Network Working Group                                   Mikael Degermark
INTERNET-DRAFT                                            Bjorn Nordgren
Expires December 18, 1996                                   Stephen Pink
                                              Lulea Technical University
                                                       Telia Research AB
                                   Swedish Institute of Computer Science
                                                                  Sweden
                                                           June 13, 1996


                      Header Compression for IPv6
                    <draft-degermark-ipv6-hc-01.txt>


Status of this Memo

   Publication of this document does not imply acceptance by the IPng
   Area of any ideas expressed within. Comments should be submitted to
   the mailing list ipng@sunroof.eng.sun.com

   Distribution of this memo is unlimited.

   This document is an Internet-Draft.  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 any
   time.  It is inappropriate to use Internet- Drafts as reference
   material or to cite them other than as ``work in progress.''

   To learn the current status of any Internet-Draft, please check the
   ``1id-abstracts.txt'' listing contained in the Internet- Drafts
   Shadow Directories on ftp.is.co.za (Africa), nic.nordu.net (Europe),
   munnari.oz.au (Pacific Rim), ds.internic.net (US East Coast), or
   ftp.isi.edu (US West Coast).

Abstract

   This document describes how to compress IPv6 headers over point-to-
   point links.  The method can be applied to IPv6 base and extension
   headers, IPv4 headers, TCP and UDP headers, and encapsulated IPv6 and
   IPv4 headers.

   A typical IPv6/UDP header can be compressed down to 4 octets
   including the UDP checksum.  A typical IPv6/TCP header can be
   compressed down to 4-6 bytes including the TCP checksum.



Degermark, Nordgren, Pink                                       [Page 1]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


   TABLE OF CONTENTS
      1.  Introduction..............................................3
      2.  Terminology...............................................5
      3.  Compression method........................................7
           3.1.  Lost packets in TCP packet streams.................7
           3.2.  Lost packets in UDP and non-TCP packet streams.....8
              3.2.1.  Compression Slow-Start........................9
              3.2.2.  Periodic Header Refreshes.....................9
              3.2.3.  Rules for sending Full Headers...............10
              3.2.4.  Cost of sending Header Refreshes.............11
      4.  Grouping packets into packet streams.....................13
           4.1.  Guidelines for grouping packets...................13
      5.  Size Issues..............................................15
           5.1.  Compression identifiers...........................15
           5.2.  Size of compression state.........................16
           5.3.  Size of full headers..............................16
              5.3.1.  Length fields in full TCP headers............18
              5.3.2.  Length fields in full non-TCP headers........18
      6.  Compressed Header Formats................................19
      7.  Compression of subheaders................................21
           7.1.  IPv6 Header.......................................23
           7.2.  IPv6 Extension Headers............................23
           7.3.  Options...........................................24
           7.4.  Hop-by-hop Options Header.........................25
           7.5.  Routing Header....................................26
           7.6.  Fragment Header...................................28
           7.7.  Destination Options Header........................29
           7.8.  No Next Header....................................29
           7.9.  Authentication Header.............................30
           7.10. Encapsulating Security Payload Header.............31
           7.11. UDP Header........................................31
           7.12. TCP Header........................................32
           7.13. IPv4 Header.......................................33
      8.  Changing compression identifiers.........................35
      9.  Rules for dropping or temporarily storing packets........36
      10. Low-loss header compression for TCP .....................37
           10.1.  The twice algorithm..............................37
           10.2.  Header Requests..................................38
      11. Links that reorder packets...............................39
           11.1.  Reordering in non-TCP packet streams.............39
           11.2.  Reordering in TCP packet streams.................39
      12. Demultiplexing...........................................41
      13. Configuration Parameters.................................42
      14. Implementation Status....................................44
      15. Acknowledgments..........................................44
      15. Security Considerations..................................44
      16. Author's Addresses.......................................44
      17. References...............................................45



Degermark, Nordgren, Pink                                       [Page 2]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


   1.  Introduction

   There are several reasons to do header compression on low- or
   medium-speed links. Header compression can

      *  Improve interactive response time

         For very low-speed links, echoing of characters may take longer
         than 100-200 ms because of the time required to transmit large
         headers. 100-200 ms is the maximum time people can tolerate
         without feeling that the system is sluggish.

      *  Allow using small packets for bulk data with good line
         efficiency

         This is important when interactive (for example Telnet) and
         bulk traffic (for example FTP) is mixed because the bulk data
         should be carried in small packets to decrease the waiting time
         when a packet with interactive data is caught behind a bulk
         data packet.

      *  Allow using small packets for delay sensitive low data-rate
         traffic

         For such applications, for example voice, the time to fill a
         packet with data is significant if packets are large.  To get
         low end-to-end delay small packets are preferred.  Without
         header compression, the smallest possible IPv6/UDP headers (48
         octets) consume 19.2 kbit/s with a packet rate of 50 packets/s.
         50 packets/s is equivalent to having 20 ms worth of voice
         samples in each packet. Tunneling or routing headers, for
         example to support mobility, will increase the bandwidth
         consumed by headers by at least 10-20 kbit/s.  This should be
         compared with the bandwidth required for the actual sound
         samples, for example 13 kbit/s with GSM encoding.  Header
         compression can reduce the bandwidth needed for headers
         significantly, in the example to about 1.7 kbit/s.

      *  Decrease header overhead.

         A common size of TCP segments for bulk transfers over medium-
         speed links is 512 bytes. When TCP segments are tunneled, for
         example because Mobile IP is used, the header is 100 bytes.
         Header compression will decrease the header overhead from 19.5
         per cent to about 1 per cent. This is a significant gain for
         line-speeds as high as a few Mbit/s.

      *  Reduce packet loss rate over lossy links.



Degermark, Nordgren, Pink                                       [Page 3]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


         Because fewer bits are sent per packet, the packet loss rate
         will be lower for a given bit-error rate. This results in
         higher throughput for TCP as the sending window can open up
         more between losses, and in fewer lost packets for UDP.

   The mechanisms described here are intended for a point-to-point link.
   However, care has been taken to allow extensions for multi-access
   links and multicast.

   Headers that can be compressed include TCP, UDP, IPv4, and IPv6 base
   and extension headers.  For TCP packets, the mechanisms of Van
   Jacobson [RFC-1144] are used to recover from loss. For non-TCP
   packets, compression slow-start and periodic header refreshes allow
   minimal periods of packet discard after loss of a header that changes
   the compression state.

   Header compression relies on many fields being constant or changing
   seldomly in consecutive packets belonging to the same packet stream.
   Fields that do not change between packets need not be transmitted at
   all.  Fields that change often with small and/or predictable values,
   e.g., TCP sequence numbers, can be encoded incrementally so that the
   number of bits needed for these fields decrease significantly.  Only
   fields that change often and randomly, e.g., checksums or
   authentication data, need to be transmitted in every header.

   The general principle of header compression is to occasionally send a
   packet with a full header; subsequent compressed headers refer to the
   full header and may contain incremental changes to the full header.























Degermark, Nordgren, Pink                                       [Page 4]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


2.  Terminology

   This section explains some terms used in this document.

   Subheader

         An IPv6 base header, an IPv6 extension header, an IPv4 header,
         a UDP header, or a TCP header.

   Header

         A chain of subheaders.

   Compress

         The act of reducing the size of a header by removing header
         fields or reducing the size of header fields. This is done in a
         way such that a decompressor can reconstruct the header if its
         compression state is identical to the compression state used
         when compressing the header.

   Decompress

         The act of reconstructing a compressed header.

   Compression identifier (CID)

         A small unique number identifying the compression state that
         should be used to decompress a compressed header. Carried in
         full headers and compressed headers.

   Compression state

         The state which the compressor uses to compress a header and
         the decompressor uses to decompress a header.  The compression
         state is the uncompressed version of the last header sent
         (compressor) or received (decompressor) over the link, except
         for fields in the header that are included "as-is" in
         compressed headers or can be inferred from, e.g., the size of
         the link-level frame.

         The compression state for a packet stream is associated with a
         compression identifier.  The compression state for non-TCP
         packet streams is also associated with a generation.







Degermark, Nordgren, Pink                                       [Page 5]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


   Generation

         For non-TCP packet streams, each new version of the compression
         state for a given CID is associated with a generation: a small
         number that is incremented whenever the compression state
         associated with that CID changes. Carried by full and
         compressed non-TCP headers.

   Packet stream

         A sequence of packets whose headers are similar and share
         compression state. For example, headers in a TCP packet stream
         have the same source and final destination address, and the
         same port numbers in the TCP header.  Similarly, headers in a
         UDP packet stream have the same source and destination address,
         and the same port numbers in the UDP header.

   Full header (header refresh)

         An uncompressed header that updates or refreshes the
         compression state for a packet stream. It carries a CID that
         will be used to identify the compression state.

         Full headers for non-TCP packet streams also carry the
         generation of the compression state they update or refresh.

   Regular header

         A normal, uncompressed, header.  Does not carry CID or
         generation association.

   Incorrect decompression

         When a compressed and then decompressed header is different
         from the uncompressed header. Usually due to mismatching
         compression state between the compressor and decompressor or
         bit errors during transmission of the compressed header.

   Differential coding

         A compression technique where the compressed value of a header
         field is the difference between the current value of the field
         and the value of the same field in the previous header
         belonging to the same packet stream. A decompressor can thus
         obtain the value of the field by adding the value in the
         compressed header to its compression state.  This technique is
         used for TCP streams but not for non-TCP streams.




Degermark, Nordgren, Pink                                       [Page 6]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


3.  Compression method

   Much of the header information stays the same over the life-time of a
   packet stream. For non-TCP packet streams almost all fields of the
   headers are constant. For TCP many fields are constant and others
   change with small and predictable values.

   To initiate compression of the headers of a packet stream, a full
   header carrying a compression identifier, CID, is transmitted over
   the link.  The compressor and decompressor store most fields of this
   full header as compression state.  The compression state consists of
   the fields of the header whose values are constant and thus need not
   be sent over the link at all, or change little between consecutive
   headers so that it uses fewer bits to send the difference from the
   previous value compared to sending the absolute value.

   Any change in fields that are expected to be constant in a packet
   stream will cause the compressor to send a full header again to
   update the compression state at the decompressor. As long as the
   compression state is the same at compressor and decompressor, headers
   can be decompressed to be exactly as they were before compression.
   However, if a full header or compressed header is damaged or lost
   during transmission, the compression state of the decompressor may
   become obsolete as it is not updated properly. Compressed headers
   will then be decompressed incorrectly.

   So a header compression scheme needs mechanisms to update the
   compression state at the decompressor and to detect or avoid
   incorrect decompression.  These mechanisms are very different for TCP
   and non-TCP streams, and are described in sections 3.1 and 3.2.

   The compression mechanisms in this document assume that packets are
   not reordered between the compressor and decompressor.  If the link
   does reorder, section 11 describes mechanisms for ordering the
   packets before decompression.  It is also assumed that the link-layer
   implementation can provide the length of packets, and that there is
   no padding in UDP packets or tunneled packets.

3.1.  Lost packets in TCP packet streams

   Since TCP headers are compressed using the difference from the
   previous TCP header, loss of a packet with a compressed or full
   header will cause subsequent compressed headers to be decompressed
   incorrectly because the compression state used for decompression was
   not incremented properly.

   Loss of a compressed TCP header will cause the TCP sequence numbers
   of subsequently decompressed TCP headers to be off by k, where k is



Degermark, Nordgren, Pink                                       [Page 7]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


   the size of the lost segment.  Such incorrectly decompressed TCP
   headers will be discarded by the TCP receiver as the TCP checksum
   reliably catches "off-by-k" errors in the sequence numbers for
   plausible k.

   TCP's repair mechanisms will eventually retransmit the discarded
   segment and the compressor peeks into the TCP headers to detect when
   TCP retransmits.  When this happens, the compressor sends a full
   header on the assumption that the retransmission was due to
   mismatching compression state at the decompressor.  [RFC-1144] has a
   good explanation of this mechanism.

3.2.  Lost packets in UDP and other non-TCP packet streams

   Incorrectly decompressed headers of UDP packets and other non-TCP
   packets are not so well-protected by checksums as TCP packets because
   differential coding is not used and there are no sequence numbers.
   The UDP checksum only covers payload, UDP header, and pseudo header.
   The pseudo header includes the source and destination addresses, the
   transport protocol type and the length of the transport packet.
   Except for those fields, large parts of the IPv6 header are not
   covered by the UDP checksum.  Moreover, other non-TCP headers lack
   checksums altogether, for example fragments.

   In order to safely avoid incorrect decompression of non-TCP headers,
   each version of the compression state for non-TCP packet streams is
   identified by a generation, a small number that is carried by the
   full headers that establish and refresh the compression state.
   Compressed headers carry the generation value of the compression
   state that were used to compress them.  When a decompressor sees that
   a compressed header carries a generation value other than the
   generation of its compression state for that packet stream, the
   compression state is not up to date and the packet must be discarded
   or stored until a full header establishes correct compression state.

   Differential coding is not used for non-TCP streams, so compressed
   non-TCP headers do not change the compression state.  Thus, loss of a
   compressed header does not invalidate subsequent packets with
   compressed headers. Moreover, the generation field changes only when
   the compression state of a full header is different from the
   compression state of the previous full header. This means that losing
   a full header will make the compression state of the decompressor
   obsolete only when the full header would actually have changed the
   compression state.

   The generation field is 7 bits long so the generation value repeats
   itself after 128 changes to the compression state. To avoid incorrect
   decompression after error bursts or other temporary disruptions, the



Degermark, Nordgren, Pink                                       [Page 8]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


   compressor must not reuse the same generation value after a shorter
   time than MIN_WRAP seconds. A decompressor which has been
   disconnected MIN_WRAP seconds or more must wait for the next full
   header before decompressing.

   Instead of reusing a generation value too soon, a compressor may
   switch to another CID or else send regular headers until MIN_WRAP
   seconds have passed.  The value of MIN_WRAP is found in section 12.

3.2.1.  Compression Slow-Start

   To allow the decompressor to recover quickly from loss of a full
   header that would have changed the compression state, full headers
   are sent periodically with an exponentially increasing period after a
   change in the compression state. This technique avoids an exchange of
   messages between compressor and decompressor used by other
   compression schemes, such as in [RFC-1553]. Such exchanges can be
   costly for wireless mobiles as more power is consumed by the
   transmitter and delay can be introduced by repeated switching between
   sending and receiving.  Moreover, techniques that require an exchange
   of messages cannot be used over simplex links, such as direct-
   broadcast satellite channels or cable TV systems, and are hard to
   adapt to multicast.

     |.|..|....|........|................|..............................
     ^
     Change   Sent packets: | with full header, . with compressed header

   The picture shows how packets are sent after change.  The compressor
   keeps a variable for each non-TCP packet stream, F_PERIOD, that keeps
   track of how many compressed headers may be sent between full
   headers.  When the headers of a non-TCP packet stream change so that
   its compression state changes, a full header is sent and F_PERIOD is
   set to one. After sending F_PERIOD compressed headers a full header
   is sent, and F_PERIOD is doubled each time a full header is sent.

3.2.2.  Periodic Header Refreshes

   To avoid losing too many packets if a receiver has lost its
   compression state, there is an upper limit, F_MAX_PERIOD, on the
   number of non-TCP packets with compressed headers that may be sent
   between header refreshes. If a packet is to be sent and F_MAX_PERIOD
   compressed headers have been sent since the last full header for this
   packet stream was sent, a full header must be sent.

   To avoid long periods of disconnection for low data rate packet
   streams, there is also an upper bound, F_MAX_TIME, on the time
   between full headers in a non-TCP packet stream. If a packet is to be



Degermark, Nordgren, Pink                                       [Page 9]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


   sent and more than F_MAX_TIME seconds have passed since the last full
   header was sent for this packet stream, a full header must be sent.
   The values of F_MAX_PERIOD and F_MAX_TIME are found in section 12.

3.2.3. Rules for sending Full Headers

   The following pseudo code can be used by the compressor to determine
   when to send a full header for a non-TCP packet stream.  The code
   maintains two variables:

      C_NUM       -- a count of the number of compressed headers sent
                     since the last full header was sent.
      F_LAST      -- the time of sending the last full header.

   and uses the functions

      current_time()       return the current time
      min(a,b)             return the smallest of a and b

   the procedures send_full_header() and send_compressed_header()
   do the obvious thing.

      if ( <this header changes the compression state> )

          C_NUM := 0;
          F_LAST := current_time();
          F_PERIOD := 1;
          send_full_header();           -- generation value incremented

      elseif ( C_NUM >= F_PERIOD )

          C_NUM := 0;
          F_LAST := current_time();
          F_PERIOD := min(2 * F_PERIOD, F_MAX_PERIOD);
          send_full_header();           -- generation value unchanged

      elseif ( current_time() > F_LAST + F_MAX_TIME )

          C_NUM := 0;
          F_LAST := current_time();
          send_full_header();           -- generation value unchanged

      else

          C_NUM := C_NUM + 1
          send_compressed_header();     -- with current generation value

      endif



Degermark, Nordgren, Pink                                      [Page 10]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


3.2.4.  Cost of sending Header Refreshes

   If every f'th packet carries a full header, H is the size of a full
   header, and C is the size of a compressed header, the average header
   size is

                 (H-C)/f + C

   For f > 1, the average header size is (H-C)/f larger than a
   compressed header.

   In a diagram where the average header size is plotted for various f
   values, there is a distinct knee in the curve, i.e., there is a limit
   beyond which further increasing f gives diminishing returns.
   F_MAX_PERIOD should be chosen to be a frequency well to the right of
   the knee of the curve.  For typical sizes of H and C, say 48 octets
   for the full header (IPv6/UDP) and 4 octets for the compressed
   header, setting F_MAX_PERIOD > 44 means that full headers will
   contribute less than a byte to the average header size. With a four-
   address routing header, F_MAX_PERIOD > 115 will have the same effect.

   The default F_MAX_PERIOD value of 256 (section 12) puts the full
   header frequency well to the right of the knee and means that full
   headers will typically contribute considerably less than a byte to
   the average header size.  For H = 48 and C = 4, full headers
   contribute about 1.4 bits to the average header size after reaching
   the steady-state header refresh frequency determined by the default
   F_MAX_PERIOD. 1.4 bits is a very small overhead.

   After a change in compression state, the exponential backoff scheme
   will initially send full headers frequently.  The default
   F_MAX_PERIOD will be reached after nine full headers and 255
   compressed headers have been sent.  This is equivalent to a little
   over 5 seconds for a typical voice stream with 20 ms worth of voice
   samples per packet.

   During the whole backoff period, full headers contribute 1.5 octets
   to the average header size, when H = 48 and C = 4.  For 20 ms voice
   samples, it takes less than 1.3 seconds until full headers contribute
   less than one octet to the average header size, and during these
   initial 1.3 seconds full headers add less than 4 octets to the
   average header size.  The cost of the exponential backoff is not
   great and as the headers of non-TCP packet streams are expected to
   change seldomly, it will be amortized over a long time.







Degermark, Nordgren, Pink                                      [Page 11]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


   The cost of header refreshes in terms of bandwidth are higher than
   similar costs for hard state schemes like [RFC-1553] where full
   headers must be acknowledged by the decompressor before compressed
   headers may be sent. Such schemes typically send one full header plus
   a few control messages when the compression state changes.  Hard
   state schemes require more types of protocol messages and an exchange
   of messages is necessary.  Hard state schemes also need to deal
   explicitly with various error conditions that soft state handles
   automatically, for instance the case of one party disappearing
   unexpectedly, a common situation on wireless links where mobiles may
   go out of range of the base station.

   The major advantage of our soft state scheme is that no handshakes
   are needed between compressor and decompressor, so the scheme can be
   used over simplex links.  The costs in terms of bandwidth are higher
   than for hard state schemes, but we feel that the simplicity of the
   decompressor, the simplicity of the protocol, and the lack of
   handshakes between compressor and decompressor justifies this small
   cost. Moreover, soft state schemes are more easily extended to
   multi-access links and multicast.































Degermark, Nordgren, Pink                                      [Page 12]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


4.  Grouping packets into packet streams

   This section explains how packets may be grouped together into packet
   streams for compression.

   To achieve the best compression rates, packets should be grouped
   together such that packets in the same packet stream have similar
   headers. If this grouping fails, the header compression performance
   will be bad, since the compression algorithm can rarely utilize the
   existing compression state for the packet stream and full headers
   must be sent frequently.

   Grouping is done by the compressor. A compressor may use whatever
   criterion it finds appropriate to group packets into packet streams.
   To determine what packet stream a packet belongs to, a compressor
   might

   a) examine the compressible chain of subheaders (see section 7),

   b) examine the contents of an upper layer protocol header that
      follows the compressible chain of subheaders, for example ICMP
      headers, DVMRP headers, or tunneled IPX headers,

   c) use information obtained from a resource manager, for example if a
      resource manager requests compression for a particular packet
      stream and provides a way to identify packets belonging to that
      packet stream,

   d) use any other relevant information, for example if routes flap and
      the hop limit (TTL) field in a packet stream changes frequently
      between n and n+k, a sender may choose to group the packets into
      two different packet streams.

   A compressor is also free not to group packets into packet streams
   for compression, letting some packets keep their regular headers and
   passing them through unmodified.

   As long as the rules for when to send full headers for a packet
   stream are followed and subheaders are compressed as specified in
   this document, the decompressor is able to reconstruct a compressed
   header correctly regardless of how packets are grouped into packet
   streams.

4.1  Guidelines for grouping packets

   In the absence of specific instructions as to which packet streams to
   compress, we offer the following quidelines for how a compressor may
   group packets into packet streams for compression.



Degermark, Nordgren, Pink                                      [Page 13]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


   Defining fields

         All fields in subheaders that are marked with DEF in section 7
         should be present and identical in all packets belonging to the
         same packet stream. The DEF marked fields include the flow
         label, source and destination addresses of IP headers, final
         destination address in routing headers, the next header field
         preceding a UDP or TCP header, port numbers, and the SPI in
         authentication and encryption headers.

   Fragmented packets

         Fragmented and unfragmented packets are never grouped together
         in the same packet stream. The Identification field of the
         Fragment header or IPv4 header is not used to identify the
         packet stream. If it was, the first fragment of a new packet
         would cause a compression slow-start.

         No field after a Fragment Header or an IPv4 header for a
         fragment should be used for grouping purposes.

   Upper protocol identification

         The first next header field identifying a header not described
         in section 7 should be used for identifying packet streams,
         i.e., all packets with the same DEF fields and the same upper
         protocol should be grouped together.

   TTL field

         A sophisticated implementation can monitor the TTL field and if
         it changes frequently use it as a DEF field. This might occur
         when there are frequent route flaps so that packets traverse
         different paths through the internet.

   Priority field

         It is concievable that the Priority field of the IPv6 header
         can change between packets with identical DEF fields when the
         Flow Label is zero. A sophisticated implementation can watch
         out for this and be prepared to use the Priority field as a DEF
         field.

   When IP packets are tunneled they are encapsulated with an additional
   IP header at the tunnel entry point and then sent to the tunnel
   endpoint. To group such packets into packet streams, the inner
   headers should also be examined to determine the packet stream.  If
   this is not done, full headers will be sent each time the headers of



Degermark, Nordgren, Pink                                      [Page 14]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


   the inner IP packet changes.  So when a packet is tunneled, the
   identifying fields of the inner subheaders should be considered in
   addition to the identifying fields of the initial IP header.

   An implementation can use other fields for identification than the
   ones described here. If too many fields are used for identification,
   performance might suffer because more CIDs will be used and the wrong
   CIDs might be reused when new flows need CIDs. If too few fields are
   used for identification, performance might suffer because there are
   too frequent changes in the compression state.

   We stress that these guidlines are only educated guesses, when IPv6
   is widely deployed and IPv6 traffic can be analyzed, we might find
   that other grouping algorithms perform better. We also stress that if
   the grouping fails, the result will be bad performance and not
   incorrect decompression. The decompressor can do its task regardless
   of how the grouping algorithm works.

5.  Size Issues

5.1.  Compression Identifiers

   Compression identifiers can be 8 or 24 bits long.  Their size is not
   relevant for finding the compression state.  An 8-bit CID with value
   two and an 24-bit CID with value two are equivalent.

   The CID spaces for TCP and non-TCP are separate, so a TCP CID and a
   non-TCP CID never identify the same compression state, even if they
   have the same value. This doubles the available CID space while using
   the same number of bits for CIDs.  It is always possible to tell
   whether a full or compressed header is for a TCP or non-TCP packet,
   so no mixups can occur.

   Non-TCP compressed headers encode the size of the CID using one bit
   in the first byte of the compressed header. The 8-bit CID allows a
   minimum compressed header size of 2 octets for non-TCP packets, the
   size bit and the 7-bit Generation value fit in the first octet and
   the CID  uses the second octet.

   For TCP the only available CID size is 8 bits.  8 bits is probably
   sufficient as TCP connections are always point-to-point.

   The 24 bit CID size is probably not needed for point-to-point links;
   it is intended for use on multi-access links where a larger CID space
   may be needed for efficient selection of CIDs.

   The major difficulty with multi-access links is that several
   compressors share the CID space of a decompressor.  CIDs can no



Degermark, Nordgren, Pink                                      [Page 15]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


   longer be selected independently by the compressors as collisions may
   occur.  This problem may be resolved by letting the decompressors
   have a separate CID space for each compressor.  Having separate CID
   spaces requires that decompressors can identify which compressor sent
   the compressed packet, perhaps by utilizing link-layer information as
   to who sent the link-layer frame.  If such information is not
   available, all compressors on the multi-access link may be
   enumerated, automatically or otherwise, and supply their number as
   part of the CID. This latter method requires a large CID space.

5.2.  Size of Compression State

   The size of the compression state should be limited to simplify
   implementation of compressor and decompressor, and put a limit on
   their memory requirements.  However, there is no upper limit on the
   size of an IPv6 header as the chain of extension headers can be
   arbitrarily long.  This is a problem as the compression state is
   essentially a stored header.

   The configurable parameter MAX_HEADER (see section 12) represents the
   maximum size of the compression state, expressed as the maximum sized
   header that can be stored as compression state.  When an IPv6 header
   is larger than MAX_HEADER, only part of it is stored as compression
   state.  An implementation must not compress more than the initial
   MAX_HEADER octets of a header.  An implementation must not partially
   compress a subheader.  Thus, the part of the header that is stored as
   compression state and compressed is the longest initial sequence of
   entire subheaders that is not larger than MAX_HEADER.

         NOTE: We are not totally happy with this solution since most of
         the [MAX_HEADER * (size of CID spaces)] sized memory is wasted
         if headers are small.  We are investigating other solutions to
         the problem of specifying tight memory requirements while
         allowing correct decompression and a low number of packet
         discards.

5.3.  Size of full headers

   It is desirable to avoid increasing the size of packets with full
   headers beyond their original size, as their size may be optimized
   for the MTU of the link. Since we assume that the link layer
   implementation provides the length of packets, we can use the length
   fields in full headers to pass the values of the CID and the
   generation to the decompressor.

   This requires that the link-layer must not add padding to the
   payload, at least not padding that can be delivered to the
   destination link user. It is also required that no extra padding is



Degermark, Nordgren, Pink                                      [Page 16]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


   added after UDP data or in tunneled packets so that the values of
   lenght fields can be calculated from the length of headers and the
   length of the link-layer frame.

   The generation requires one octet and the CID may require up to 3
   octets.  Length fields of 2 octets occur in the IPv6 Base Header, the
   IPv4 header, and the UDP header.

   A full TCP header will thus have at least 2 octets available in the
   IPv6 base header to pass the 8 bit CID, which is sufficient.  [RFC-
   1144] uses the 8 bit Protocol field of the IPv4 header to pass the
   CID. We cannot use the corresponding method for IPv6 as the sequence
   of extension headers is not fixed and CID values are not disjoint
   from the legal values of Next Header fields.

   An IPv6/UDP or IPv4/UDP packet will have 4 octets available to pass
   the generation and the CID, so all CID sizes may be used. Fragmented
   or encrypted packet streams may have only 2 octets available to pass
   the generation and CID.  Thus, 8-bit CIDs may be the only CID sizes
   that can be used for such packet streams.  When IPv6/IPv4 or
   IPv4/IPv6 tunneling is used, there will be at least 4 octets
   available, and all CID sizes may be used for non-TCP packet streams.

   The 7-bit generation value is passed in the higher order octet of the
   first length field in the full header. When only one length field is
   available, the 8-bit CID is passed in the low order octet.  When two
   length fields are available, the lowest two octets of the CID are
   passed in the second length field and the low order octet of the
   first length field carries the highest octet of the CID.






















Degermark, Nordgren, Pink                                      [Page 17]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


5.3.1.  Length fields in full TCP headers

   Full TCP headers with one length field:

                                    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
                     Length field   | LSB of pkt nr |      CID      |
                                    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

   Full TCP headers with two or more length fields:

                                    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
               First length field   | LSB of pkt nr |      CID      |
                                    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

                                    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
               Second length field  | MSB of pkt nr |       0       |
                                    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

   See section 11 for a description of how the pkt nr field is used.


5.3.2.  Length fields in full non-TCP headers

   Full non-TCP headers with 1 octet CID field:

                                    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
               First length field   |0|  Generation |      CID      |
                                    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

   Full non-TCP headers with 3 octet CID field:

                                    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
               First length field   |1|  Generation |   MSB of CID  |
                                    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

                                    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
               Second length field  |   two lowest octets of CID    |
                                    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+













Degermark, Nordgren, Pink                                      [Page 18]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


6.  Compressed Header Formats

   This section uses some terminology (DELTA, RANDOM) defined in section
   7.

   a) Compressed TCP header format (same as [RFC 1144]):

         +-+-+-+-+-+-+-+-+- - - - - - - -
         | |C I P S A W U|  CID (if C=1) |
         +-+-+-+-+-+-+-+-+- - - - - - - -
         | RANDOM fields, if any (see section 7)
          - - - - - - - - - - - -
         | Sequence Number Delta (if S=1)
          - - - - - - - - - - - -
         | Acknowledgment Number Delta (if A=1)
          - - - - - - - - - - - -
         | Window (if W=1)
         +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
         |         TCP Checksum          |
         +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
         | Urgent Pointer (present if U=1)
          - - - - - - - - - - - -

   The flags in the first byte (CIPSAWU) have the same meaning as in
   [RFC-1144], regardless of whether the TCP segments are carried by
   IPv6 or IPv4. The compression state associated with the CID keeps
   track of the IP version. The order between delta fields specified
   here is different from [RFC-1144].

   The initial flag byte is followed by an 8-bit CID if the C flag is
   set, after which the RANDOM and DELTA fields of the subheaders follow
   in the same order as they occur in the original uncompressed header.
   The C-flag must always be set on links that can reorder packets.

   The I flag is zero unless an IPv4 header is present as there is no
   Identification field in the IPv6 header. If there are more than one
   IPv4 header present, only the Identification field of the IPv4 header
   closest to the TCP header is delta encoded, other Identification
   fields are RANDOM. The delta of the Identification field is placed
   among the RANDOM fields at the position corresponding to the IPv4
   header.

   If the I flag is set, the delta of the IPv4 Identifier value is
   placed at the position that corresponds to the place of the IPv4
   subheader in the uncompressed header.

   See section 7.12 and [RFC-1144] for further information about how to
   compress TCP headers.



Degermark, Nordgren, Pink                                      [Page 19]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


   b) Compressed non-TCP header, 8 bit CID:

       +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
       |0|  Generation |      CID      |
       +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
       | RANDOM fields, if any (see section 7)
        - - - - - - - - - - - -


   c) Compressed non-TCP header, 24 bit CID:

       +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
       |1|  Generation |                      CID                      |
       +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
       | RANDOM fields, if any (see section 7)
        - - - - - - - - - - - -

   Compressed non-TCP headers are followed by relevant RANDOM fields
   (see section 7) placed in the same order as they occur in the
   original uncompressed header, followed by the payload.































Degermark, Nordgren, Pink                                      [Page 20]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


7.  Compression of subheaders

   This section gives rules for how the compressible chain of subheaders
   is compressed.  Subheaders that may be compressed include IPv6 base
   and extension headers, TCP headers, UDP headers, and IPv4 headers.
   The compressible chain of subheaders extends from the beginning of
   the header

   a) up to but not including the first header that is not an IPv4
      header, an IPv6 base or extension header, a TCP header, or a UDP
      header, or

   b) up to and including the first TCP header, UDP header, Fragment
      Header, Encapsulating Security Payload Header, or IPv4 header for
      a fragment,

   whichever gives the shorter chain. For example, rules a) and b) both
   fit a chain of subheaders that contain a Fragment Header and ends at
   a tunneled IPX packet. Since rule b) gives a shorter chain, the
   compressible chain of subheaders stops at the Fragment Header.

   The following subsections are a systematic classification of how all
   fields in subheaders are expected to change.

   NOCHANGE    The field is not expected to change. Any change means
               that a full header must be sent to update the compression
               state.

   DELTA       The field may change often but usually the difference
               from the field in the previous header is small, so that
               it is cheaper to send the change from the previous value
               rather than the current value.  This type of compression
               is not used for non-TCP packet streams.

   RANDOM      The field should be included "as-is" in compressed
               headers, usually because it changes unpredictably.

   INFERRED    The field contains a value that can be inferred from
               other values, for example the size of the frame carrying
               the packet, and thus need not be included in the
               compressed header.










Degermark, Nordgren, Pink                                      [Page 21]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


   The classification implies how a compressed header is constructed. No
   field that is NOCHANGE or INFERRED is present in a compressed header.
   A compressor obtains the values of NOCHANGE fields from the
   compression state identified by the compression identifier, and
   obtains the values of INFERRED fields from the link-layer
   implementation, e.g., from the size of the link-layer frame, or from
   other fields, e.g., by recalculating the IPv4 header checksum.  DELTA
   fields are encoded as the difference to the value in the previous
   packet in the same packet stream, the decompressor adds the value in
   the compressed header to the value in its compression state to obtain
   the proper value.  RANDOM fields are sent "as-is" in the compressed
   header.  DELTA and RANDOM fields occur in the same order in the
   compressed header as they occur in the full header.

   There is currently little experience with actual IPv6 traffic, so
   this classification may change as IPv6 traffic can be observed.

   Fields that may be used to identify what packet stream a packet
   belongs to according to section 4.1 are marked with the word DEF. To
   a compressor using the guidelines from section 4.1, any difference in
   corresponding DEF fields between two packets implies that they belong
   to different packet streams. Moreover, if a DEF field is present in
   one packet but not in another, the packets belong to different packet
   streams.



























Degermark, Nordgren, Pink                                      [Page 22]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


7.1.  IPv6 Header [IPv6, section 3]

      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |Version| Prio. |                   Flow Label                  |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |         Payload Length        |  Next Header  |   Hop Limit   |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                                                               |
      +                                                               +
      |                                                               |
      +                         Source Address                        +
      |                                                               |
      +                                                               +
      |                                                               |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                                                               |
      +                                                               +
      |                                                               |
      +                      Destination Address                      +
      |                                                               |
      +                                                               +
      |                                                               |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

         Version                   NOCHANGE (DEF)
         Prio                      NOCHANGE
         Flow Label                NOCHANGE (DEF)
         Payload Length            INFERRED
         Next Header               NOCHANGE
         Hop Limit                 NOCHANGE
         Source Address            NOCHANGE (DEF)
         Destination Address       NOCHANGE (DEF)


   The Payload Length field of encapsulated headers must correspond to
   the length value of the encapsulating header. If not, the header
   chain cannot be compressed.

   This classification implies that the whole IPv6 base header can be
   compressed away.

7.2.  IPv6 Extension Headers [IPv6, section 4]

   What extension headers are present and the relative order of them is
   not expected to change in a packet stream.  Whenever there is a
   change, a full packet header must be sent.  All Next Header fields in
   IPv6 base header and IPv6 extension headers are NOCHANGE.




Degermark, Nordgren, Pink                                      [Page 23]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


7.3.  Options [IPv6, section 4.2]

   The contents of Hop-by-hop Options and Destination Options extension
   headers are encoded with TLV "options":

         +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+- - - - - - - - -
         |  Option Type  |  Opt Data Len |  Option Data
         +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+- - - - - - - - -

   Option Type and Opt Data Len fields are assumed to be fixed for a
   given packet stream, so they are classified as NOCHANGE.  The Option
   data is RANDOM unless specified otherwise below.

   Padding

      Pad1 option

         +-+-+-+-+-+-+-+-+
         |       0       |
         +-+-+-+-+-+-+-+-+

         Entire option is NOCHANGE.

      PadN option

         +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+- - - - - - - - -
         |       1       |  Opt Data Len |  Option Data
         +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+- - - - - - - - -

         All fields are NOCHANGE.





















Degermark, Nordgren, Pink                                      [Page 24]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


7.4.  Hop-by-Hop Options Header [IPv6, section 4.3]


      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |  Next Header  |  Hdr Ext Len  |                               |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+                               +
      |                                                               |
      .                                                               .
      .                            Options                            .
      .                                                               .
      |                                                               |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

      Next Header          NOCHANGE
      Hdr Ext Len          NOCHANGE

      Options              TLV coded values and padding.
                           Classified according to 7.3 above, unless
                           being a Jumbo Payload option (see below).

   Jumbo Payload option

                                      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
                                      |      194      |Opt Data Len=4 |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                     Jumbo Payload Length                      |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

      First two fields are NOCHANGE and Jumbo Payload Length INFERRED.
      (frame length must be supplied by link layer implementation).

         NOTE: It is silly to compress the headers of a packet carrying
         a Jumbo Payload Option since the relative header overhead is
         negligible. Moreover, it is usually a bad idea to send such
         large packets over low- and medium-speed links.
















Degermark, Nordgren, Pink                                      [Page 25]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


7.5.  Routing Header [IPv6, section 4.4]

      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |  Next Header  |  Hdr Ext Len  |  Routing Type | Segments Left |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                                                               |
      .                                                               .
      .                       type-specific data                      .
      .                                                               .
      |                                                               |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

   If the Routing Type is not recognized, it is impossible to determine
   the final Destination Address unless the Segments Left field has the
   value zero, in which case the Destination Address is the final
   Destination Address. For unknown routing types, all fields of the
   Routing Header are NOCHANGE.

   If the Flow Label is zero and the Segments Left field is not zero,
   the packet stream can be identified only when the Routing Type is 0,
   the only Routing Type currently defined.

   The fields of the Type 0 Routing Header are classified as follows:

      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |  Next Header  |  Hdr Ext Len  | Routing Type=0| Segments Left |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |   Reserved    |             Strict/Loose Bit Mask             |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                                                               |
      +                                                               +
      |                                                               |
      +                           Address[1]                          +
      |                                                               |
      +                                                               +
      |                                                               |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                                                               |
      +                                                               +
      |                                                               |
      +                           Address[2]                          +
      |                                                               |
      +                                                               +
      |                                                               |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      .                               .                               .
      .                               .                               .
      .                               .                               .



Degermark, Nordgren, Pink                                      [Page 26]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


      .                               .                               .
      .                               .                               .
      .                               .                               .
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                                                               |
      +                                                               +
      |                                                               |
      +                           Address[n]                          +
      |                                                               |
      +                                                               +
      |                                                               |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

      Next Header          NOCHANGE
      Routing Type         NOCHANGE
      Hdr Ext Len          NOCHANGE
      Segments Left        NOCHANGE

      Reserved             NOCHANGE

      Strict/Loose Bit Mask
                           NOCHANGE

      Address[1..n]        NOCHANGE
                             (Address[n] is DEF if (Segments Left > 0))

   This classification allows Routing Headers of type 0 to be compressed
   away completely. This is a big win as the maximum size of the Routing
   Header is 392 octets.






















Degermark, Nordgren, Pink                                      [Page 27]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


7.6.  Fragment Header [IPv6, section 4.5]

   The first fragment of a packet has Fragment Offset = 0 and the chain
   of subheaders extends beyond its Fragment Header. If a fragment is
   not the first (Fragment Offset not 0), there are no subsequent
   subheaders (unless the chain of subheaders in the first fragment
   didn't fit entirely in the first fragment).

   Since packets may be reordered before reaching the compression point,
   and some fragments may follow other routes through the network, a
   compressor cannot rely on seeing the first fragment before other
   fragments. This implies that information in subheaders following the
   Fragment Header of the first fragment cannot be examined to determine
   the proper packet stream for other fragments.

   It is possible to design compression schemes that can compress
   subheaders after the Fragment Header, at least in the first fragment,
   but to avoid complicating the rules for sending full headers and the
   rules for compression and decompression, the chain of subheaders that
   follow a Fragment Header must not be compressed.

   The fields of the Fragment Header are classified as follows.

      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |  Next Header  |   Reserved    |      Fragment Offset    |Res|M|
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                         Identification                        |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

      Next Header          NOCHANGE
      Reserved             NOCHANGE
      Res                  RANDOM
      M flag               RANDOM
      Fragment Offset      RANDOM
      Identification       RANDOM   (it is never compressed away)

   This classification implies that a Fragment Header is compressed down
   to 6 octets. The minimum IPv6 MTU is 576 octets so most fragments
   will be at least 576 octets. Since the 6 octet overhead of the
   compressed fragment header is amortized over a fairly large packet,
   the additional complexity of more sophisticated compression schemes
   is not justifiable.

         NOTE: The Identification field is RANDOM instead of NOCHANGE to
         avoid one compression slow-start per packet.






Degermark, Nordgren, Pink                                      [Page 28]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


   Grouping of fragments according to the guidelines in section 4.1

      Fragments and unfragmented packets should not be grouped together.

      Port numbers cannot be used to identify the packet stream because
      port numbers are not present in every fragment.  To adhere to the
      uniqueness rules for the Identification value, a fragmented packet
      stream is identified by the combination of Source Address and
      (final) Destination Address.

            NOTE: The Identification value is NOT used to identify the
            packet stream. This avoids using a new CID for each packet
            and saves the cost of the associated compression slow-start.
            We hope that the unfragmentable part of the headers will not
            change too frequently, if it does thrashing may occur.

7.7.  Destination Options Header [IPv6, section 4.6]

        +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
        |  Next Header  |  Hdr Ext Len  |                               |
        +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+                               +
        |                                                               |
        .                                                               .
        .                            Options                            .
        .                                                               .
        |                                                               |
        +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

        Next Header          NOCHANGE
        Hdr Ext Len          NOCHANGE

        Options              TLV coded values and padding.
                             Compressed according to 7.3 above.

   The only Destination Options defined in [IPv6] are the padding
   options.  When further Destination Options are defined, more clever
   compression techniques may be defined.

7.8.  No Next Header [IPv6, section 4.7]

   Covered by rules for IPv6 Header Extensions (7.2).










Degermark, Nordgren, Pink                                      [Page 29]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


7.9.  Authentication Header [RFC-1826, section 3.2]

       1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
      +---------------+---------------+---------------+---------------+
      | Next Header   | Length        |           RESERVED            |
      +---------------+---------------+---------------+---------------+
      |                Security Parameters Index (SPI)                |
      +---------------+---------------+---------------+---------------+
      |                                                               |
      +     Authentication Data (variable number of 32-bit words)     |
      |                                                               |
      +---------------+---------------+---------------+---------------+

      Next Header          NOCHANGE
      Length               NOCHANGE
      Reserved             NOCHANGE
      SPI                  NOCHANGE (DEF)
      Authentication Data  RANDOM

   [RFC-1828] specifies how to do authentication with keyed MD5, the
   authentication method all IPv6 implementations must support.  For
   this method, the Authentication Data is 128 bits. These 128 bits are
   sent "as-is" in compressed headers and thus increase the size of the
   compressed header by 16 octets.  This is typically more than a 300%
   increase in the size of the compressed header.


























Degermark, Nordgren, Pink                                      [Page 30]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


7.10.  Encapsulating Security Payload Header [RFC-1827, section 3.1]

   This header implies that the subsequent parts of the packet are
   encrypted. Thus, no further header compression is possible on
   subsequent headers as encryption is typically already performed when
   the compressor sees the packet.

   However, when the ESP Header is used in tunnel mode an entire IP
   packet is encrypted, and the headers of that packet may be compressed
   before the packet is encrypted at the entry point of the tunnel.
   This means that it must be possible to feed an IP packet and its
   length to the decompressor, as if it came from the link-layer. The
   mechanisms for dealing with reordering described in section 11 should
   be used, as packets are likely to be reordered in a tunnel.

      +---------------+---------------+---------------+---------------+
      |        Security Association Identifier (SPI), 32 bits         |
      +===============+===============+===============+===============+
      |            Opaque Transform Data, variable length             |
      +---------------+---------------+---------------+---------------+

      SPI                          NOCHANGE (DEF)
      Opaque Transform Data        RANDOM

   Everything after the SPI is encrypted, and is not compressed.


7.11.  UDP Header

   The UDP header is described in [RFC-768].

   The Next Header field (IPv6) or Protocol field (IPv4) in the
   preceding subheader is DEF.

      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |          Source Port          |       Destination Port        |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |            Length             |           Checksum            |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

      Source Port          NOCHANGE (DEF)
      Destination Port     NOCHANGE (DEF)
      Length               INFERRED
      Checksum             RANDOM

   The UDP header is compressed down to 2 octets, the UDP checksum.  The
   Length field of the UDP header must correspond to the Length field(s)
   of preceding subheaders.



Degermark, Nordgren, Pink                                      [Page 31]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


7.12.  TCP Header

   The TCP header is described in [RFC-793].

   The Next Header field (IbPv6) or Protocol field (IPv4) in the
   preceding subheader is DEF.

      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |          Source Port          |       Destination Port        |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                        Sequence Number                        |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                    Acknowledgment Number                      |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      | Offset| Reserved  |U|A|P|R|S|F|            Window             |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |           Checksum            |         Urgent Pointer        |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                    Options                    |    Padding    |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

   There are two ways to compress the TCP header. One uses the
   differential encoding techniques of Jacobsson, described in [RFC-
   1144], the only difference being the placement of the compressed TCP
   header fields (see section 6).

      Source Port           NOCHANGE  (DEF)
      Destination Port      NOCHANGE  (DEF)
      Sequence Number       DELTA
      Acknowledgment Number DELTA
      Offset                INFERRED
      Reserved              NOCHANGE
      U,A,P,R,S,F           see [RFC-1144]
      WINDOW                DELTA, see [RFC-1144]
      Checksum              RANDOM
      Urgent Pointer        see [RFC-1144]
      Options               see [RFC-1144]
      Padding               NOCHANGE

   The other method classifies the Source Port and Destination Port as
   NOCHANGE and the rest of the TCP header fields as RANDOM. This is a
   simple way to deal with links that reorder. The packet stream is
   considered to be a non-TCP packet stream (!), so a non-TCP CID should
   be used and format b), c), or d) must be used for the compressed
   header.  The rules for sending full headers in section 3.2.3 apply.

   Section 11 describes a method to deal with reordering that only adds
   two bytes to compressed headers.



Degermark, Nordgren, Pink                                      [Page 32]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


7.13.  IPv4 header [RFC-791, section 3.1]

   As we expect many IPv6 packets to be encapsulated in IPv4 packets,
   and many IPv4 packets to be encapsulated in IPv6 packets, it is
   important to be able to compress IPv4 headers.

       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
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |Version|  IHL  |Type of Service|          Total Length         |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |         Identification        |Flags|      Fragment Offset    |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |  Time to Live |    Protocol   |         Header Checksum       |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                       Source Address                          |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                    Destination Address                        |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                    Options                    |    Padding    |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

   There are two ways to compress the IPv4 header

   a) If the IPv4 header is not for a fragment (MF flag is not set and
      Fragment Offset is zero) and there are no options (IHL is 5), it
      is classified as follows

         Version              NOCHANGE   (DEF)
         IHL                  NOCHANGE   (DEF, must be 5)
         Type of Service      NOCHANGE
         Total Length         INFERRED   (from link-layer implementation
                                          or encapsulating IP header)

         Identification       DELTA/     (for TCP)
                              RANDOM     (for non-TCP)

         Flags                NOCHANGE   (MF bit must not be set)
         Fragment Offset      NOCHANGE   (must be zero)
         Time to Live         NOCHANGE
         Protocol             NOCHANGE
         Header Checksum      INFERRED   (calculated from other fields)
         Source Address       NOCHANGE   (DEF)
         Destination Address  NOCHANGE   (DEF)
         Options, Padding                (not present)






Degermark, Nordgren, Pink                                      [Page 33]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


   b) If the IPv4 header is for a fragment (MF bit set or Fragment
      Offset nonzero), or there are options (IHL > 5), all fields are
      RANDOM (i.e., they are sent as-is and not compressed).  If the
      IPv4 header is for a fragment it ends the compressible chain of
      subheaders, i.e., it is the last subheader to be compressed.  If
      the IPv4 header has options but is not for a fragment it does not
      end the compressible chain of subheaders, so subsequent subheaders
      will be compressed.

   A compressor that follows the guidelines of section 4.1 will in case
   a) use the Version, Source Address and Destination Address to define
   the packet stream, together with the fact that there are no IPv4
   options and that this is not a fragment.

   Case b) can define two kinds of packet streams depending on whether
   the IPv4 header is for a fragment or not.

   If the IPv4 header in case b) is for a fragment, the compressor uses
   that fact together with the Version, Source Address, and Destination
   Address to determine the packet stream.

   If the IPv4 header in case b) is not for a fragment, it must have
   options. The compressor uses that fact, but not the size of the
   options, together with the Version, Source Address, and Destination
   Address to determine the packet stream.


























Degermark, Nordgren, Pink                                      [Page 34]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


8.  Changing compression identifiers

   On a point-to-point link, the compressor has total knowledge of what
   CIDs are in use at the decompressor and can change what CID a packet
   stream uses or reuse CIDs at will. To get the highest possible
   compression rate, 4-bit CIDs should be allocated to the non-TCP
   packet streams with the highest packet rates.  However, changing a
   CID to a new packet stream involves paying the cost of at least one
   exponential backoff, so it should not be done too frequently.

   Each non-TCP CID is associated with a compression state with a
   generation value. To avoid too rapid generation wrap-around and
   potential incorrect decompression, an implementation must avoid
   wrap-around of the generation value in less than MIN_WRAP seconds
   (see section 12).

   To aid in avoiding wrap-around, the generation value associated with
   a non-TCP CID should be maintained when changing a CID to a new
   packet stream.  A compressor must increment the generation value by
   one when switching to a new non-TCP packet stream.

   An implementation must not reset the generation value when allocating
   a CID to a new packet stream.




























Degermark, Nordgren, Pink                                      [Page 35]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


9.  Rules for dropping or temporarily storing packets

   When a decompressor receives a packet with a compressed TCP header
   with CID C, it must be discarded when the compression state for C has
   not been initialized by a full header.

   When a decompressor receives a packet with a compressed non-TCP
   header with CID C and generation G, the header must not be
   decompressed using the current compression state when

      a) the decompressor has been disconnected from the compressor for
         more than MIN_WRAP seconds, because the compression state might
         be obsolete even if it has generation G.

      b) the compression state for C has a generation other than G.

   In case a) and b) the packet can either be

     i)  discarded immediately, or else

     ii) stored temporarily until the compression state is updated by a
         packet with a full non-TCP header with CID C and generation G,
         after which the header can be decompressed.

         Packets stored in this manner must be discarded when

           *)  receiving full or compressed non-TCP headers with CID C
               and a generation other than G,

           *)  the decompressor has not received packets with CID C in
               the last MIN_WRAP seconds.

   When full headers are lost, a decompressor may receive compressed
   non-TCP headers with a generation value other than the generation of
   its compression state.  Rule ii) allows the decompressor to store
   such headers until they can be decompressed using correct compression
   state.














Degermark, Nordgren, Pink                                      [Page 36]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


10. Low-loss header compression for TCP

   Since fewer bits are transmitted per packet with header compression,
   the packet loss rate is lower with header compression than without,
   for a fixed bit-error rate.  This is beneficial for links with high
   bit-error rates such as wireless links.

   However, since TCP headers are compressed using differential
   encoding, a single lost TCP segment can ruin an entire TCP sending
   window because the compression state is not incremented properly at
   the decompressor.  Subsequent headers will therefore be decompressed
   to be different than before compression and discarded by the TCP
   receiver.  A TCP connection in the wide area where the last hop is
   over a medium-speed lossy link, for example a wireless LAN, can have
   poor performance with header compression because the sending window
   is relatively large and the bit-error rate high.

   This section describes two simple mechanisms for quick repair of the
   compression state. With these mechanisms header compression will
   improve TCP throughput over lossy links as well as links with low
   bit-error rates.

10.1.  The "twice" algorithm

   The decompressor can compute the TCP checksum to determine if its
   compression state is not updated properly. If the checksum fails, the
   error is  assumed to caused by a lost segment that did not update the
   compression state properly. The delta of the current segment is then
   added to the compression state again on the assumption that the lost
   segment contained the same delta as the current. By decompressing and
   computing the TCP checksum again, the decompressor checks if the
   repair succeeded or if the delta should be applied once more.

   Analysis of traces of various TCP bulk transfers show that applying
   the delta of the current segment one or two times will repair the
   compression state for between 83 and 99 per cent of all single-
   segment losses in the data stream. For the acknowledgment stream, the
   success rate is smaller due to the delayed ack mechanism of TCP. The
   "twice" mechanism repairs the compression state for 99 - 53 per cent
   of the losses in the acknowledgment stream.  A sophisticated
   implementation of this idea would determine whether the TCP stream is
   an acknowledgment or data stream and determine the segment size by
   observing the stream of full and compressed headers.  Trying deltas
   that are small multiples of the segment size will result in even
   higher repair success rates for acknowledgment streams.






Degermark, Nordgren, Pink                                      [Page 37]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


10.2.  Header Requests

   The relativley low success rate for the "twice" algorithm for TCP
   acknowledgment streams calls for an additional mechanism for
   repairing the compression state at the decompressor. When the
   decompressor fails to repair the compression state after a loss, the
   decompressor requests a full header from the compressor.

               Node A          lossy link          Node B
         ------------------                  ------------------
        |                  |      Acks      |                  |
      >--->  Compressor  >--------------------> Decompressor >---->
        |                  |                |                  |
        |                  |      Data      |                  |
      <---< Decompressor <--------------------<  Compressor  <----<
        |                  |                |                  |
         ------------------                  ------------------

   The most common configuration is likely to be that the TCP
   acknowledgment and data streams pass through the same nodes on each
   side of a lossy link such as wireless.  There will then be a
   compressor/decompressor pair on each side of the link.

   Assume that an acknowledgement is damaged on the lossy link from node
   A to node B. The link-level checksum detects the damaged frame and
   discards it. Also, assume that the decompressor in node B fails to
   repair its compression state when the next compressed acknowledgment
   arrives. The decompressor in node B will then ask the compressor in
   node B to set a bit in a full header in the corresponding TCP stream
   going in the opposite direction, in this case the data stream. When
   the decompressor in node A sees the bit, it asks its companion
   compressor to send a full header in the packet stream with the
   corresponding IP addresses and port numbers.  The full header updates
   the compression state for the decompressor in node B and
   acknowledgments start to flow again.

   In this manner the TCP acknowledgment stream is repaired after a
   roundtrip-time over the lossy link. A wide-area TCP connection is not
   likely to get a timeout signal during this time, and the TCP transfer
   can continue without backing off.

   The header request mechanism will not work when routes are not
   symmetric and the TCP streams do not visit the same nodes. This
   situation is perfectly normal. Sending a header request is a hint
   that it is a good idea to send a full header in the corresponding TCP
   packet stream. However, header requests is only a performance
   improving mechanism and it is safe to ignore such hints if no
   corresponding TCP packet stream can be found.  Compressors must



Degermark, Nordgren, Pink                                      [Page 38]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


   silently ignore requests for full headers in TCP streams they are not
   compressing. Decompressors must ignore requests for full headers when
   they cannot contact a suitable compressor.

   What bit to use for requesting a full header is specified in section
   12.  The corresponding TCP stream going in the opposite direction is
   identified by examining the source and final destination addresses of
   the innermost IP header plus the TCP source and destination port
   numbers.  If one TCP stream has the values (SA1, DA1, SP1, DP1) in
   these fields and another has the values (SA2, DA2, SP2, DP2), they
   correspond to each other if and only if SA1 = DA2, DA1 = SA2, SP1 =
   DP2, and DP1 = SP2.

11.  Links that reorder packets

   Some links reorder packets, for example multi-hop radio links that
   use deflection routing to route around congested nodes.  Packets
   routed different ways can then arrive at the destination in a
   different order than they were sent.

11.1.  Reordering in non-TCP packet streams

   Compressed non-TCP headers do not change the compression state, and
   neither do full headers that refresh it.  It is only when a full
   header that changes the compression state arrives out of order that
   there can be problems.  There are two cases:

      - A packet with a full header with generation G arrives *after* a
         packet with a compressed header with generation G.  This case
         is covered by rule b) ii) in section 9.

      - A packet with a full header with generation G arrives *before* a
         packet with a compressed header with generation G-1 (modulo
         128).  The decompressor can then keep both versions of the
         compression state around for a while to be able to decompress
         subsequent compressed headers with generation G-1 (modulo 128).
         The old compression state must be discarded after MIN_WRAP
         seconds.

11.2.  Reordering in TCP packet streams

   Section 7.12 describes a simple method to deal with reordering in TCP
   streams.  It requires using the methods from 11.1 for dealing with
   reordering in non-TCP streams.  Compressed headers will typically be
   18 bytes with that method, which is significantly larger than the
   usual 4-6 bytes.

   To achieve better compression rates the following method, adding only



Degermark, Nordgren, Pink                                      [Page 39]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


   two bytes to the compressed header for a total of 6-8 bytes, can be
   used.  A packet sequence number, incremented by one for every packet
   in the TCP stream, is associated with each compressed and full
   header.  This allows the decompressor to place the packets in the
   correct sequence and it can apply their deltas to the compression
   state in the correct order.  A simple sliding window scheme can be
   used to place the packets in the correct order.

   Two bytes are needed for the packet sequence numbers.  One byte gives
   only 256 sequence numbers.  In a sliding window scheme the window
   should be no larger than half of the sequence number space, so
   packets can not arrive more than 127 positions out-of-sequence. This
   is equivalent to a delay of 260 ms on 2 Mbit/s links with 512 byte
   segments.  Delays of that order are not uncommon on multi-hop radio
   links.  However, two bytes giving 2^16 = 65536 values should be
   sufficient.

   Full TCP headers will only have space for one byte of sequence number
   when there is no tunneling. It is not feasible to increase the size
   of full headers since the packet size might be optimized for the MTU
   of the link. Therefore only the least significant byte of the packet
   sequence number can be placed in such full headers. We believe that
   such full headers can be positioned correctly frequently enough with
   only the least significant byte of the packet sequence number
   available.  Experiments are needed to confirm this.

   We skip over packet sequence numbers where the least significant byte
   is zero. This allows the compressor to signal that it is adding
   packet sequence numbers to compressed headers by placing a value
   other than zero in a full header.

   Avoiding zero also takes care of a problem that can occur when the
   TCP window scale option is used to enlarge the TCP window. When
   exactly 2^16 bytes of TCP data is lost, a compressed header will be
   decompressed incorrectly without being detected by the TCP checksum.
   TCP segments are usually a power of two.  So by using a packet
   sequence number space that is not a power of two either the sequence
   number or the packet sequence number will differ when 2^16 bytes are
   lost. When a compressor sees the window scale option on a SYN
   segment, it must use packet sequence numbers when subsequently
   compressing that packet stream.

   In compressed TCP headers the two byte packet sequence number is
   placed immediately after the CID. The CID must not be omitted when
   packet sequence numbers are used.  See section 5.3 for placement of
   packet sequence numbers in full headers.





Degermark, Nordgren, Pink                                      [Page 40]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


12.  Demultiplexing

   It is necessary to distinguish packets with regular IPv4 headers,
   regular IPv6 headers, full IPv6 packets, full IPv4 packets,
   compressed TCP packets, and compressed non-TCP packets. It is also
   desirable to find one bit for requesting a header refresh as
   described in section 11.2.

   The decision to use a distinct ethertype (or equivalent) for IPv6 has
   already been taken, which means that link-layers must be able to
   indicate that a packet is an IPv6 packet.

   IPv6 header compression requires that the link-layer implementation
   can indicate two new kinds of packets: COMPRESSED_TCP for format a)
   in section 6, and COMPRESSED_NON_TCP for formats b) and c) in section
   6.

   Full headers and header requests are indicated by special encodings
   of the first four bits (Version field) in a packet indicated to be an
   IPv6 packet.

      Version   Type of header
      -------   --------------

      0110      regular IPv6 header

      1100      full IPv4 header
      1101      full IPv4 header, requests full header

      1110      full IPv6 header
      1111      full IPv6 header, requests full header


   If the link-layer cannot indicate these packet types, all packets
   with compressed headers must start with a byte indicating the packet
   type, followed by the compressed header.

      First byte   Type of compressed header
      ----------   -------------------------

      00000000     compressed TCP header (format a)
      00000001     compressed non-TCP header (formats b or c)









Degermark, Nordgren, Pink                                      [Page 41]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


13.  Configuration Parameters

   The following parameters are fixed for all implementations of this
   header compression scheme.

      MIN_WRAP     - minimum time of generation value wrap around

           3 seconds.

   The following parameters can be negotiated between the compressor and
   decompressor. If not negotiated their values must be as specified by
   DEFAULT.

      F_MAX_PERIOD - Largest number of compressed headers that may be
                     sent without a header refresh.

           DEFAULT is 256


      F_MAX_TIME   - Compressed headers may not be sent more than
                     F_MAX_TIME seconds after sending last full header.

           DEFAULT is 5 seconds

           NOTE:  F_MAX_PERIOD and F_MAX_TIME should be lower when it is
                  likely that a decompressor loses its state.


      MAX_HEADER   - The largest header size (in 8-octet units) that may
                     be compressed.

           DEFAULT is 25  (200 octets).

           Minimum is 16  (128 octets), which covers

                           IPv6 Base Header
                           Routing Header with two addresses
                           Keyed MD5 Authentication Header
                           Minimum TCP Header


      TCP_SPACE    - Size of allowed CID space for TCP streams.

           TCP_SPACE is the 2-logarithm of size of TCP CID space.

           DEFAULT is 4, giving a maximum TCP CID of 2^4-1 = 15

           TCP_SPACE must be at least 2 and at most 8.



Degermark, Nordgren, Pink                                      [Page 42]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


      NON_TCP_SPACE    - Size of allowed CID space for non-TCP streams.

           NON_TCP_SPACE is the 2-logarithm of the size of the
                   non-TCP CID space.

           DEFAULT is 4, giving a maximum non-TCP CID of 2^4-1 = 15

           NON_TCP_SPACE must be at least 2 and at most 24.


      ENABLE_CID_COMPRESSION  - may TCP CIDs be compressed away?

           DEFAULT no.


      EXPECT_REORDERING       - The mechanisms in section 11 are used.

           DEFAULT no.

































Degermark, Nordgren, Pink                                      [Page 43]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


14.  Implementation Status

   A prototype using UDP as the link layer has been operational since
   March 1996. We will have a NetBSD implementation for PPP for the
   Montreal IETF meeting, and will issue a revised draft describing the
   PPP specifics in July.

15.  Acknowledgments

   This protocol uses many ideas originated by Van Jacobson in the
   design of header compression for TCP/IP [RFC-1144].

   We thank Craig Partridge for pointing out a problem that can occur
   when the TCP window scale option is used.  A solution to this problem
   relying on the packet sequence numbers used for reordering is
   described in section 11.2.

16.  Security Considerations

   We advise against identifying packet streams with the aid of
   information that is encrypted even if such information happens to be
   available to the compressor. Doing so would expose traffic patterns.

17.  Author's Addresses

   Mikael Degermark                            Tel: +46 920 91188
   CDT/Dept of Computer Science                Fax: +46 920 72191
   Lulea University                            EMail: micke@sm.luth.se
   S-971 87 Lulea, Sweden

   Bjorn Nordgren                              Tel: +46 920 75400
   CDT/Telia Research AB                       Fax: +46 920 75490
   Aurorum 6                                   EMail: bcn@lulea.trab.se
   S-977 75 Lulea, Sweden

   Stephen Pink                                Tel: +46 8 752 15 59
   CDT/Swedish Institute of Computer Science   Fax: +46 8 751 72 30
   PO Box 1263                                 Mobile: +46 70 532 0007
   S-164 28 Kista, Sweden                      EMail: steve@sics.se












Degermark, Nordgren, Pink                                      [Page 44]


INTERNET-DRAFT        Header Compression for IPv6          June 13, 1996


18.  References

   [RFC-768]   J. Postel, User Datagram Protocol, RFC 761, August 1980.

   [RFC-791]   J. Postel, Internet Protocol, RFC 791, September 1981.

   [RFC-793]   J. Postel, Transmission Control Protocol, RFC 793,
               September 1981.

   [RFC-1144]  V. Jacobson, Compressing TCP/IP Headers for Low-Speed
               Serial Links, RFC 1144, February 1990.

   [RFC-1553]  A. Mathur, M. Lewis, Compressing IPX Headers Over WAN
               Media (CIPX), RFC 1553, December 1993.

   [RFC-1700]  J. Reynolds and J. Postel, Assigned Numbers, RFC-1700,
               October 1994.

   [RFC-1826]  R. Atkinson, IP Authentication Header, RFC 1826, August
               1995.

   [RFC-1827]  R. Atkinson, IP Encapsulating Security Protocol (ESP),
               RFC 1827, August 1995.

   [RFC-1828]  Metzger, W. Simpson, IP Authentication using Keyed MD5,
               RFC 1828, August 1995.

   [IPv6]      S. Deering, R. Hinden, Internet Protocol, Version 6
               (IPv6) Specification, RFC 1883, December 1995.

   [ICMPv6]    A. Conta, S. Deering, Internet Control Message Protocol
               (ICMPv6) for the Internet Protocol Version 6 (IPv6), RFC
               1885, December 1995.




This draft expires December 18, 1996













Degermark, Nordgren, Pink                                      [Page 45]