section 14). Hence much of the discussion below applies only to that scenario. All the fields in the RTP header of FEC packets are used Adam H. Li [Page 10] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 according to RFC 3550 [1], with some of them further clarified below. Marker: This field is not used for this payload type, and SHALL be set to 0. SSRC: The SSRC value SHALL be the same as the SSRC value of the media stream it protects. Sequence number: The sequence number has the standard definition - it MUST be one higher than the sequence number in the previously transmitted FEC packet. Timestamp: The timestamp MUST be set to the value of the media RTP clock at the instant the FEC packet is transmitted. Thus, the TS value in FEC packets is always monotonically increasing. Payload type: The payload type for the FEC packets is determined through dynamic, out of band means. According to RFC 3550 [1], RTP participants that cannot recognize a payload type must discard it. This provides backwards compatibility. The FEC mechanisms can then be used in a multicast group with mixed FEC-capable and FEC- incapable receivers, particularly when the FEC protection is sent as redundant encoding (see Section 14). In such cases, the FEC protection will have a payload type which is not recognized by the FEC-incapable receivers, and will thus be disregarded. 7.4. FEC Header of FEC Packets This header is 12 or 16 octets, depending on whether the long-mask flag (the L bit, see below) is set. The format of the header is shown in Figure 2 and consists of extension flag (E bit), long-mask flag (L bit), P recovery field, X recovery field, CC recovery field, M recovery field, PT recovery field, SN base field, TS recovery field, and the mask field. 0 1 2 3 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 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |E|L|P|X| CC |M| PT recovery | SN base | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | TS recovery | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | length recovery | mask | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | mask cont. (present only when L = 1) | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Figure 3: FEC Header Format Adam H. Li [Page 11] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 The E bit indicates the mode of the current FEC packet. The E bits for the baseline mode packets MUST be set to 0. The E bits for the extended mode packets MUST be set to 1. The L bit indicates whether the long mask is used. When the L bit is not set, the mask is 16-bit long. When the L bit is set, the mask is then 48-bit long. The P recovery field, the X recovery field, the CC recovery field, the M recovery field, and the PT recovery field are obtained via the protection operation applied to the P, X, CC, M, and PT values of the media packets associated with the FEC packet. The SN base field MUST be set to the lowest sequence number, taking wrap around into account, of those media packets protected by FEC (at all levels for the extended mode). This allows for the FEC operation to extend over any string of at most 16 packets when the L bit is not set or 48 packets when the L bit is set. The TS recovery field is computed via the protection operation applied to the timestamps of the media packets associated with this FEC packet. This allows the timestamp to be completely recovered. The length recovery field is used to determine the length of any recovered packets. It is computed via the protection operation applied to the unsigned network-ordered 16 bit representation of the sums of the lengths (in bytes) of the media payload, CSRC list, extension and padding of media packets associated with this FEC packet (in other words, the CSRC list, RTP extension, and padding of the media payload packets, if present, are "counted" as part of the payload). This allows the FEC procedure to be applied even when the lengths of the media packets are not identical. For example, assume an FEC packet is being generated by xor'ing two media packets together. The length of the two media packets are 3 (0b011) and 5 (0b101) bytes, respectively. The length recovery field is then encoded as 0b011 xor 0b101 = 0b110. The mask field in the FEC header indicates which packets are associated (or associated at level 0 if extended mode is used) with the FEC packet. It is either 16 bits or 48 bits depending on whether the L bit is set. If bit i in the mask is set to 1, then the media packet with sequence number N + i is associated with this FEC packet, where N is the SN Base field in the FEC packet header. The most significant bit of the mask corresponds to i=0, and the least significant to i=15 when the L bit is not set or i=47 when the L bit is set. 7.5. ULP Level Header of FEC Packets The ULP Level Header is 2 octets for ULP level 0, and 4 or 8 octets (depending on whether the L bit is set in the FEC header) for ULP Adam H. Li [Page 12] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 level 1 and higher. The formats of the headers are shown in Figure 3 and Figure 4. Figure 3 shows the ULP FEC level header with level 0. It consists of only one field for the protection length. The Protection Length field is 16 bits. It indicates the protection length provided by the ULP FEC for Level 0 (i.e., the payload length after the SSRC field in the header). 0 1 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Protection Length | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Figure 3: ULP Level Header Format (Level 0) Figure 4 shows the ULP FEC level header with level 1 and higher. It consists of a Protection Length field and a mask field (for level 1 and higher headers). The protection length field is 16-bit long. The mask field is 16-bit long (when the L bit is not set) or 48-bit long (when the L bit is set). Its meaning is the same as the mask field in the main FEC header, except it now indicates which packets are protected by the FEC at the current level. 0 1 2 3 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 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Protection Length | mask | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | mask cont. (present only when L = 1) | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Figure 4: ULP Level Header Format (Level 1 and higher) The setting of mask field when the extended mode FEC is used shall follow the following rules: a. A media packet SHALL be protected only once at each protection level higher than level 0. A media packet MAY be protected more than once at level 0 by different packets, providing the protection lengths of level 0 of these packets are equal. b. For a media packet to be protected at level p, it must also be protect at level p-1 in any FEC packets. Please note that the protection level p for a media packet can be in a FEC Adam H. Li [Page 13] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 packet that is different from the one which contains protection level p-1 for the same media packet. c. If an ULP FEC packet contains protection at level p, it must also contain protection at level p-1. Please that the combination of payload packets that are protected in level p may be different from those of level p-1. One example of the protection combinations is illustrated in Figure 5 below. In this case, eight payload packets are protected by four FEC packets in three different level. Payload Packet # | FEC packet which protects at level | 0 1 2 ---------------------+--------------------------------------- 0 | 0 1 3 1 | 0 1 3 2 | 1 1 3 3 | 1 1 3 4 | 2 3 3 5 | 2 3 3 6 | 3 3 3 7 | 3 3 3 Figure 5: An example of protection combination In this example, FEC packet #0 and #2 only have protection level 0. FEC packet #1 has protection level 0 and 1, and FEC packet #3 has three protection levels. Read across the table, it is shown that payload packet #0 is protected by FEC packet #0 at level 0, by FEC packet #1 at level 1, and FEC packet #3 at level 2, and so on. Also, it can be easily seen from the table that FEC packet #1 protects at level 0 payload packets #2 and #3, at level 1 payload packets #0-#4, and so on. For more examples with more details, please refer to Section 10 Examples. The payload of the ULP FEC packet of each level is the protection operation applied to the concatenation of the CSRC list, RTP extension, media payload, and padding of the media packets associated with the ULP FEC packet. Details are described in the next section on the protection operation. Note that it's possible for the FEC packet to be slightly larger than the media packets it protects (due to the presence of the FEC header). This could cause difficulties if this results in the FEC packet exceeding the Maximum Transmission Unit size for the path along which it is sent. Adam H. Li [Page 14] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 8. Protection Operation The protection operation involves copying the payload, padding it with zeros, and computing the parity (XOR) across the resulting bit strings. The resulting bit string is used to generate the ULP FEC packet. The following procedure MAY be followed for the protection operation. Other procedures MAY be used, but the end result MUST be identical to the one described here. The protection operation can be performed in one of the two modes: the baseline mode and the extended mode. 8.1. Protection Operation in the Baseline Mode The protection operation in the baseline mode is performed as the following. For each media packet to be protected, a bit string is generated by concatenating the following fields together in the order specified: o The first 64 bits of the RTP header (64 bits) o Unsigned network-ordered 16 bit representation of the media packet length in bytes minus 12 (for the fixed RTP header), i.e., the sum of the lengths of all the following if present: the CSRC List, extension header, RTP payload, and RTP padding (16 bits) o if CC is nonzero, the CSRC List (variable length) o if X is 1, the Header Extension (variable length) o the payload (variable length) o the RTP padding, if present (variable length) If the lengths of the bit strings are not equal, each bit string that is shorter than length of the longest, MUST be padded to that length with octet 0. The pad MUST be added at the end of the bit string. The parity operation is then applied across the bit strings. The result is the bit string used to build the FEC packet. We will call this the FEC bit string. The first (most significant) two bits in the FEC bit string are skipped. The next bit in the FEC bit string is written into the P recovery bit of the FEC header in the FEC packet. The next bit in the FEC bit string is written into the E recovery bit of the FEC Adam H. Li [Page 15] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 header. The next four bits of the FEC bit string are written into the CC recovery field of the FEC header. The next bit is written into the M recovery bit of the FEC header. The next 7 bits of the FEC bit string are written into the PT recovery field in the FEC header. The next 16 bits are skipped. The next 32 bits of the FEC bit string are written into the TS recovery field in the FEC header. The next 16 bits are written into the length recovery field in the packet header. The remaining bits are set to be the payload of the FEC packet. 8.2. Protection Operation in the Extended Mode 8.2.1. Extended Mode - Protection Level 0 Protection operation in the extended mode on protection level 0 is very similar to the protection operation in the baseline mode as described above. The only difference is that the protection is only applied to the payload RTP packet up to the protection length plus 96 bits (The 96-bit corresponds to the length of the RTP header upto the end of the SSRC field). That means the padding of the shorter packages is necessary only up to the point of the payload RTP packet at the protection length plus 96 bits. Likewise for the parity operation and copying of the resulting parity string into the FEC packets. 8.2.2. Extended Mode - Protection Level 1 and Higher The protected data of the associated packets are copied into the bit strings. If any of these packets ends before the Protection Length of the current level is reached, the bit string is padded to that length. Octet 0 MUST be used for the padding. The padding MUST be added at the end of the bit string. The parity operation is applied across the bit strings of the corresponding packets as generated above. The resulting FEC bit string of that level is then appended to the payload of the FEC packet. 9. Recovery Procedures The FEC packets allow end systems to recover from the loss of media packets. All of the header fields of the missing packets, including CSRC lists, extensions, padding bits, marker and payload type, are recoverable. This section describes the procedure for performing this recovery. Recovery requires two distinct operations. The first determines which packets (media and FEC) must be combined in order to recover a missing packet. Once this is done, the second step is to actually reconstruct the data. The second step MUST be performed as described below. The first step MAY be based on any algorithm chosen by the Adam H. Li [Page 16] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 implementer. Different algorithms result in a tradeoff between complexity and the ability to recover missing packets, if possible. In the baseline mode, the lost payload packets are always fully recovered when it is recoverable from the FEC data. In the extended mode, however, the lost payload packets may be recovered in full or in parts depending on the data lose situation due to the nature of unequal error protection. The partial recovery of the packet can be detected by checking the recovery length of the packet retrieved from the FEC header against the actual length of the recovered payload data. The applications that use the extended mode need to have the capability of utilizing partially recovered data in order to take advantage of the unequal error protection capacity of the extended mode. 9.1. Reconstruction in the Baseline Mode Let T be the list of packets (FEC and media) which can be combined to recover some media packet xi. The procedure is as follows: 1. For the media packets in T, compute the bit string as described in the protection operation in the baseline mode of the previous section. 2. For the FEC packet in T, compute the bit string by concatenating the first 80 bits of the FEC header with the FEC payload. 3. If any of the bit strings generated from the media packets are shorter than the bit string generated from the FEC packet, pad them to be the same length as the bit string generated from the FEC. The padding of octet 0 MUST be added at the end of the bit string. 4. Perform the exclusive-or (parity) operation across the bit strings, resulting in a recovery bit string. 5. Create a new packet with the standard 12 byte RTP header and no payload. 6. Set the version of the new packet to 2. Skip the first two bits in the recovery bit string. 7. Set the Padding bit in the new packet to the next bit in the recovery bit string. 8. Set the Extension bit in the new packet to the next bit in the recovery bit string. 9. Set the CC field to the next four bits in the recovery bit string. Adam H. Li [Page 17] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 10. Set the marker bit in the new packet to the next bit in the recovery bit string. 11. Set the payload type in the new packet to the next 7 bits in the recovery bit string. 12. Set the SN field in the new packet to xi. Skip the next 16 bits in the recovery bit string. 13. Set the TS field in the new packet to the next 32 bits in the recovery bit string. 14. Take the next 16 bits of the recovery bit string. Whatever unsigned integer this represents (assuming network-order), take that many bytes from the recovery bit string and append them to the new packet. This represents the CSRC list, extension, payload, and the padding of the RTP payload. 15. Set the SSRC of the new packet to the SSRC of the media stream it's protecting, i.e., the SSRC of the media stream to which the FEC stream is associated to. This procedure will completely recover both the header and payload of an RTP packet. 9.2. Reconstruction in the Extended Mode 9.2.1. Extended Mode - Reconstruction of Level 0 The reconstruction in the extended mode is the same as the reconstruction in the baseline mode, except that in step 14, instead of copying a number of bytes as recovered from the length recovery field, a string of number of bytes of the Protection Length of level 0 is copied to the new packet. This procedure will recover both the header and payload of an RTP packet up to the Protection Length of level 0. 9.2.2. Extended Mode - Reconstruction of Level 1 and Higher Let T be the list of packets (FEC and media) which can be combined to recover some media packet xi at certain protection level. The procedure is as follows: 1. For the media packet in T, get the protection length of that level. Copy the data of the that protection level (data of the length read following the level header) to the bit strings. 2. If any of the bit strings generated from the media packets are shorter than the Protection Length of the current level, Adam H. Li [Page 18] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 pad them to that length. The padding of octet 0 MUST be added at the end of the bit string. 3. Perform the exclusive-or (parity) operation across the bit strings, resulting in a recovery bit string. 4. The recovery bit string of the current protection level as generated above is copied and concatenated with the recovery bit string of all the lower levels to form the (fully or partially) recovered payload. The reconstruction operation of the lower level MUST be performed before those of higher level is performed. 5. The total length of the packet is recovered from recovery operation at protection level 0 of the packet. If the recovery operation does not recover the packet to its full length, the un-recovered part of the packet SHOULD be filled (with any data chosen by the implementation) to the total length of the packet. For the FEC data protection in the extended mode, the data protected at lower protection level is almost always recoverable if the higher level protected data is recoverable. This procedure (together with the procedure for the lower protection levels) will usually recover both the header and payload of an RTP packet up to the Protection Length of the current level. 10. Examples In the first three examples considered below (Section 10.1, 10.2, and 10.3), we assume the FEC streams are sent through a separate RTP session as described in Section 14.1. For these examples we assume that 4 media packets are to be sent, A, B, C and D, from SSRC 2. Their sequence numbers are 8, 9, 10 and 11, respectively, and have timestamps of 3, 5, 7 and 9, respectively. Packet A and C uses payload type 11, and packet B and D uses payload type 18. Packet A has 200 bytes of payload, packet B 140, packet C 100 and packet D 340. Packet A and C have their marker bit set. The fourth example, is to illustrate when the FEC data is sent as redundant data with the payload packets. 10.1. A Baseline Mode Example We can protect the four payload packet with one FEC packet in the baseline mode. The scheme is as shown in Figure 6. Adam H. Li [Page 19] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 +-------------------+ : Packet A | | : +-------------+-----+ : Packet B | | : +---------+---+ : Packet C | | : +---------+-----------------------+ Packet D | | +---------------------------------+ : +---------------------------------+ Packet FEC | | +---------------------------------+ : : :<------------- L0 -------------->: Figure 6 FEC scheme in the baseline mode An FEC packet is generated from these four packets. We assume that payload type 127 is used to indicate an FEC packet. The resulting RTP header is shown in Figure 7. The FEC header in the FEC packet is shown in Figure 8. 0 1 2 3 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 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |1 0|0|0|0 0 0 0|0|1 1 1 1 1 1 1|0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1| +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 1| +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0| +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Version: 2 Padding: 0 Extension: 0 Marker: 0 PT: 127 SN: 1 TS: 9 SSRC: 2 Figure 7: RTP Header of FEC Packet (baseline mode) Adam H. Li [Page 20] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 0 1 2 3 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 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |0|0|0|0|0 0 0 0|0|0 0 0 0 0 0 0|0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0| +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0| +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |0 0 0 0 0 0 0 1 0 1 1 1 0 1 0 0|1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0| +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ E: 0 [FEC in baseline mode] L: 0 [short 16-bit mask] P rec.: 0 [0 XOR 0 XOR 0 XOR 0] X rec.: 0 [0 XOR 0 XOR 0 XOR 0] CC rec.: 0 [0 XOR 0 XOR 0 XOR 0] M rec.: 0 [1 XOR 0 XOR 1 XOR 0] PT rec.: 0 [11 XOR 18 XOR 11 XOR 18] SN base: 8 [min(8,9,10,11)] TS rec.: 8 [3 XOR 5 XOR 7 XOR 9] len. rec.: 372 [200 XOR 140 XOR 100 XOR 340] mask: 61440 [with Bit 1, 2, 3, and 4 marked accordingly for Packet 8, 9, 10, and 11] Figure 8: FEC Header of FEC Packet (baseline mode) 10.2. An Extended Mode Example With One Protection Level Suppose we want to protect the data of length L0 = 70 bytes of them at the beginning of these packets, as illustrated in Figure 9 below. +------:------------+ Packet A | : | +------:------+-----+ Packet B | : | +------:--+---+ Packet C | : | +------:--+-----------------------+ Packet D | : | +------:--------------------------+ : +------+ Packet FEC | | +------+ : : :<-L0->: Figure 9 ULP FEC scheme with only protection level 0 Adam H. Li [Page 21] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 The resulting FEC packet will have the same RTP header as shown in Figure 7. The FEC header in the FEC packet is shown in Figure 10. It is the same as the FEC header in Figure 8, except that the E bit is set to 1. The ULP header level 0 present before the parity data of level 0 in the FEC packet is as shown in Figure 11. 0 1 2 3 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 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |1|0|0|0|0 0 0 0|0|0 0 0 0 0 0 0|0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0| +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0| +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |0 0 0 0 0 0 0 1 0 1 1 1 0 1 0 0|1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0| +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ E: 1 [FEC in extended mode] L: 0 [short 16-bit mask] P rec.: 0 [0 XOR 0 XOR 0 XOR 0] X rec.: 0 [0 XOR 0 XOR 0 XOR 0] CC rec.: 0 [0 XOR 0 XOR 0 XOR 0] M rec.: 0 [1 XOR 0 XOR 1 XOR 0] PT rec.: 0 [11 XOR 18 XOR 11 XOR 18] SN base: 8 [min(8,9,10,11)] TS rec.: 8 [3 XOR 5 XOR 7 XOR 9] len. rec.: 372 [200 XOR 140 XOR 100 XOR 340] mask: 61440 [with Bit 1, 2, 3, and 4 marked accordingly for Packet 8, 9, 10, and 11] Figure 10: FEC Header of FEC Packet (extended mode) 0 1 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |0 0 0 0 0 0 0 0 0 1 0 0 0 1 1 0| +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ L0: 70 The payload length for level 0 is 70 bytes. Figure 11: ULP Level Header (Level 0) Adam H. Li [Page 22] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 10.3. An Extended Mode Example With Two Protection Levels A more complex example is to use extended mode FEC at two levels. The level 0 FEC will provide greater protection to the beginning part of the payload packets. The level 1 FEC will apply additional protection to the rest of the packets. This is illustrated in Figure 12. In this example, we take L0 = 70 and L1 = 90. +------:--------:---+ Packet A | : : | +------:------+-:---+ Packet B | : | : +------:--+---+ : : : +------+ : ULP #1 | | : +------+ : : : +------:--+ : Packet C | : | : +------:--+-----:-----------------+ Packet D | : : | +------:--------:-----------------+ : : +------:--------+ ULP #2 | : | +------:--------+ : : : :<-L0->:<--L1-->: Figure 12 ULP FEC scheme with protection level 0 and level 1 This will result in two extended mode FEC packets - #1 and #2. The resulting ULP FEC packet #1 will have the RTP header as shown in Figure 13. The FEC header for ULP FEC packet #1 will be as shown in Figure 14. The level 0 ULP header for #1 will be shown in Figure 15. 0 1 2 3 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 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |1 0|0|0|0 0 0 0|1|1 1 1 1 1 1 1|0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1| +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1| +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0| +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Version: 2 Adam H. Li [Page 23] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 Padding: 0 Extension: 0 Marker: 1 PT: 127 SN: 1 TS: 5 SSRC: 2 Figure 13: RTP Header of FEC Packet #1 0 1 2 3 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 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |1|0|0|0|0 0 0 0|0|0 0 1 1 0 0 1|0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0| +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0| +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |0 0 0 0 0 0 0 0 0 1 0 0 0 1 0 0|1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0| +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ E: 1 [FEC in extended mode] L: 0 [short 16-bit mask] P rec.: 0 [0 XOR 0 XOR 0 XOR 0] X rec.: 0 [0 XOR 0 XOR 0 XOR 0] CC rec.: 0 [0 XOR 0 XOR 0 XOR 0] M rec.: 0 [1 XOR 0 XOR 1 XOR 0] PT rec.: 25 [11 XOR 18] SN base: 8 [min(8,9)] TS rec.: 6 [3 XOR 5] len. rec.: 68 [200 XOR 140] mask: 49152 [with Bit 1 and 2 marked accordingly for Packet 8 and 9] Figure 14: FEC Header of ULP FEC Packet #1 0 1 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |0 0 0 0 0 0 0 0 0 1 0 0 0 1 1 0| +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ L0: 70 The payload length for level 0 is 70 bytes. Figure 15: ULP Level Header (Level 0) for FEC Packet #1 The resulting FEC packet #2 will have the RTP header as shown in Figure 16. The FEC header for FEC packet #2 will be as shown in Adam H. Li [Page 24] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 Figure 17. The level 0 ULP header for #2 will be shown in Figure 18. The level 1 ULP header for #2 will be shown in Figure 19. 0 1 2 3 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 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |1 0|0|0|0 0 0 0|1|1 1 1 1 1 1 1|0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0| +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 1| +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0| +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Version: 2 Padding: 0 Extension: 0 Marker: 1 PT: 127 SN: 2 TS: 9 SSRC: 2 Figure 16: RTP Header of FEC Packet #2 0 1 2 3 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 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |1|0|0|0|0 0 0 0|0|0 0 1 1 0 0 1|0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0| +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 0| +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |0 0 0 0 0 0 0 1 0 0 1 1 0 0 0 0|0 0 1 1 0 0 0 0 0 0 0 0 0 0 0 0| +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ E: 1 [FEC in extended mode] L: 0 [short 16-bit mask] P rec.: 0 [0 XOR 0 XOR 0 XOR 0] X rec.: 0 [0 XOR 0 XOR 0 XOR 0] CC rec.: 0 [0 XOR 0 XOR 0 XOR 0] M rec.: 0 [1 XOR 0 XOR 1 XOR 0] PT rec.: 25 [11 XOR 18] SN base: 8 [min(8,9,10,11)] TS rec.: 14 [7 XOR 9] len. rec.: 304 [100 XOR 340] mask: 12288 [with Bit 3 and 4 marked accordingly for Packet 10 and 11] Figure 17: FEC Header of FEC Packet #2 Adam H. Li [Page 25] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 0 1 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |0 0 0 0 0 0 0 0 0 1 0 0 0 1 1 0| +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ L0: 70 The payload length for level 0 is 70 bytes. Figure 18: ULP Level Header (Level 0) for FEC Packet #2 0 1 2 3 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 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |0 0 0 0 0 0 0 0 0 1 0 1 1 0 1 0|1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0| +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ L1: 90 mask: 61440 [with Bit 1, 2, 3, and 4 marked accordingly for Packet 8, 9, 10, and 11] The payload length for level 1 is 90 bytes. Figure 19: ULP Level Header (Level 1) for FEC Packet #2 10.4. An Extended Mode Example With FEC As Redundant Coding This example illustrates FEC sent as redundant coding in the same stream as the payload. We assume that five media packets are to be sent, A, B, C, D, and E, from SSRC 2. Their sequence numbers are 8, 9, 10, 11, and 12, respectively, and with timestamps of 3, 5, 7, 9, and 11, respectively. All the media data are coded with primary coding (and FEC as redundant coding only protects the primary coding) and uses payload type 11. Packet A has 200 bytes of payload, packet B 140, packet C 100, packet D 340, and packet E 160. Packet A and C have their marker bit set. The FEC scheme we use will be the extended mode with one level as illustrated by Figure 9 in Section 10.2. The protection length L0 = 70 octets in the beginning of the packet. A redundant coding packetization is used with payload type 100. The payload type of the FEC is assumed to be 127. The first four RED packets, RED #1 through RED #4, each contains an individual media packet, A, B, C, or D, respectively. The FEC data protecting the media data in the first four media packets is generated. The fifth packet, RED #5, contains this FEC data as redundant coding along with media packet E. Adam H. Li [Page 26] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 RED Packet #1: Media Packet A RED Packet #2: Media Packet B RED Packet #3: Media Packet C RED Packet #4: Media Packet D RED Packet #5: FEC Packet, Media Packet E RED packet #1 through #4 will have the structure as shown in Figure 20. The RTP header of the RED packet #1 is as shown in Figure 21, with all the other RED packets in similar format with corresponding sequence numbers and time stamps. The primary encoding block header of the RED packets is as shown in Figure 22. +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | RTP Header (RED) - 6 octets | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Primary Encoding Block Header (RED) - 1 octet | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Media Packet Data | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Figure 20: RED Packet Structure - Media Data Only 0 1 2 3 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 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |1 0|0|0|0 0 0 0|0|1 1 0 0 1 0 0|0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1| +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1| +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0| +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Version: 2 Padding: 0 Extension: 0 Marker: 0 [Even though media packet A has marker set] PT: 100 [Payload type for RED] SN: 1 TS: 5 SSRC: 2 Figure 21: RTP Header of RED Packet #1 Adam H. Li [Page 27] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 0 1 2 3 4 5 6 7 +-+-+-+-+-+-+-+-+ |0|0 0 0 1 0 1 1| +-+-+-+-+-+-+-+-+ F bit: 0 [This is the primary coding data] Block PT: 11 [The payload type of media] Figure 22: Primary Encoding Block Header The FEC data is generated not directly from the RED packets, but from the virtual RTP packets containing the media packet data. Those virtual RTP packets can be very easily generated from the RED packets both with or without redundant coding included. The conversion from RED packets to virtual RTP packets is simply done by (1) removing any RED block headers and redundant coding data, and (2) replace the PT in the RTP header with the PT of the primary coding. Note: In the payload format for redundant coding as specified by RFC 2198 the marker bit is lost as soon as the primary coding is carried in the RED packets. So the marker bit can not be recovered regardless the FEC is used or not. As mentioned above, RED packet #5 will contain the FEC data (that protects media packet A, B, C, and D) as well as the data of media packet E. The structure of RED packet #5 is as illustrated in Figure 23. +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | RTP Header (RED) - 6 octets | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Redundant Encoding Block Header (RED) - 4 octet | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | FEC Packet Data | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Primary Encoding Block Header (RED) - 1 octet | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Media Packet Data | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Figure 23: RED Packet Structure - With FEC Data The RTP header of the RED packets with FEC included is the same as shown in Figure 20, with their corresponding sequence numbers and time stamps. In RED packet #5, the redundant encoding block header for the FEC packet data block is as shown below in Figure 24. It will be followed by the FEC packet data which, in this case, includes an FEC header (12 octets as shown in Figure 25), ULP Level 0 header (2 Adam H. Li [Page 28] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 octets as shown in Figure 26) and the ULP Level 0 data (70 octets as set for Level 0). 0 1 2 3 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 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |1|1 1 1 1 1 1 1|0 0 0 0 0 0 0 0 0 0 0 0 0 0|0 0 0 1 0 1 0 1 1 0| +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ F bit: 1 [This is the redundant coding data] Block PT: 127 [The dynamic payload type for FEC] TS Offset: 0 [The instance at which the FEC data is transmitted] Block Len: 84 [FEC header (12 octets) plus ULP Level 0 header (2 octets) and ULP Level 0 data (70 octets)] Figure 24: Redundant Encoding Block Header 11. Security There are two ways to use FEC with encryption in secure communications: one way is to apply the FEC on already encrypted payloads, and the other way is to apply the FEC before the encryption. Since the protected payload of this FEC are RTP packets, applying FEC on encrypted payloads in primarily related to the secure RTP (SRTP) [12]. Because the FEC applies XOR across the payload, the FEC packets should be cryptographically as secure as the original payload. In such cases, additional encryption of the FEC packets is not necessary. In the following discussion, it is assumed that the FEC is applied to the payload before the encryption. The use of FEC has implications on the usage and changing of keys for encryption. As the FEC packets do consist of a separate stream, there are a number of combinations on the usage of encryption. These include: o The FEC stream may be encrypted, while the media stream is not. o The media stream may be encrypted, while the FEC stream is not. o The media stream and FEC stream are both encrypted, but using different keys. o The media stream and FEC stream are both encrypted, but using the same key. The first three of these would require all application level signaling protocols used to be aware of the usage of FEC, and to Adam H. Li [Page 29] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 thus exchange keys and negotiate encryption usage on the media and FEC streams separately. In the final case, no such additional mechanisms are needed. The first two cases present a layering violation, as ULP FEC packets should be treated no differently than other RTP packets. Encrypting just one stream may also make certain known-plaintext attacks possible. For these reasons, applications utilizing encryption SHOULD encrypt both streams. The changing of encryption keys is another crucial issue needs to be addressed. Consider the case where two packets a and b are sent along with the FEC packet that protects them. The keys used to encrypt a and b are different, so which key should be used to decode the FEC packet? In general, old keys need to be cached, so that when the keys change for the media stream, the old key can be used until it is determined that the key has changed for the ULP FEC packets as well. The sender and the receiver need to define how the encryption is performed and how the keys are used. The integrity of the FEC packets can have a big impact on the reconstruction operation. Changing some bits in the FEC payload can have significant effect on the calculation and the correct recovery of the payload packets. For example, change the length recovery field can result in the recovery of a packet which is too long. Also, the computational complexity of the recovery can be easily effected for up to at least one order of magnitude. 12. Congestion Considerations Another issue with the use of FEC is its impact on network congestion. In many situations, the packet loss in the network is induced by congestions. In such scenarios, adding FEC when encountering increasing network losses should be avoided. If it is used on a widespread basis, this can result in increased congestion and eventual congestion collapse. The applications may include stronger protections while at the same time reduce the bandwidth for the payload packets. In any event, implementations MUST NOT substantially increase the total amount of bandwidth in use (including the payload and the FEC) as network losses increase. The general congestion control considerations for transporting RTP data apply, see RTP [1] and any applicable RTP profile like AVP [13]. An additional requirement if best-effort service is being used is: users of this payload format MUST monitor packet loss to ensure that the packet loss rate is within acceptable parameters. Packet loss is considered acceptable if a TCP flow across the same network path, and experiencing the same network conditions, would achieve an average throughput, measured on a reasonable timescale, that is not less than the RTP flow is achieving. This condition can be satisfied by implementing congestion control mechanisms to adapt the transmission rate (or the number of layers subscribed for a layered Adam H. Li [Page 30] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 multicast session), or by arranging for a receiver to leave the session if the loss rate is unacceptably high. 13. MIME Registrations Four new MIME sub-types as described in this section are to be registered with IANA. This registration is done using the registration template [3] and following RFC 3555 [4]. 13.1. Registration of audio/ulpfec MIME media type name: audio MIME subtype name: ulpfec Required parameters: Rate: The RTP timestamp rate which is used to mark the time of transmission of the FEC packet in separate stream. In cases it is sent as redundant data to another stream the rate SHALL be the same as the primary encoding it is used to protect. When used in a separate stream the rate SHALL be larger than 1000 Hz to provide sufficient resolution to RTCP operations. The selected rate MAY be any value above 1000 Hz but is RECOMMENDED to match the rate of the media this stream protects. Optional parameters: none Encoding considerations: This format is framed (see Section 4.8 in the template document [3]) and contains binary data. Restrictions on usage: This media type depends on RTP framing, and hence is only defined for transfer via RTP [1]. Transport within other framing protocols SHALL NOT be defined as this is a robustness mechanism for RTP. Security considerations: the same security considerations apply to these MIME registrations as to the payloads for them, as detailed in RFC xxxx. Interoperability considerations: none Published specification: RFC xxxx. Applications which use this media type: Audio and video streaming tools which seek to improve resiliency to loss by sending additional data with the media stream. Additional information: none Adam H. Li [Page 31] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 Person & email address to contact for further information: Adam Li adamli@hyervision.com IETF Audio/Video Transport Working Group Intended usage: COMMON Author: Adam Li adamli@hyervision.com Change controller: IETF Audio/Video Transport Working Group delegated from the IESG. 13.2. Registration of video/ulpfec MIME media type name: video MIME subtype name: ulpfec Required parameters: Rate: The RTP timestamp rate which is used to mark the time of transmission of the FEC packet in separate stream. In cases it is sent as redundant data to another stream the rate SHALL be the same as the primary encoding it is used to protect. When used in a separate stream the rate SHALL be larger than 1000 Hz to provide sufficient resolution to RTCP operations. The selected rate MAY be any value above 1000 Hz but is RECOMMENDED to match the rate of the media this stream protects. Optional parameters: none Encoding considerations: This format is framed (see Section 4.8 in the template document [3]) and contains binary data. Restrictions on usage: This media type depends on RTP framing, and hence is only defined for transfer via RTP [1]. Transport within other framing protocols SHALL NOT be defined as this is a robustness mechanism for RTP. Security considerations: the same security considerations apply to these MIME registrations as to the payloads for them, as detailed in RFC xxxx. Interoperability considerations: none Published specification: RFC xxxx. Applications which use this media type: Audio and video streaming tools which seek to improve resiliency to loss by sending additional data with the media stream. Adam H. Li [Page 32] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 Additional information: none Person & email address to contact for further information: Adam Li adamli@hyervision.com IETF Audio/Video Transport Working Group Intended usage: COMMON Author: Adam Li adamli@hyervision.com Change controller: IETF Audio/Video Transport Working Group delegated from the IESG. 13.3. Registration of text/ulpfec MIME media type name: text MIME subtype name: ulpfec Required parameters: Rate: The RTP timestamp rate which is used to mark the time of transmission of the FEC packet in separate stream. In cases it is sent as redundant data to another stream the rate SHALL be the same as the primary encoding it is used to protect. When used in a separate stream the rate SHALL be larger than 1000 Hz to provide sufficient resolution to RTCP operations. The selected rate MAY be any value above 1000 Hz but is RECOMMENDED to match the rate of the media this stream protects. Optional parameters: none Encoding considerations: This format is framed (see Section 4.8 in the template document [3]) and contains binary data. Restrictions on usage: This media type depends on RTP framing, and hence is only defined for transfer via RTP [1]. Transport within other framing protocols SHALL NOT be defined as this is a robustness mechanism for RTP. Security considerations: the same security considerations apply to these MIME registrations as to the payloads for them, as detailed in RFC xxxx. Interoperability considerations: none Published specification: RFC xxxx. Adam H. Li [Page 33] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 Applications which use this media type: Audio and video streaming tools which seek to improve resiliency to loss by sending additional data with the media stream. Additional information: none Person & email address to contact for further information: Adam Li adamli@hyervision.com IETF Audio/Video Transport Working Group Intended usage: COMMON Author: Adam Li adamli@hyervision.com Change controller: IETF Audio/Video Transport Working Group delegated from the IESG. 13.4. Registration of application/ulpfec MIME media type name: application MIME subtype name: ulpfec Required parameters: Rate: The RTP timestamp rate which is used to mark the time of transmission of the FEC packet in separate stream. In cases it is sent as redundant data to another stream the rate SHALL be the same as the primary encoding it is used to protect. When used in a separate stream the rate SHALL be larger than 1000 Hz to provide sufficient resolution to RTCP operations. The selected rate MAY be any value above 1000 Hz but is RECOMMENDED to match the rate of the media this stream protects. Optional parameters: none Encoding considerations: This format is framed (see Section 4.8 in the template document [3]) and contains binary data. Restrictions on usage: This media type depends on RTP framing, and hence is only defined for transfer via RTP [1]. Transport within other framing protocols SHALL NOT be defined as this is a robustness mechanism for RTP. Security considerations: the same security considerations apply to these MIME registrations as to the payloads for them, as detailed in RFC xxxx. Interoperability considerations: none Adam H. Li [Page 34] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 Published specification: RFC xxxx. Applications which use this media type: Audio and video streaming tools which seek to improve resiliency to loss by sending additional data with the media stream. Additional information: none Person & email address to contact for further information: Adam Li adamli@hyervision.com IETF Audio/Video Transport Working Group Intended usage: COMMON Author: Adam Li adamli@hyervision.com Change controller: IETF Audio/Video Transport Working Group delegated from the IESG. 14. Multiplexing of FEC The FEC packets can be sent to the receiver along with the protected payload primarily in one of the two ways: as a separate stream, or in the same stream as redundant encoding. 14.1. FEC as a Separate Stream When the FEC packets are sent in a separate stream, several pieces of information must be conveyed: o The address and port where the FEC is being sent to o The payload type number for the FEC o Which media stream the FEC is protecting There is no static payload type assignment for FEC, so dynamic payload type numbers MUST be used. The SSRC of the FEC stream MUST be set to that of the protected payload stream. The association of the FEC stream with its corresponding stream is done by line grouping in SDP [5] with the FEC semantics [6] or other external means. Following the principles as discussed in Section 5.2 of RFC 3550 [1], multiplexing of the FEC stream and its associated payload stream is usually provided by the destination transport address (network address and port number) which is different for each RTP session. Sending FEC together with the payload in one single RTP session and multiplex only by SSRC or payload type precludes: (1) Adam H. Li [Page 35] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 the use of different network paths or network resource allocations for the payload and the FEC protection data; (2) reception of a subset of the media if desired, particularly for the hosts which do not understand FEC; and (3) receiver implementations that use separate processes for the different media. In additional, multiplexing FEC with payload data streams will affect the timing and sequence number space of the original payload stream, which is usually undesirable. So the FEC stream and the payload stream SHOULD be sent through two separate RTP session, and multiplexing them by payload type into one single RTP session SHOULD be avoided. In additional, the FEC and the payload MUST NOT be multiplexed by SSRC into one single RTP session since they always have the same SSRC. 14.2. FEC as Redundant Encoding When the FEC stream is being sent as a secondary codec in the redundant encoding format, this must be signaled through SDP. To do this, the procedures defined in RFC 2198 [7] are used to signal the use of redundant encoding. The FEC payload type is indicated in the same fashion as any other secondary codec. The FEC MUST protect only the main codec, with the payload of FEC engine coming from virtual RTP packets created from the main codec data. The virtual RTP packets can be very easily converted from the RFC 2198 packets by simply (1) removing all the additional headers and the redundant coding data, and (2) replacing the payload type in the RTP header with that of the primary codec. Note: In the payload format for redundant coding as specified by RFC 2198 the marker bit is lost as soon as the primary coding is carried in the RED packets. So the marker bit can not be recovered regardless the FEC is used or not. Because the FEC data (including the ULP header) is sent in the same packets as the protected payload. The FEC data is associated with the protected payload by being bundled in the same stream. 15. Indicating FEC Usage in SDP FEC packets contain RTP packets with dynamic payload type values. In addition, the FEC packets can be sent on separate multicast groups or separate ports from the media. The FEC can even be carried in packets containing media using the redundant encoding payload format [7]. These configuration options MUST be indicated out of band. This section describes how this can be accomplished using the Session Description Protocol (SDP), specified in RFC 2327 [8]. 15.1. FEC as a Separate Stream In the case discussed in this subsection, the FEC packets are sent as a separate stream. This means that they can be sent on a different port and/or multicast group from the media. Adam H. Li [Page 36] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 Just like any media stream, the port number and the payload type number for the FEC stream is conveyed in its m line in the SDP. There is no static payload type assignment for FEC, so dynamic payload type numbers MUST be used. The binding to the number is indicated by an rtpmap attribute. The name used in this binding is "ulpfec". The address that the FEC stream is on is conveyed in its corresponding c line. The association relationship between the FEC stream and the payload stream it protects is conveyed through media line grouping in SDP (RFC 3388) [5] using FEC semantics (RFC yyyy) [6]. The FEC stream and the protected payload stream forms an FEC group. The following is an example SDP for FEC application in a multicast session: v=0 o=adam 289083124 289083124 IN IP4 host.example.com s=ULP FEC Seminar t=0 0 c=IN IP4 224.2.17.12/127 a=group:FEC 1 2 a=group:FEC 3 4 m=audio 30000 RTP/AVP 0 a=mid:1 m=application 30002 RTP/AVP 100 a=rtpmap:100 ulpfec/8000 a=mid:2 m=video 30004 RTP/AVP 31 a=mid:3 m=application 30004 RTP/AVP 101 c=IN IP4 224.2.17.13/127 a=rtpmap:101 ulpfec/8000 a=mid:4 The presence of two a=group lines in this SDP indicates that there FEC groups. The first FEC group, as indicated by the "a=group:FEC 1 2" line, consists of stream 1 (an audio stream using PCM) and stream 2 (the protecting FEC stream). The FEC stream is sent to the same multicast group and has the same TTL as the audio, but on a port number two higher. The second FEC group, as indicated by the "a=group:FEC 3 4" line, consists of stream 3 (an video stream) and stream 4 (the protecting FEC stream). The FEC stream is sent to a different multicast address, but has the same port number (30004) as the payload video stream. 15.2. FEC as Redundant Encoding When the FEC stream is sent as a secondary codec in the redundant encoding format, this can signaled through SDP. To do this, the procedures defined in RFC 2198 [7] are used to signal the use of redundant encoding. The FEC payload type is indicated in the same Adam H. Li [Page 37] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 fashion as any other secondary codec. An rtpmap attribute MUST be used to indicate a dynamic payload type number for the FEC packets. The FEC MUST protect only the main codec. For example: m=audio 12345 RTP/AVP 121 0 5 100 a=rtpmap:121 red/8000/1 a=rtpmap:100 ulpfec/8000 a=fmtp:121 0/5/100 This SDP indicates that there is a single audio stream, which can consist of PCM (media format 0) , DVI (media format 5), the redundant encodings (indicated by media format 121, which is bound to red through the rtpmap attribute), or FEC (media format 100, which is bound to ulpfec through the rtpmap attribute). Although the FEC format is specified as a possible coding for this stream, the FEC MUST NOT be sent by itself for this stream. Its presence in the m line is required only because non-primary codecs must be listed here according to RFC 2198. The fmtp attribute indicates that the redundant encodings format can be used, with DVI as a secondary coding and FEC as a tertiary encoding. 15.3. Offer / Answer Consideration Some considerations are needed when SDP is used for offer / answer [RFC 3264] exchange. When the FEC is sent as a separate stream and signaled through media line grouping in SDP (RFC 3388) [5] using FEC semantics (RFC yyyy) [6], the offering side MUST implement both RFC 3388 and RFC yyyy. The rules for offer / answer in RFC 3388 and RFC YYYY SHALL be followed with the below additional consideration. For all offers with FEC, the answerer MAY refuse the separate FEC session by setting the port to 0, and remove the "a=group" attribute that groups that FEC session with the RTP session being protected. If the answerer accepts the usage of FEC, the answer simply accepts the FEC RTP session and the grouping in the offer by including them in the answer. Note that the rejection of FEC RTP session does not prevent the media sessions from being accepted and used without FEC. When the FEC stream is sent as a secondary codec in the redundant encoding format (RFC 2198) [7], the offering side can indicate the FEC stream as specified in Section 14.2. The answer MAY reject the FEC stream by removing the payload type for the FEC stream. To accept the usage of FEC, the answerer must in the answer include the FEC payload type. Note that in cases the redundancy payload format [7] is used with FEC as the only secondary codec, when the FEC stream is rejected the redundant encoding payload type SHOULD also be removed. Adam H. Li [Page 38] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 16. Application Statement The generic FEC algorithm specified in this document is designed to deal with any type of packet loss occurring in transmission. This FEC algorithm is fully interoperable between the hosts that are FEC- capable and those that are not. Since the media payload is not altered and the protection is sent as additional information, the receivers that are unaware of the generic FEC as specified in this document can simply ignore the additional FEC information and process the main media payload. This interoperability is particularly important for backward compatibility with existing hosts, and also in the scenario where many different hosts need to communicate with each other at the same time, such as during multicast. The generic FEC algorithm specified in this document is also a generic protection algorithm with the following features: (1) it is independent of the nature of the media being protected, whether that media is audio, video, or otherwise, (2) it is flexible enough to support a wide variety of FEC mechanisms and settings, (3) it is designed for adaptivity, so that the FEC parameters can be modified easily without resorting to out of band signaling, and (4) it supports a number of different mechanisms for transporting the FEC packets. The extended mode of FEC (also called ULP) further generalizes and extends the generic FEC algorithm here, and provides user with Unequal Error Protection capabilities. Some other algorithms may also provide the Unequal Error Protection capabilities thought other means. For example, an Unequal Erasure Protection (UXP) scheme has been proposed in the AVT Working Group in "An RTP Payload Format for Erasure-Resilient Transmission of Progressive Multimedia Streams". The UXP scheme applies unequal error protection to the media payloads by interleaving the payload stream to be protected with the additional redundancy information obtained using Reed-Solomon operations. By altering the structure of the protected media payload, the UXP scheme sacrifices the backward compatibility with terminals that do not support UXP. This makes it more difficult to apply UXP when backward compatibility is desired. In the case of ULP, however, the media payload remains un-altered and can always be used by the terminals. The extra protection can simply be ignored if the receiving terminals do not support ULP. At the same time, also because the structure of the media payload is altered in UXP, UXP offers the unique ability to change packet size independent of the original media payload structure and protection applied, and is only subject to the protocol overhead constraint. This property is useful in scenarios when altering the packet size of the media at transport level is desired. Adam H. Li [Page 39] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 Because of the interleaving used in UXP, delays will be introduced at both the encoding and decoding sides. For UXP, all data within a transmission block need to arrive before encoding can begin, and a reasonable number of packets must be received before a transmission block can be decoded. The ULP scheme introduces little delay at the encoding side. On the decoding side, correctly received packets can be delivered immediately. Delay is only introduced in ULP when packet losses occur. Because UXP is an interleaved scheme, the un-recoverable errors occurring in data protected by UXP usually result in a number of corrupted holes in the payload stream. In ULP, on the other hand, the unrecoverable errors due to packet loss in the bitstream usually appear as contiguous missing pieces at the end of the packets. Depending on the encoding of the media payload stream, many applications may find it easier to parse and extract data from a packet with only a contiguous piece missing at the end than a packet with multiple corrupted holes, especially when the holes are not coincident with the independently decodable fragment boundaries. The exclusive-or (XOR) parity check operation used by ULP is simpler and faster than the more complex operations required by Reed-Solomon codes. This makes ULP more suitable for applications where computational cost is a constraint. As discussed above, both the ULP and the UXP schemes apply unequal error protection to the RTP media stream, but each uses a different technique. Both schemes have their own unique characteristics, and each can be applied to scenarios with different requirements. 17. Acknowledgments The following authors have made significant contributions to this document: Adam H. Li, Fang Liu, John D. Villasenor, Dong-Seek Park, Jeong-Hoon, Yung-Lyul Lee, Jonathan D. Rosenberg, and Henning Schulzrinne. The authors would also like to acknowledge the suggestions from many people, particularly Magnus Westerlund, Stephen Casner, Colin Perkins, Tao Tian, Matthieu Tisserand, Stephen Wenger, Jay Fahlen, and Jeffery Tseng. 18. Bibliography 18.1. Normative References [1] H. Schulzrinne, S. Casner, R. Frederick, and V. Jacobson, "RTP: a transport protocol for real-time applications," Request for Comments (Proposed Standard) 3550, Internet Engineering Task Force, July 2003. Adam H. Li [Page 40] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 [2] S. Bradner, "Key words for use in RFCs to indicate requirement levels," Request for Comments (Best Current Practice) 2119, Internet Engineering Task Force, March 1997. [3] N. Freed and J. Klensin, "Media Type Specifications and Registration Procedures", IETF work in progress. [4] S. Casner, and P. Hoschka, "MIME type registration of RTP payload formats", Request for Comments (Proposed Standard) 3555, Internet Engineering Task Force, July 2003. [5] G. Camarillo, J. Holler, and H. Schulzrinne, "Grouping of Media Lines in the Session Description Protocol (SDP)", RFC 3388, December 2002. [6] A. Li, "FEC Grouping Semantics in SDP", IETF work in progress. [7] C. Perkins, I. Kouvelas, O. Hodson, V. Hardman, M. Handley, J.C. Bolot, A. Vega-Garcia, and S. Fosse-Parisis, "RTP Payload for Redundant Audio Data", Request for Comments (Proposed Standard) 2198, Internet Engineering Task Force, September 1997. [8] M. Handley, and V. Jacobson, "SDP: Session Description Protocol", Request for Comments (Proposed Standard) 2327, Internet Engineering Task Force, April 1998. 18.2. Informative References [9] J. Rosenberg and H. Schulzrinne, "An RTP Payload Format for Generic Forward Error Correction," Request for Comments (Proposed Standard) 2733, Internet Engineering Task Force, December 1999. [10] C. Perkins and O. Hodson, "Options for repair of streaming media", Request for Comments (Informational) 2354, Internet Engineering Task Force, June 1998. [11] J. Rosenberg and H. Schulzrine, "Registration of parityfec MIME types", Request for Comments (Proposed Standard) 3009, Internet Engineering Task Force, November 2000. [12] M. Baugher, D. McGrew, M. Naslund, E. Carrara, K. Norrman, "The Secure Real-time Transport Protocol (SRTP)", Request for Comments (Proposed Standard) 3711, Internet Engineering Task Force, March 2004. [13] H. Schulzrinne and S. Casner, "RTP Profile for Audio and Video Conferences with Minimal Control", Request for Comments (Proposed Standard) 3551, Internet Engineering Task Force, July 2003. Adam H. Li [Page 41] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 19. Author's Addresses Adam H. Li 10194 Wateridge Circle #152 San Diego, CA 92121 USA Phone: +1 858 622 9038 Email: adamli@hyervision.com Adam H. Li [Page 42] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 Copyright Statement Copyright (C) The Internet Society (2005). This document is subject to the rights, licenses and restrictions contained in BCP 78, and except as set forth therein, the authors retain all their rights." Disclaimer of Validity This document and the information contained herein are provided on an "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY AND THE INTERNET ENGINEERING TASK FORCE DISCLAIM ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. The IETF takes no position regarding the validity or scope of any Intellectual Property Rights or other rights that might be claimed to pertain to the implementation or use of the technology described in this document or the extent to which any license under such rights might or might not be available; nor does it represent that it has made any independent effort to identify any such rights. Information on the procedures with respect to rights in RFC documents can be found in BCP 78 and BCP 79. Copies of IPR disclosures made to the IETF Secretariat and any assurances of licenses to be made available, or the result of an attempt made to obtain a general license or permission for the use of such proprietary rights by implementers or users of this specification can be obtained from the IETF on-line IPR repository at http://www.ietf.org/ipr. The IETF invites any interested party to bring to its attention any copyrights, patents or patent applications, or other proprietary rights that may cover technology that may be required to implement this standard. Please address the information to the IETF at ietf-ipr@ietf.org. Adam H. Li [Page 43] I-Draft An RTP Payload Format for Generic FEC Nov 28, 2005 RFC Editor Considerations The RFC editor is kindly requested to perform the following editing to this draft: - Replace all occurrences of xxxx with the RFC number this document receives. - Replace reference [6] and all occurrences of RFC yyyy with the corresponding title and RFC number of that ID when it is published. - Remove this and the next section "Changes". Changes Compared to the previous version of this document, draft-ietf-avt- ulp-12.txt, the following changes have been made: (1) A example on using FEC with RFC 2198 is added. (2) A few minor editorial changes are made. This Internet-Draft expires May 28, 2006. Adam H. Li [Page 44]