Ericsson

8400 boulevard Decarie Montreal, QC H4P 2N2 Canada daniel.migault@ericsson.com

LMU Munich

Oettingenstr. 67 80538 Munchen Germany guggemos@nm.ifi.lmu.de http://www.nm.ifi.lmu.de/~guggemos

Universitaet Bremen TZI

Postfach 330440 Bremen D-28359 Germany +49-421-218-63921 cabo@tzi.org

Google LLC

1600 Amphitheatre Parkway Mountain View California 94043 USA dschinazi.ietf@gmail.com

Security ipsecme With the use of encrypted ESP for secure IP communication, the compression of IP payload is only possible with complex frameworks, such as RObust Header Compression (ROHC). Such frameworks are too complex for numerous use cases and especially for IoT scenarios, which makes IPsec not being used here, although it offers architectural benefits. ESP Header Compression (EHC) defines a flexible framework to compress communications protected with IPsec/ESP. Compression and decompression is defined by EHC Rules orchestrated by EHC Strategies. The necessary state is hold within the IPsec Security Association and can be negotiated during key agreement, e.g. with IKEv2. The document specifies the necessary parameters of the EHC Context to allow compression of ESP and the most common included protocols, such as IPv4, IPv6, UDP and TCP and the corresponding EHC Rules. It also defines the Diet-ESP EHC Strategy which compresses up to 32 bytes per packet for traditional IPv6 VPN and up to 66 bytes for IPv6 VPN sent over a single TCP or UDP session.

The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 when, and only when, they appear in all capitals, as shown here.

IPsec/ESP secures communications either using end-to-end security or by building a VPN, where the traffic is carried to a secure domain via a security gateway. IPsec/ESP was not designed to minimize its associated networking overhead. In fact, bandwidth optimization often adds computational overhead that may negatively impact large infrastructures in which bandwidth usage is not a constraint. On the other hand, in IoT communications, sending extra bytes can significantly impact the battery life of devices and thus the life time of the device. The document describes a framework that optimizes the networking overhead associated to IPsec/ESP for these devices. Most compression mechanisms work with dynamic compression contexts. Some mechanisms, such as ROHC , agree and dynamically change the context over a dedicated channel. Others, such as 6LowPAN, send the context together with the actual protocol information in a separate compression header. Those mechanism fail when it comes to compress encrypted payloads as appearing in ESP. This is found to be a major reason, why IPsec and in particular ESP is not widely developed in environments where bandwidth saving is a critical task, such as in IoT scenarios. ESP Header Compression (EHC) chooses another form of context agreement, which is similar to the one defined by Static Context Header Compression (SCHC). It works with a static compression context agreed for a specific Security Association. The context itself can be negotiated during the key agreement, which allows only minimal the changes to the actual ESP implementation. EHC itself is defined as a framework that specifically compresses ESP protected communications. EHC is highly flexible to address any use case where compression is necessary. EHC takes advantage of the negotiation between the communication endpoint to agree on the cryptographic parameters, which in some cases already includes parameters that remain constant during the communications (like layer 4 ports, or IP addresses) and can thus be used as part of the compression context. Only additional, EHC specific parameters need to be agreed for the purpose of compression. In addition EHC Rules define how fields may be compressed and decompressed given the provided parameters. Finally, EHC defines EHC Strategy which defines how a set of EHC Rule is coordinated. This document specifies EHC Context parameters for the most common Layer 3 and 4 protocols and the associated EHC Rules. Additionally, an EHC Strategy called Diet-ESP is defined, which compresses up to 32 bytes per packet for traditional VPN and up to 66 bytes for VPN set over a single TCP or UDP session. Its main purpose is a maximum level of compression with a minimum of additional agreement. This is achieved by defining a default usage of existing IPsec SA parameters wherever possible.

This document uses the following terminology: ESP Header Compression Internet of Things If not stated otherwise, IP means IPv6. Least Significant Bytes Most Significant Bytes IPsec Security Association Database IPsec Security Association IPsec Security Policy Database IPsec Traffic Selector ESP Security Parameter Index ESP Sequence Number ESP Padding ESP Pad Length Next Header Initialization Vector Implicit Initialization Vector Integrity Check Value Virtual Private Network

ESP Header Compression (EHC) compresses IPsec ESP packets, thus reducing the size of the packet sent on the wire, while carrying an equivalent level of information with an equivalent level of security. EHC is able to compress any protocol encapsulated in ESP and ESP itself. Concerned fields include those of the ESP protocol, as well as other protocols in the ESP payload such as the IP header when the tunnel mode is used, but also upper layer protocols, such as the UDP or the TCP header. Non ESP fields may be compressed by ESP under certain circumstances, but EHC is not intended to provide a generic way outside of ESP to compress these protocols. Compression of the unprotected IP header and the unencrypted ESP header may be performed by mechanism such as 6LoWPAN , SCHC , ROHC or 6LoWPAN-GHC . EHC is based on a static compression context, EHC Rules coordinated by an EHC Strategy: Stores the information of a specific header field which can be compressed by EHC. This can be specific header values such as IP addresses or L4 ports do not have to be send on the wire at all, or compression information for fields which can be partially compressed, such as sequence numbers. Defines how the information of the EHC Context is used to compress a specific field. It defines compression functions, such as "elided", "least significant byte" and others, being applied on the header field. Is applied to efficiently coordinate EHC Context and EHC Strategy. The EHC Strategy "Diet-ESP" defined in this document utilizes the information in the IPsec SA to pre-define the EHC Context without explicitly exchanging the EHC Context. As depicted in , the EHC Strategy - Diet-ESP in our case - and the EHC Context are agreed upon between the two peers, e.g. during key exchange. The EHC Rules are to be implemented on the peers and do not require further agreement.

EHC Context | | EHC Rules | EHC Rules | | | | | v v v v +====================+ +====================+ | ESP | | ESP | +====================+ +====================+ | < pre-esp > | | < pre-esp > | +--------------------+ +--------------------+ | < clear text esp > | | < clear text esp > | +--------------------+ +--------------------+ | < encryption > | | < encryption > | +--------------------+ +--------------------+ | < post-esp > | | < post-esp > | +--------------------+ +--------------------+]]>

In , the ESP stack is represented by various sub layers describing the packet processing inside the ESP: represents treatment performed to a non ESP packet, i.e. before ESP encapsulation or decapsulation is being performed. Any compression of protocols not specific to but encrypted by ESP, such as L4 and higher protocols, is performed here. designates the ESP encapsulation / decapsulation processing performed on an non encrypted ESP packet. This layer includes compression for fields which are included during the ESP encapsulation. A typical example is the later encrypted Tunnel IP header and the fields of the ESP trailer. designates the encryption/decryption phase This layer could include compression of encryption information (e.g. Initialization Vector, etc.), but this is currently out of scope of this document. the processing performed on an ESP encrypted packet. This layer includes compression of the ESP header. EHC Rules may be processed at any of these layers and thus impact differently the standard ESP. More specifically, EHC Rules performed at the "pre-esp" or "post-esp" layer do not require the current ESP stack to be updated and can simply be appended to the current ESP stack. On the other hand, EHC Rules at the "clear text esp" may require modification of the current ESP stack. The set of EHC rules described in this document as well as the EHC Strategies may be extended in the future. Nothing prevents such EHC Rules and Strategies to be updated.

Signalling the compression of a certain ESP packet is crucial for correct decompression at the sender. Situation where decompression may fail unforeseen are various, such as IP fragmentation, UDP options just to name a few. With EHC, the agreement of the level or occurrence of compression is left the negotiation protocol (e.g. IKEv2). In order to achieve per-packet signalization of the compression level, this document proposes new IPsec modes "Compressed Transport" and "Compressed Tunnel", which are meant to be agreed during the negotiation of the EHC Contex and EHC Strategy. This can lead to multiple SAs, and thus, multiple SPIs for different levels of compression agreed with the EHC Context. The receiver can detect the level of compression of an incoming packet by looking up the used EHC Context and EHC strategy in the corresponding SA. If the sender detects that the de-compression can not be guaranteed with a given EHC Context and EHC Strategy, it MUST NOT apply compression. If an SA with IPsec Mode "Tunnel"/"Transport" is available, the sender SHOULD send the packet uncompressed, rather than discard the packet. When there is no uncompressed SA available, the packet MUST be dropped.

The EHC Context provides the necessary information so the two peers can proceed to the appropriated compression and decompression defined by the EHC Strategy. The EHC Context is defined on a per-SA basis. A context can be defined for any protocol encapsulated with ESP and for ESP itself. For each header field, a context attribute is provided to the EHC Context in order to allow compression and decompression. Most power of EHC lies in the fact, that the attributes for some protocols are already available in the IPsec SA (e.g. IP addresses in the Traffic Selector). Such attributes are designated by "Yes" in the "In SA" column. All others need to be negotiated separately in order to allow EHC to work properly. As this document is limited to the Diet-ESP strategy, the EHC Context in this section used by the Diet-ESP Strategy to activate specific EHC Rules as well as to execute the EHC Rules by providing the necessary parameters..

Context Attribute In SA Possible Values ipsec_mode Yes "Tunnel", "Transport" outer_version Yes "IPv4", "IPv6" esp_spi Yes ESP SPI esp_spi_lsb No 0, 1, 2, 3, 4 esp_sn Yes ESP Sequence Number esp_sn_lsb No 0, 1, 2, 3, 4 esp_sn_gen No "Time", "Incremental" esp_align No 8, 16, 24, 32 esp_encr Yes ESP Encryption Algorithm

Parameters associated to the Inner IP addresses are only specified when the SA has been configured with the tunnel mode. As a result when ipsec_mode is set to "Transport" the parameters below MUST NOT be considered and are considered as "Undefined" Context Attribute In SA Possible Values ip_version Yes "IPv4", "IPv6"

Context Attribute In SA Possible Values ip6_tcfl_comp No "Outer", "Value", "UnComp" ip6_tc No IPv6 Traffic Class ip6_fl No IPv6 Flow Label ip6_hl_comp No "Outer", "Value", "UnComp" ip6_hl No Hop Limit Value ip6_src Yes IPv6 Source Address ip6_dst Yes IPv6 Destination Address ip6_tcfl_comp indicates how Traffic Class and Flow Label fields of the inner IP Header are expected to be compressed. "Outer" indicates Traffic Class and Flow Label are read from the outer IP header, "Value" indicates these values are provided by the Diet-ESP Context, while "Uncompress" indicates that no compression occurs and these values are read in the inner IP inner header. ip6_hl_comp indicates how Hop Limit field of the inner IP Header is expected to be compressed. (see ip6_tcfl_comp). ip6_dst designates the Destination IPv6 Address of the inner IP header. The IP address is provided by the TS, and can be defined as a range of IP addresses. Compression is only considered when ip6_dst indicates a single IP Address. When the TS defines more than a single IP address ip6_dst is considered as "Unspecified" and its value MUST NOT be considered for compression.

Context Attribute In SA Possible Values ip4_options No "Options", "No_Options" ip4_id No IPv4 Identification ip4_id_lsb No 0,1,2 ip4_ttl_comp No "Outer", "Value", "UnComp" ip4_ttl No IPv4 Time To Live ip4_src Yes IPv4 Source Address ip4_dst Yes IPv4 Destination Address ip4_frag_enable No "True", "False" ip4_options specifies if the IPv4 header contains any options. If set to "No_Options", the first 8 bit of the IPv4 header (being the IP version and IP header length) are compressed. If set to "Options" this bits are sent uncompressed. ip4_ttl indicates how the Time To Live field of the inner IP Header is expected to be compressed. (see ip6_hl_comp).

The following parameters are provided by the SA but the SA may specify single value or a range of values. When the SA specifies a range of values, these parameters MUST NOT be considered and are considered as Unspecified. Context Attribute In SA Possible Values l4_proto Yes IPv6/ESP Next Header,IPv4 Protocol l4_src Yes UDP/UDP-Lite/TCP Source Port l4_dst Yes UDP/UDP-Lite/TCP Destination Port

For UDP, there are no additional parameters necessary than the ones in

Context Attribute In SA Possible Values udplite_coverage No 8-6535, "Length", "uncompressed" udplite_coverage: For UDP-Lite, the checksum can have different coverages, which is defined by the "Checksum Coverage" field which replaces the "Length" field of UDP. This context field defines the coverage in advance by either a specific value (8-16535), the actual length of the UDP-Lite payload ("Length" or 0) or as uncompressed. Note that udplite coverage is indicated on a packet basis and cannot be greater than the UDP length. In this case udplite_coverage is negotiated for all packets and the actual coverage for a given UDP packet is derived as the minimum value between udplite_coverage and the length of the UDP packet.

Context Attribute In SA Possible Values tcp_sn No TCP Sequence Number tcp_ack No TCP Acknowledgment Number tcp_lsb No 0, 1, 2, 3, 4 tcp_options No "True", "False" tcp_urgent No "True", "False" tcp_sn holds the current Sequence Number of the TCP session. tcp_ack holds the current Acknowledgement Number of the TCP session. tcp_lsb holds the number of lsb of tcp_sn and tcp_ack sent on the wire. tcp_options says if options are enabled in the current TCP session. If tcp_options is set to "False" the Options field in TCP can be elided. tcp_urgent says if the urgent pointer is enabled in the current TCP session. If tcp_urgent is set to "False" the Urgent Pointer field in TCP can be elided.

This section describes the EHC Rules involved in Diet-ESP. The EHC Rules defined by Diet-ESP may be used in the future by EHC Strategies other than Diet-ESP, so they are described in an independent way. A EHC Rule defines the compression and decompression of one or more fields and EHC Rules are represented this way:

The EHC Rule is designated by a name (EHC_RULE_NAME) and the concerned Fields (f1, ..., fm). Each field compression and decompression is represented by an Action (a1, ..., am). The Parameters indicate the necessary parameters for the action to perform both the compression and the decompression. The table below provides a high level description of the Actions used by Diet-ESP. As these Action may take different arguments and may operate differently for each field a compete description is provided in the next sections as part of the EHC Rule description. Function Compression Decompression send-value No No elided Not send Get from EHC Context lsb(_lsb_size) Sent LSB Get from EHC Context lower Not send Get from lower layer checksum Not send Compute checksum. padding(_align) Compute padding Get padding send-value designates an action that does not perform any compression or decompression of a field. elided designates an action where both peers have a local value of the field. The compression of the field consists in removing the field, and the decompression consists in retrieving the field value from a known local value. The local value may be stored in a EHC Context or defined by the EHC Rule (like a zero value for example). lsb designates an action where both peers have a local value of the field, but the compression consists in sending only the LSB bytes instead of the whole field. The decompression consists in retrieving the field from the LSB sent as well as some other additional local values. lower designates an action where the compression consists in not sending the field. The decompression consists in retrieving the field from the lower layers of the packet. A typical example is when both IP and UDP carry the length of the payload, then the length of the UDP payload can be inferred from the one of the IP layer. checksum designates an action where the compression consists in not sending a checksum field. The decompression consists in re-computing the checksum. ESP provides an integrity-check based on signature of the ESP payload (ICV). This makes removing checksum possible, without harming the checksum mechanism. padding designates an action that computes the padding of the ESP packet. The function is specific to the ESP. For all actions, the function can be performed only when the appropriated parameters and fields are provided. When a field or a parameters does not have an appropriated value its value is designated as "Unspecified". Specifically some fields such as inner IP addresses, ports or transport protocols are agreed during the SA negotiation and are specified by the SA. Their value in the SA may take various values that are not appropriated to enable a compression. For example, when these fields are defined as a range of values, or by selectors such as OPAQUE or ANY these fields cannot be retrieved from a local value. Instead, when they are defined as a "Single" value (i.e a single IP address, or a single port number or a single transport protocol number) compression and decompression can be performed. These SA related fields are considered as "Unspecified" when not limited to a "Single" value. When a field or a parameter is "Unspecified", the EHC Rule MUST NOT be activated. This is the purpose of the EHC Strategy to avoid ending in such case. In any case, when one of these condition is not met, the EHC Rule MUST NOT perform any compression or decompression action and the packet MUST be discarded. When possible, an error SHOULD be raised and logged.

This section describes the EHC Rules for ESP which are summed up in the table below. EHC Rule Field Action Parameters ESP_SPI SPI lsb esp_spi_lsb, esp_spi ESP_SN Sequence Number lsb esp_sn_lsb, esp_sn_gen, esp_sn ESP_NH Next Header elided l4_proto, ipsec_mode ESP_PAD Pad Length, Padding padding esp_align, esp_encr ESP_SPI designates the EHC Rule compressing / decompressing the SPI. ESP_SPI is performed in the "post-esp" phase. The SPI is compressed using "lsb". The sending peer only places the LSB bytes of the SPI and the receiving peer retrieve the SPI from the LSB bytes carried in the packets as well as from the SPI value stored in the SA. The SPI MUST be retrieved as its full value is included in the signature check. The two peers MUST agree on the number of LSB bytes to be sent: "esp_spi_lsb". Upon agreeing on "esp_spi_lsb", the receiving peer MUST NOT agree on a value not carrying sufficient information to retrieve the full SPI. ESP_SN designates the EHC Rule compressing / decompressing the ESP Sequence Number. ESP_SN is performed in the "post-esp" phase. ESP_SN is only activated if the SN ("esp_sn"), the LSB significant bytes ("esp_sn_lsb") and the method used to generate the SN ("esp_sn_gen") are defined. The Sequence Number is compressed using "lsb". Similarly to the SPI, the Sequence Number MUST be retrieved in order to complete the signature check of the ESP packet. Unlike the SPI, the Sequence Number is not agreed by the peers, but is changing for every packet. As a result, in order to retrieve the Sequence Number from the LSB "esp_sn_lsb", the peers MUST agree on generating Sequence Number in a similar way. This is negotiated with "esp_sn_gen" and the receiver MUST ensure that "esp_sn_lsb" is big enough to absorb minor packet losses or time differences between the peers. ESP_NH designates the EHC Rule compressing / decompressing the ESP Next Header. ESP_NH is performed in the "clear text esp" phase. ESP_NH is only activated if the Next Header is specified. The Next Header can be specified as IP (IPv4 or IPv6) when the IPsec tunnel mode is used ("ipsec_mode" set to "Tunnel") or when the transport mode ("ipsec_mode" set to "Transport") is used when the Traffic Selector defines a "Single" Protocol ID ("l4_proto"). The Next Header, is compressed using "elided". The Next Header indicates the Header in the Payload Data. When the Tunnel mode is chosen, the type of the header is known to be an IP header. Similarly, the TS may also hold transport layer protocol, which specifies the Next Header value for Transport mode. The Next Header value is only there to provide sufficient information for decapsulating ESP. In other words decompressing this fields would occur in the "clear text esp" phase and striped but directly removed again by the ESP stack. For these reasons, implementation may simply omit decompressing this field. ESP_PAD designates the EHC Rule compressing / decompressing the Pad Length and Padding fields. ESP_PAD is performed in the "clear text esp" phase. Pad Length and Padding define the padding. The purpose of padding is to respect a 32 bit alignment for ESP or block sizes of the used cryptographic suite. As the ESP trailer is encrypted, Padding and Pad Length MUST to be performed by ESP and not by the encryption algorithm. Thus, ESP_PAD always needs to respect the cipher alignment ("esp_encr"), if applicable. Compression may be performed especially when device support alignment smaller than 32 bit. Such alignment is designated as "esp_align" and the padding bytes are the necessary bytes so the ESP packet has a length that is a multiple of "esp_align". When "esp_align" is set to an 8-bit alignment padding bytes are not necessary, and Padding as well as Pad Length are removed. For values that are different from 8-bit alignment, padding bytes needs to be computed according to the ESP packet length why ESP_PAD MUST be the last action of "clear text esp". The resulting number of padding byte is then expressed in Padding and Pad Length fields with Pad Length set to padding bytes number - 1 and Padding is generated as described in . Combining the Pad Length and Padding fields could potentially add an overhead on fixed size padding. In fact some applications may only send the same type of fixed size data, in which case the Pad Length would not be necessary to be specified. However, the only corner case Pad Length fields would actually add an overhead is when padding is expected to be of zero size. In this case, specifying an 8-bit alignment solve this issue.

All IPv4 EHC Rules MUST be performed during the "clear text esp" phase. The EHC Rules are only defined for compressing the inner IPv4 header and thus can only be used when the SA is using the Tunnel mode. EHC Rule Field Action Parameters IP4_OPT_DIS Version elided ip_version Header Length elided IP4_LENGTH Total Length lower IP4_ID Identification lsb ip4_id, ip4_id_lsb IP4_FRAG_DIS Flags elided Fragment Offset elided IP4_TTL_OUTER Time To Live elided ip4_ttl IP4_TTL_VALUE Time To Live elided ip4_ttl IP4_PROT Protocol elided l4_proto IP4_CHECK Header Checksum checksum IP4_SRC Source Address elided ip4_src IP4_DST Dest. Address elided ip4_dst IP4_OPT_DIS designates that the IPv4 header does not include any options and indicates if the first byte of the IPv4 header - consisting of IP version and IPv4 Header Length, are compressed. The Version "ip_version" is defined by the SA and is thus compressed using "elided". If the header does not contain any options, it is compressed with "elided" and decompressed to "20", the default length of the IPv4 header. If the header does contains some options, the length is not compressed. IP4_LENGTH designates the EHC Rule compressing / decompressing the Total Length Field of the inner IPv4 header. The Total Length is compressed by the sender and not sent. The receiver decompresses it by recomputing the Total Length from the outer IP header. The outer IP header can be IPv4 or IPv6 and IP4_LENGTH MUST support both versions if both versions are supported by the device. Note that the length of the inner IP payload may also be subject to updates if decompression of the upper layers occurs. IP4_ID designates the EHC Rule compressing / decompressing the Identification Field. IP4_ID is only activated if the ID ("ip4_id"), the LSB significant bytes ("ip4_id_lsb") are defined. Upon agreeing on "ip4_id_lsb", the receiving peer MUST NOT agree on a value not carrying sufficient information to retrieve the full IP Identification. Note also that unlike the ESP SN, the IPv4 Identification is not part of the SA. As a result, when the ID is compressed, its value MUST be stored in the EHC Context. The reserved attribute for that is "ip4_id" IP4_FRAG_DIS designates that the inner IPv4 header does not support fragmentation. If activated, IP4_FRAG_DIS indicates compression of Flags and Fragment Offset field in the IPv4 header which consists of 2 bytes. Both fields are compressed with "elided" and decompressed with their default value according to , which is 0b010 for Flags and 0 for Fragment Offset. IP4_TTL_OUTER designates an EHC Rule compressing / decompressing the Time To Live field of the inner IP header. If the outer IP header is an IPv6 header, the Hop Limit is used for decompression. The Time To Live field is compressed / decompressed using "lower", thus the field is not sent. The receiver decompresses it by reading its value from the outer IP header (TTL in case of IPv4 or HL in case of IPv6). IP4_TTL_VALUE designates an EHC Rule compressing / decompressing the Time To Live field of the inner IP header. IP4_TTL_VALUE is only activated when the Hop Limit ("ip4_ttl") has been agreed. Time To Live is compressed / decompressed using the "elided" method. IP4_PROTO designates the EHC Rule compressing / decompressing the Protocol field of the inner IPv4 header. IP4_PROTO is only activated if the Protocol is specified, that is when the Traffic Selectors defines a "Single" Protocol ID ("l4_proto"). When the Protocol ID identified by the SA has a "Single" value, the Protocol is compressed and decompressed using the "elided" method. IP4_CHECK designates the EHC rule compressing / decompressing the Header Checksum field of the inner IPv4 header. The IPv4 header checksum is not sent by the sender and the receiver computes from the decompressed inner IPv4 header. IP4_CHECK MUST compute the checksum and not fill the checksum field with zeros. As a result, IP4_CHECK is the last decompressing EHC Rule to be performed on the decompressed IPv4 header. IP4_SRC compresses the source IP address of the inner IPv4 header. IP4_SRC_IP is only be activated when the Traffic Selectors agreed by the SA defines a "Single" source IP address ("ip4_src"). The Source IP address is compressed / decompressed using the "elided" method. IP4_DST works in a similar way as IP4_SRC_IP but for the destination IP address ("ip4_dst")

All IPv6 EHC Rules MUST be performed during the "clear text esp" phase. The EHC Rules are only defined for compressing the inner IPv6 header and thus can only be used when the SA is using the Tunnel mode. EHC Rule Field Action Parameters IP6_OUTER Version elided ip_version Traffic Class lower Flow Label lower IP6_VALUE Version elided ip_version Traffic Class elided ip6_tc Flow Label elided ip6_fl IP6_LENGTH Payload Length lower IP6_NH Next Header elided l4_proto IP6_HL_OUTER Hop Limit lower IP6_HL_VALUE Hop Limit elided ip6_hl IP6_SRC Source Address elided ip6_src IP6_DST Dest. Address elided ip6_dst IP6_OUTER designates an EHC Rule for compressing / decompressing the first 32 bits of the inner IPv6 header formed by the Version, Traffic Class and Flow Label. IP6_OUTER only proceeds to compression when both the outer and inner IP header are IPv6 header. When the outer IP header is an IPv4, the compression is bypassed. Bypassing enables to proceed to compression of IPv4 and IPv6 traffic in a VPN use case with a single SA. The Version "ip_version" is defined by the SA and is thus compressed using "elided". The other parameters Traffic Class and Flow Label are compressed using "lower". More specifically, the fields are not sent. The receiver decompresses them by reading their value from the outer IPv6 header. IP6_VALUE designates an EHC Rule for compressing / decompressing the first 32 bits of the inner IPv6 header formed by the Version, Traffic Class and Flow Label. IP6_VALUE is only activated if the Version of the inner IP header agreed by the SA is set to "Version 6" ("ip_version" set to "Version 6") and the specific values of the Traffic Class ("ip6_tc") and the Flow Label ("ip6_fl") are specified. With IP6_VALUE all fields are compressed and decompressed using "elided". Version is provided by the SA ("ip_version") while other fields are explicitly provided (ip6_tc, ip6_fl. IP6_LENGTH designates the EHC Rule compressing / decompressing the Payload Length Field of the inner IPv6 header. The Payload Length is compressed by the sender and is not sent. The receiver decompress it by recomputing the Payload Length from the outer IP header. The IP header can be IPv4 or IPv6 and IP6_LENGTH MUST support both versions if both versions are supported by the device. Note that the length of the inner IP payload may also be subject to updates if decompression of the upper layers occurs. IP6_NH designates the EHC Rule compressing / decompressing the Next Header field of the inner IPv6 header. IP6_NH is only activated if the Next Header is specified, that is when the Traffic Selectors defines a "Single" Protocol ID ("l4_proto"). When the Protocol ID identified by the SA has a "Single" value, the Next Header is compressed and decompressed using the "elided" method. IP6_HL_OUTER designates an EHC Rule compressing / decompressing the Hop Limit field of the inner IP header. If the outer IP header is an IPv4 header, the Time To Live is used for decompression. The Hop Limit field is compressed / decompressed using the "lower". More specifically, the fields are not sent. The receiver decompresses them by reading their value from the outer IPv6 header. IP6_HL_VALUE designates an EHC Rule compressing / decompressing the Hop Limit field of the inner IP header. IP6_HL_VALUE is only activated when the Hop Limit ("ip6_hl") has been agreed. The Hop Limit is compressed / decompressed using the "elided" method. IP6_SRC compresses the source IP address of the inner IP header. IP6_SRC_IP is only be activated when the Traffic Selectors agreed by the SA defines a "Single" source IP address ("ip6_src"). The Source IP address is compressed / decompressed using the "elided" method. IP6_DST works in a similar way as IP6_SRC_IP but for the destination IP address ("ip6_dst")

All UDP EHC Rules MUST be performed during the "pre-esp" phase. The EHC Rules are only defined when the Traffic Selectors agreed during the SA negotiation results in "Single" Protocol ID ("l4_proto") which is set to UDP (17). EHC Rule Field Action Parameters UDP_SRC Source Port elided l4_source UDP_DST Dest. Port elided l4_dest UDP_LENGTH Length lower UDP_CHECK UDP Checksum checksum UDP_SRC designates the EHC Rule that compresses / decompresses the UDP Source Port. UDP_SRC is only activated when the Source Port agreed by the SA negotiation ("l4_src") is "Single". The Source Port is then compressed / decompressed using the "elided" method. UDP_DST works in a similar way as UDP_SRC but for the Destination Port ("l4_dst"). UDP_LENGTH designates the EHC Rule compressing / decompressing the Length Field of the UDP header. The length is compressed by the sender and is not sent. The receiver decompresses it by recomputing the Length from the IP address header. The IP address can be IPv4 or IPv6 and UDP_LENGTH MUST support both versions if both versions are supported by the device. UDP_CHECK designates the EHC Rule compressing / decompressing the UDP Checksum. The UDP Checksum is not sent by the sender and the receiver computes from the decompressed UDP payload. UDP_CHECK MUST compute the checksum and not fill the checksum field with zeros. As a result, UDP_CHECK is the last decompressing EHC Rule to be performed on the decompressed UDP Payload.

All UDP-lite EHC Rules MUST be performed during the "pre-esp" phase. The EHC Rules are only defined when the Traffic Selectors agreed during the SA negotiation results in a "Single" Protocol ID ("l4_proto") which is set to UDPLite (136). EHC Rule Field Action Parameters UDP-LITE_SRC Source Port elided l4_source UDP-LITE_DST Dest. Port elided l4_dest UDP-LITE_COVERAGE Checksum Coverage elided udplite_coverage UDP-LITE_CHECK UDP-Lite Checksum checksum UDP-LITE_SRC works similarly to UDP_SRC UDP-LITE_DST works similarly to UDP_DST UDP-LITE_COVERAGE designates the EHC Rule compressing / decompressing the UDP-Lite Coverage field. UDP-LITE_COVERAGE is only activated when the Coverage ("udplite_coverage") has been agreed with a valid value. The Coverage is compressed / decompressed using the "elided" method. UDP-LITE_CHECK designates the EHC Rule compressing / decompressing the UDP-Lite checksum. UDP-LITE_CHECK is only activated if the Coverage is defined either elided or sent. UDP-LITE_CHECK computes the checksum using "checksum" according to the uncompressed UDP packet and the value of the Coverage.

All TCP EHC Rules MUST be performed during the "pre-esp" phase. The EHC Rules are only defined when the Traffic Selectors agreed during the SA negotiation results in a"Single" Protocol ID ("l4_proto") which is set to TCP (6). EHC Rule Field Action Parameters TCP_SRC Source Port elided l4_source TCP_DST Dest. Port elided l4_dest TCP_SN Sequence Number lsb tcp_sn, tcp_lsb TCP_ACK Acknowledgment Number lsb tcp_ack, tcp_lsb TCP_OPTIONS Data Offset elided tcp_options Reserved Bits elided TCP_CHECK TCP Checksum checksum TCP_URGENT TCP Urgent Field elided tcp_urgent TCP_SRC works similarly to UDP_SRC. TCP_DST works similarly to UDP_DST. TCP_SN designates the EHC Rule compressing / decompressing the TCP Sequence Number. TCP_SN is only activated if the SN ("tcp_sn") and the LSB significant bytes ("tcp_lsb") are defined. The TCP SN is compressed using "lsb". The sending peer only places the LSB bytes of the TCP SN ("tcp_sn") and the receiving peer retrieve the TCP SN from the LSB bytes carried in the packets as well as from the TCP SN value stored in EHC Context ("tcp_sn"). The two peers MUST agree on the number of LSB bytes to be sent: "tcp_lsb". Upon agreeing on "tcp_lsb", the receiving peer MUST NOT agree on a value not carrying sufficient information to retrieve the full TCP SN. Note also that unlike the ESP SN, the TCP SN is not part of the SA. As a result, when the SN is compressed, the value of the TCP SN MUST be stored in the EHC Context. The reserved attribute for that is "tcp_sn" TCP_ACK designates the EHC Rule compressing / decompressing the TCP Acknowledgment Number and works similarly to TCP SN. Note that "tcp_lsb" is agreed for both TCP SN and TCP Acknowledgment. Similarly the value of the complete TCP Acknowledgment Number MUST be stored in the "tcp_ack" attribute of the EHC Context. TCP_OPTIONS designates the EHC Rule compressing / decompressing TCP options related fields such as Data Offset and Reserved Bits. TCP_OPTION can only be activated when the TCP Option ("tcp_options") is defined. When "tcp_options" is set to "False" and indicates there are no TCP Options, the Data Offsets and Reserved Bits are compressed / decompressed using the "elided" method with Data Offset and Reserved Bits set to zero. TCP_CHECK designates the EHC Rule compressing / decompressing the TCP Checksum. TCP_CHECK works similarly as UDP_CHECK. TCP_URGENT designates the EHC Rule compressing / decompressing the urgent related information. When "tcp_urgent" is set to "False" and indicates there are no TCP Urgent related information, the Urgent Pointer is then "elided" and filled with zeros.

From the attributes of the EHC Context, Diet-ESP defined as an EHC Strategy, which EHC Rules to apply. The EHC Strategy is defined for outbound packets which compresses the packet as well as for inbound packet where the decompression occurs. Diet-ESP results from a compromise between compression efficiency, ease to configure Diet-ESP and the various use cases considered. In order to achieve a great simplicity, Diet-ESP favors compression methods that required fewer configuration: For IPv6, ip6_tcfl_comp and ip6_hl_com to "Outer" so that ip6_tc, ip6_fl and ip6_hl can be derived from the packet. Similarly, ip4_ttl_comp has is set to "Outer" so ip4_tll can be derived from the packet. Diet-ESP limits compression method to those foreseen as the most commonly used. As such, esp_sn_gen has been set to "Incremental" as this is the most common method used to generate SN. The other method would be "Time". Diet-ESP limits compression to the most foreseen scenarios. IPv4 compression has been limited in favor of IPv6 as constraint devices have largely adopted IPv6, and the gain versus the complexity to deploy IPv4 inner IP addresses has not been proved. As a result some compressions for IPv4 are not considered by DIet-ESP. This involved compression of the IPv4 options by setting ip4_options to "No_Options". Similarly IPv4 ID compression has not been enabled by setting ip4_id and ip4_id_lsb to "Unspecified". Diet-ESP negotiated values shared by different rules such as tcp_lsb which is shared for TCP ACK as well as for the TCP SN. Diet-ESP defines a logic to set the necessary parameters from those agreed by the standard ESP agreement, which limits the setting of parameters. The following tables shows, which EHC Rules are activated by default for the supported protocols ESP, IPv4, IPv6, UDP, UDP-Lite and TCP when using the Diet-ESP strategy and which ones are activated due to certain circumstances or explicit negotiation ESP: EHC Rule Activated if Parameter Value ESP_SPI Diet-ESP esp_spi_lsb Negotiated esp_spi In SA ESP_SN Diet-ESP esp_sn_lsb Negotiated esp_sn_gen Negotiated esp_sn In SA ESP_NH Diet-ESP ipsec_mode In SA l4_proto In SA ESP_PAD Diet-ESP esp_align Negotiated esp_encr In SA IPv4: EHC Rule Activated if Parameter Value IP4_OPT_DIS ip_version==4 ip_version In SA IP4_LENGTH ip_version==4 None IP4_FRAG_DIS ip_version==4 None IP4_TTL_OUTER ip_version==4 None IP4_TTL_OUTER ip_version==4 l4_proto In SA IP4_CHECK ip_version==4 None IP4_SRC ip_version==4 ip4_src In SA IP4_DST ip_version==4 ip4_dst In SA IPv6: EHC Rule Activated if Parameter Value IP6_OUTER ip_version==6 ip_version In SA IP6_LENGTH ip_version==6 None IP6_NH ip_version==6 l4_proto In SA IP6_HL_OUTER ip_version==6 None IP6_SRC ip_version==6 ip6_src In SA IP6_DST ip_version==6 ip6_dst In SA UDP: EHC Rule Activated if Parameter Value UDP_SRC l4_proto==17 l4_source In SA UDP_DST l4_proto==17 l4_dest In SA UDP_LENGTH l4_proto==17 None UDP_CHECK l4_proto==17 None UDP-Lite: EHC Rule Activated if Parameter Value UDP_LITE_SRC l4_proto==136 l4_source In SA UDP_LITE_DST l4_proto==136 l4_dest In SA UDP_LITE_COVERAGE l4_proto==136 udplite_coverage Negotiated UDP_LITE_CHECK l4_proto==136 None TCP: EHC Rule Activated if Parameter Value TCP_SRC l4_proto==6 l4_source In SA TCP_DST l4_proto==6 l4_dest In SA TCP_SN l4_proto==6 tcp_sn In SA tcp_lsb Negotiated TCP_ACK l4_proto==6 tcp_ack In SA tcp_lsb Negotiated TCP_OPTIONS l4_proto==6 tcp_options Negotiated TCP_CHECK l4_proto==6 None TCP_URGENT l4_proto==6 tcp_urgent Negotiated Thus, the parameters that the two peers needs to agree on are: esp_sn_lsb esp_spi_lsb esp_align udplite_coverage tcp_lsb tcp_options tcp_urgent Implementation may differ from the description below. However, the outcome MUST remain the same.

Diet-ESP compression is defined as follows: In phase "pre-esp": Match the inbound packet with the SA and determine if the Diet-ESP EHC Strategy has been activated. If the Diet-ESP EHC Strategy has been activated proceed to next step, otherwise skip all steps associated to Diet-ESP and proceed to the standard ESP as defined in In phase "pre-esp": If "l4_proto" designates a "Single" Protocol ID (UDP, TCP or UDP-Lite), proceed to the compression of the specific layer. Otherwise, the transport layer is not compressed. In phase "clear text esp": If "ipsec_mode" is set to "Tunnel" mode, determine "ip_version" the IP version of the inner IP addresses and proceed to the appropriated inner IP address compression. In phase "clear text esp" and "post-esp": Proceed to the ESP compression. UDP compression is defined as below: If "l4_src" designates a "Single" Source Port, apply UDP_SRC to compress the Source Port. If "l4_dst" designates a "Single" Destination Port, apply UDP_DST to compress the Destination Port. Apply UDP_CHECK to compress the Checksum. Apply UDP_LENGTH to compress the Length. UDP-lite compression is defined as below: If "l4_src" designates a "Single" Source Port, apply the UDP-LITE_SRC to compress the Source Port. If "l4_dst" designates a "Single" Destination Port, apply the UDP-LITE_DST, to compress the Destination Port. If "udplite_coverage" is specified, apply the UDP-LITE_COVERAGE, to compress the Coverage. Apply UDP-LITE_CHECK to compress the Checksum. TCP compression is defined as below: If "l4_src" designates a "Single" Source Port than apply the TCP_SRC to compress the Source Port. If "l4_dst" designates a "Single" Destination Port than apply the TCP_DST to compress the Destination Port. If "tcp_lsb" is lower than 4, then "tcp_sn" "tcp_ack" attributes of the Diet-ESP Context are updated with the value provided from the packet before applying the TCP_SN and the TCP_ACK EHC Rules. If "tcp_options" is set to "False" apply the TCP_OPTIONS EHC Rule. If "tcp_urgent" is set to "False" apply the TCP_URGENT EHC Rule. Apply TCP_CHECK to compress the Checksum. Inner IPv6 Header compression is defined as below: If "ip6_src" designates a "Single" Source IP address, apply the IP6_SRC to compress the IPv6 Source Address. If "ip6_dst" designates a "Single" Destination IP address, apply the IP6_DST to decompress the IPv6 Destination Address. Apply IPv6_HL_OUTER to compress the Hop Limit. If "l4_proto" designates a "Single" Protocol ID (UDP, TCP or UDP-Lite), apply IP6_NH to compress the Next Header. Apply, IP6_LENGTH to compress the Length. Apply IP6_OUTER to compress Version, Traffic Class and Flow Label. Inner IPv4 Header compression is defined as below: Apply, IP4_LENGTH to compress the Length. Apply IP4__TTL_OUTER to compress Time To Live. Apply, IP4_CHECK to compress the IPv4 header checksum. If "ip4_src" designates a "Single" Source IP address, apply the IP4_SRC to compress the IPv4 Source Address. If "ip4_dst" designates a "Single" Destination IP address, apply the IP4_DST to decompress the IPv4 Destination Address. ESP compression is defined as below: In phase "clear text esp": If "ipsec_mode" is set to "Tunnel" or "l4_proto" is set to a "Single value - eventually different from TCP, UDP or UDP-Lite, apply ESP_NH, to compress the Next Header. In phase "clear text esp": If "esp_encr" specify an encryption algorithm that does not provide padding, then apply ESP_PAD to compress the Pad Length and Padding. Proceed to the ESP encryption as defined in . In phase "post-esp: If "esp_sn_lsb" is different from 4, then apply ESP_SN. To compress the ESP SN. In phase "post-esp": If "esp_spi_lsb" is different from 4, then apply ESP_SPI to compress the SPI.

Diet-ESP decompression is defined as follows: Match the inbound packet with the SA and determine if the Diet-ESP EHC Strategy has been activated. When Diet-ESP is activated this means that the "esp_spi_lsb" are sufficient to index the SA and proceed to next step, otherwise skip all steps associated to Diet-ESP and proceed to the standard ESP as defined in In phase "clear text esp" and "post-esp": Proceed to the ESP decompression. In phase "clear text esp": If "ipsec_mode" is set to "Tunnel" mode, determine "ip_version" the IP version of the inner IP addresses and proceed to the appropriated inner IP address decompression, except for the computation of the checksums and length. In phase "pre-esp": If "l4_proto" designates a "Single" Protocol ID (UDP, TCP or UDP-Lite), proceed to the decompression of the specific layer, except for the computation of the checksums and length replaced by zero fields. In phase "pre-esp": Proceed to the decompression of the checksums and length. ESP decompression is defined as follows: In phase "post-esp": If "esp_spi_lsb" is different from 4, then apply ESP_SPI to decompress the SPI. In phase "post-esp: If "esp_sn_lsb" is different from 4, then apply ESP_SN. To decompress the ESP SN. Proceed to the ESP signature validation and decryption as defined in . In phase "clear text esp": If "ipsec_mode" is set to "Tunnel" or "l4_proto" is set to a "Single value - eventually different from TCP, UDP or UDP-Lite, apply ESP_NH, to decompress the Next Header. In phase "clear text esp": If "esp_encr" specify an encryption algorithm that does not provide padding, then apply ESP_PAD to compress the Pad Length and Padding. Extract the ESP Data Payload and apply decompression EHC Rule to the ESP Data Payload. UDP decompression is defined as follows: If "l4_src" designates a "Single" Source Port, apply UDP_SRC to decompress the Source Port. If "l4_dst" designates a "Single" Destination Port, apply UDP_DST to decompress the Destination Port. Apply UDP_LENGTH to compress the Length. The length value is computed from the length provided by the lower layer, with the additional added bytes during the UDP decompression including the length size. Apply UDP_CHECK to decompress the Checksum. Update the Length of the lower layers: If "ipsec_mode" is set to "Transport" mode, update the Length of the outer IP header (IPv4 or IPv6). The Length is incremented by the number of bytes generated by the decompression of the transport layer. If "ipsec_mode" is set to "Tunnel" mode, update the Length of the inner IP address (IPv4 or IPv6) as well as the outer IP header (IPv4 or IPv6). The Length is incremented by the number of bytes generated by the decompression of the transport layer. UDP-Lite decompression is defined as follows: If "l4_src" designates a "Single" Source Port, apply the UDP-LITE_SRC to decompress the Source Port. If "l4_dst" designates a "Single" Destination Port, apply the UDP-LITE_DST, to decompress the Destination Port. If "udplite_coverage" is specified, apply the UDP-LITE_COVERAGE, to decompress the Coverage. Apply UDP-LITE_CHECK to compress the Checksum. Update the Length of the lower layers as defined in UDP. TCP decompression is defined as follows: If "l4_src" designates a "Single" Source Port than apply the TCP_SRC to decompress the Source Port. If "l4_dst" designates a "Single" Destination Port than apply the TCP_DST to decompress the Destination Port. If "tcp_lsb" is lower than 4, apply TCP_SN and the TCP_ACK to decompress the TCP Sequence Number and the TCP Acknowledgment Number. If "tcp_options" is set to "False" apply TCP_OPTIONS to decompress Data Offset and Reserved Bits. If "tcp_urgent" is set to "False" apply the TCP_URGENT to decompress the Urgent Pointer. Apply TCP_CHECK to decompress the Checksum. Inner IPv6 decompression is defined as follows: Apply IP6_OUTER to decompress Version, Traffic Class and Flow Label. Set the Length to zero. If "l4_proto" designates a "Single" Protocol ID (UDP, TCP or UDP-Lite), apply IP6_NH to decompress the Next Header. Hop Limit is decompressed with IP6_HL_OUTER (with "ip6_hl_comp" set to "Outer"). If the "ip6_src" designates a "Single" Source IP address, apply the IP6_SRC to de compress the IPv6 Source Address. If the "ip6_dst" designates a "Single" Destination IP address than apply the IP6_DST to decompress the IPv6 Destination Address. Apply, IP6_LENGTH to provide the replace the zero length value by its appropriated appropriated value. The Length value considers the length provided by the lower layers to which are added the additional bytes due to the decompression, minus the length of the inner IP6 Header. Inner IPv4 decompression is defined as follows: Apply, IP4_LENGTH to provide the replace the zero length value by its appropriated appropriated value. The Length value considers the length provided by the lower layers to which are added the additional bytes due to the decompression, minus the length of the inner IPv4 Header. The value computed from the lower layer will have to be overwritten in case further decompression occurs. Apply IP4_TTL_OUTER to decompress Time To Live. If "l4_proto" designates a "Single" Protocol ID (UDP, TCP or UDP-Lite), apply IP4_PROT to decompress the Protocol Field. If "ip4_src" designates a "Single" Source IP address, apply the IP4_SRC to de compress the IPv4 Source Address. If "ip4_dst" designates a "Single" Destination IP address than apply the IP4_DST to decompress the IPv4 Destination Address. Apply IP4_CHECK to decompress the checksum of the IPv4 header

There are no IANA consideration for this document.

This section lists security considerations related to the Diet-ESP protocol. The Security Parameter Index (SPI) is used by the receiver to index the Security Association that contains appropriated cryptographic material. If the SPI is not found, the packet is rejected as no further checks can be performed. In EHC, the value of the SPI is not reduced, but compressed why the SPI value may not be fully provided between the compressor and the de-compressor. On the other hand, its uncompressed value is provided to the ESP-procession and no weakness is introduced to ESP itself. On an implementation perspective, it is strongly recommended that decompression is deterministic. Compression and decompression adds some additional treatment to the ESP packet, which might be used by an attacker. In order to minimize the load associated to decompression, decompression is expected to be deterministic. The incoming compressed SPI with the associated IP addresses should output a single and unique uncompressed SPI value. If an uncompressed SPI values have to be considered, then the receiver could end in n signature checks which may be used by an attacker for a DoS attack. The Sequence Number (SN) is used as an anti-replay attack mechanism. Compression and decompression of the SN is already part of the standard ESP namely the Extended Sequence Number (ESN). The SN in a standard ESP packet is 32 bit long, whether EHC enables to reduce it to 0 bytes and the main limitation to the compression a deterministic decompression. SN compression consists in indicating the least significant bits of the uncompressed SN on the wire. The size of the compressed SN must consider the maximum reordering index such that the probability that a later sent packet arrives before an earlier one. In addition the size of SN should also consider maximum consecutive packets lost during transmission. In the case of ESP, this number is set to 2^32 which is, in most real world case, largely over-provisioned. When the compression of the SN is not appropriately provisioned, the most significant bit value may be de-synchronized between the sending and receiving parties. Although IKEv2 provides some re-synchronization mechanisms, in case of IoT the de-synchronization will most likely result in a renegotiation and thus DoS possibilities. Note that IoT communication may also use some external parameters, i.e. other than the compressed SN, to define whether a packet be considered or not and eventually derive the SN. One such scenario may be the use of time windows. Suppose a device is expected to send some information every hour or every week. In this case, for example, the SN may be compressed to zero bytes. Instead the SN may be derived by incrementing the SN every hour after the last received valid packet. Considering the time the packet is received make it possible to consider the time derivation of the sensor clock. If TIME is used as the method to generate the SN, the receiver MUST ensure that the esp_sn_lsb is big enough to resist time differences between the nodes. Note also that the anti-replay mechanism needs to define the size of the anti-replay window. provides guidance to set the window size and are similar to those used to define the size of the compressed SN.

Until Diet-ESP is deployed outside the scope of IoT and small devices, the use of a compressed SPI may provide an indication that one of the endpoint is a sensor. Such information may be used, for example, to evaluate the number of appliances deployed, or - in addition with other information, such as the time interval, the geographic location - be used to derive the type of data transmitted. If incremented for each ESP packet, the SN may leak some information like the amount of transmitted data or the age of the sensor. The age of the sensor may be correlated with the software used and the potential bugs. On the other hand, re-keying will re-initialize the SN, but the cost of a re-keying may not be negligible and thus, frequent re-keying can be considered. In addition to the re-key operation, the SN may be generated in order to reduce the accuracy of the information leaked. In fact, the SN does not have to be incremented by one for each packet it just has to be an increasing function. Using a function such as a TIME may prevent characterizing the age or the use of the sensor. Note that the use of such function may also impact the compression efficiency and result in larger compressed SN. Another possibility may also consists in compressing the SN to the low order bytes to reduce the information related to the age or the number of packets being exchanged.

We would like to thank Orange and Universitee Pierre et Marie Curie for initiating the work on Diet-ESP. We Would like to thank Sylvain Killian for implementing an open source Diet-ESP on Contiki and testing it on the FIT IoT-LAB funded by the French Ministry of Higher Education and Research. We thank the IoT-Lab Team and the INRIA for maintaining the FIT IoT-LAB platform and for providing feed backs in an efficient way. We would like to thank Bob Moskowitz for not copyrighting Diet HIP. The "Diet" terminology is from him. We would like to thank those we received many useful feed backs among others: Dominique Bartel, Anna Minaburo, Suresh Krishnan, Samita Chakrabarti, Michael Richarson, Tero Kivinen.

This section considers a IoT IPv6 probe hosting a UDP application. The probe is dedicated to a single application and establishes a single UDP session. As a result, inner IP addresses and UDP Ports have a "Single" value and can be easily compressed. The probes sets an IPsec VPN using IPv6 addresses in order to connect its secure domain - typically a Home Gateway. The use of IPv6 for inner and outer IP addresses, enables to infer inner IP fields from the outer IP address. The probes encrypts with AES-CCM_8 . AES-CCM does not have padding, so the padding is performed by ESP. The probes uses an 8 bit alignment which enables to fully compress the ESP Trailer. In addition, as the probe SA is indexed using the outer IP addresses (or eventually the radio identifiers) which enables to fully compress the SPI. As the probe provides information every hour, the Sequence Number using time can be derived from the received time, which enables to fully compress the SN. represents the original UDP packet and represents the corresponding packet compressed with Diet-ESP. The compression with Diet-ESP results in a reduction of 61 bytes overhead. With IPv4 inner IP addressed Diet-ESP results in an 45 byte overhead reduction. Further compression may be done for example by using an implicit IV and by compressing the outer IP addresses (not represented) on the figure. In addition, application data may also be compressed with mechanisms outside of the scope of Diet-ESP.

The following table illustrates the activated rules and the attributes of the Diet-ESP Context that needs an explicit agreement to achieve the compression. All other attributes used by the rules are part of the SA agreement. Parameters of not activated rules are left "Unspecified". EHC Rule Context Attribute Value ESP_SPI esp_spi_lsb 0 ESP_SN esp_sn_lsb 0 esp_sn_gen ESP_NH ESP_PAD esp_align 8 IP6_OUTER ip6_tcfl_comp ip6_hl_comp IP6_LENGTH IP6_NH IP6_HL_OUTER IP6_SRC IP6_DST UDP_SRC UDP_DST UDP_LENGTH UDP_CHECK

This section considers the same probe as described in but instead of using UDP as a transport layer, the probe uses TCP. In this case TCP is used with no options, no urgent pointers and the SN and ACK Number are compressed to 2 bytes as the throughput is expected to be low. represents the original TCP packet and represents the corresponding packet compressed with Diet-ESP. The compression with Diet-ESP results in a reduction of 66 bytes overhead. With IPv4 inner address Diet-ESP results in a 50 byte overhead reduction.

The following table illustrates the activated rules and the attributes of the Diet-ESP Context that needs an explicit agreement to achieve the compression. All other attributes used by the rules are part of the SA agreement. Parameters of not activated rules are left "Unspecified". Note for simplicity, tcp_sn and tcp_ack are negotiated to start with 0, but it could be any other value as well. EHC Rule Context Attribute Value ESP_SPI esp_spi_lsb 0 ESP_SN esp_sn_lsb 0 esp_sn_gen ESP_NH ESP_PAD esp_align 8 IP6_OUTER ip6_tcfl_comp ip6_hl_comp IP6_LENGTH IP6_NH IP6_HL_OUTER IP6_SRC IP6_DST TCP_SRC TCP_DST TCP_SN tcp_lsb 2 tcp_sn 0 TCP_ACK tcp_lsb 2 tcp_ack 0 TCP_OPTIONS tcp_options "False" TCP_CHECK TCP_URGENT tcp_urgent "False"

This section illustrates the case of an company VPN. The VPN is typically set by a remote host that forwards all its traffic to the security gateway. As transport protocols are "Unspecified", compression is limited to ESP and the inner IP header. For the inner IP header, the Destination IP address is "Unspecified" so the compression of the inner IP address excludes the Destination IP address. Similarly, the inner IP Next Header cannot be compressed as the transport layer is not specified. For ESP, the security gateway may only have a sufficiently low number of remote users with relatively low throughput in which case SPI and SN can be compressed to 2 bytes. As throughput remains relatively low, the alignment may also set to 8 bits.

represents the original TCP packet with IPv6 inner IP addresses and represents the corresponding packet compressed with Diet-ESP. The compression with Diet-ESP results in a reduction of 32 bytes.

The following table illustrates the activated rules and the attributes of the Diet-ESP Context that needs an explicit agreement to achieve the compression. All other attributes used by the rules are part of the SA agreement. Parameters of not activated rules are left "Unspecified". EHC Rule Context Attribute Value ESP_SPI esp_spi_lsb 2 ESP_SN esp_sn_lsb 2 esp_sn_gen ESP_NH ESP_PAD esp_align 8 IP6_OUTER ip6_tcfl_comp IP6_LENGTH IP6_HL_OUTER ip6_hl_comp IP6_SRC

If the compressed inner IP header is an IPv6, but the outer IP header is an IPv4 header, the activated rules differ, as IP6_OUTER cannot be used. Instead, ip6_tcfl_comp and ip6_hl_comp are set to "Value". The resulting ESP packet is the same as in . EHC Rule Context Attribute Value ESP_SPI esp_spi_lsb 2 ESP_SN esp_sn_lsb 2 esp_sn_gen ESP_NH ESP_PAD esp_align 8 IP6_OUTER ip6_tcfl_comp IP6_LENGTH IP6_HL_OUTER ip6_hl_comp IP6_SRC

represents the original TCP packet with IPv4 inner IP addresses and represents the corresponding packet compressed with Diet-ESP. The compression with Diet-ESP results in a reduction of 24 bytes.

The following table illustrates the activated rules and the attributes of the Diet-ESP Context that needs an explicit agreement to achieve the compression. All other attributes used by the rules are part of the SA agreement. Parameters of not activated rules are left "Unspecified". EHC Rule Context Attribute Value ESP_SPI esp_spi_lsb 2 ESP_SN esp_sn_lsb 2 esp_sn_gen "Incremental" ESP_NH ESP_PAD esp_align 8 IP4_OPT_DIS IP4_LENGTH IP4_FRAG_DIS IP4_TTL_OUTER IP4_CHECK IP4_SRC

If the compressed inner IP header is an IPv4, but the outer IP header is an IPv6 header, the activated rules differ, as IP4_TTL_OUTER cannot be used. Instead, IP4_TTL_VALUE is used. The resulting ESP packet is the same as in . EHC Rule Context Attribute Value ESP_SPI esp_spi_lsb 2 ESP_SN esp_sn_lsb 2 esp_sn_gen "Incremental" ESP_NH ESP_PAD esp_align 8 IP4_OPT_DIS IP4_LENGTH IP4_FRAG_DIS IP4_CHECK IP4_SRC