Internet Engineering Task Force T. Hardjono Internet-Draft MIT Intended status: Informational M. Hargreaves Expires: 8 July 2024 Quant Network N. Smith Intel V. Ramakrishna IBM 5 January 2024 Secure Asset Transfer (SAT) Interoperability Architecture draft-ietf-satp-architecture-02 Abstract This document proposes an interoperability architecture for the secure transfer of assets between two networks or systems based on the gateway model. Status of This Memo This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79. Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet- Drafts is at https://datatracker.ietf.org/drafts/current/. 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." This Internet-Draft will expire on 8 July 2024. Copyright Notice Copyright (c) 2024 IETF Trust and the persons identified as the document authors. All rights reserved. Hardjono, et al. Expires 8 July 2024 [Page 1] Internet-Draft SAT Architecture January 2024 This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (https://trustee.ietf.org/ license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Revised BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Revised BSD License. Table of Contents 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2 2. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 3 3. Assumptions and Principles . . . . . . . . . . . . . . . . . 4 3.1. Design Principles . . . . . . . . . . . . . . . . . . . . 4 3.2. Operational Assumptions . . . . . . . . . . . . . . . . . 5 3.3. Assumptions Regarding Gateway Operators . . . . . . . . . 5 4. Gateway Interoperability Modes . . . . . . . . . . . . . . . 6 5. Architecture . . . . . . . . . . . . . . . . . . . . . . . . 7 5.1. Goal of Architecture . . . . . . . . . . . . . . . . . . 7 5.2. Overview of Asset Transfer . . . . . . . . . . . . . . . 8 5.3. Desirable Properties of Asset Transfer . . . . . . . . . 8 5.4. Event log-data, crash recovery and backup gateways . . . 10 5.5. Overview of the Stages in Asset Transfer . . . . . . . . 10 6. Transfer Initiation Claims negotiations (Stage-1) . . . . . . 12 7. Asset Lock Assertion and Receipt (Stage 2) . . . . . . . . . 15 8. Commitment Preparation and Finalization (Stage 3) . . . . . . 16 9. The Commitment sub-protocol . . . . . . . . . . . . . . . . . 18 10. Security Considerations . . . . . . . . . . . . . . . . . . . 19 11. Policy Considerations . . . . . . . . . . . . . . . . . . . . 19 12. Threat Model Considerations . . . . . . . . . . . . . . . . . 20 13. Compatibility Considerations . . . . . . . . . . . . . . . . 20 14. References . . . . . . . . . . . . . . . . . . . . . . . . . 20 14.1. Normative References . . . . . . . . . . . . . . . . . . 20 14.2. Informative References . . . . . . . . . . . . . . . . . 21 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 23 1. Introduction This document proposes an interoperability architecture based on gateways, which are points of interconnection between networks or systems. There are several services that may be offered by a gateway, one of which being the direct transfer of a digital asset from one network to another via pairs of gateways without a mediating third party. Hardjono, et al. Expires 8 July 2024 [Page 2] Internet-Draft SAT Architecture January 2024 A given network or system may have one or more gateways to perform a unidirectional direct transfer of digital assets to another network possessing one or more compatible gateway. Both gateways must implement a secure asset transfer protocol that must satisfy certain security, privacy and atomicity requirements. The purpose of this architecture document is to provide technical framework within which to define the required properties of a gateway that supports the secure asset transfer protocol. 2. Terminology There following are some terminology used in the current document. We borrow terminology from NIST and ISO as much as possible, introducing new terms only when needed: * Asset network (system): The network or system where a digital asset is utilized. * Asset Transfer Protocol: The protocol used to transfer (move) a digital asset from one network to another using gateways. * Origin network: The current network where the digital asset is located. * Destination network: The network to which a digital asset is to be transferred. * Resource Domain: The collection of resources and entities participating within an asset network. The domain denotes a boundary for permissible or authorized actions on resources. * Interior Resources: The various interior protocols, data structures and cryptographic constructs that are a core part of an asset network or system. * Exterior Resources: The various protocols, data structures and cryptographic constructs that are outside of (external to) the network or system. * Gateway: The collection of services which connects to a minimum of one network or system, and which implements the secure asset transfer protocol. * Entity public-key pair: This the private-public key pairs of an entity, where the public-key is available and verifiable outside the network. Among others, it may be utilized for interactions Hardjono, et al. Expires 8 July 2024 [Page 3] Internet-Draft SAT Architecture January 2024 other entities from outside the network. The term is used to distinguish this public-key from other key-pairs belonging to the same entity, but which is only available within the (private) network. * Originator: Person or organization in an origin network seeking the transfer of a digital asset to a beneficiary located in a remote network. * Beneficiary: Person or organization in an destination network seeking to receive the transfer of a digital asset to from an originator located in a remote network. * Gateway device identity: The identity of the device implementing the gateway functions. The term is used in the sense of IDevID (IEEE 802.1AR) or EK/AIK (in TPM1.2 and TPM2.0) [IDevID]. * Gateway owner: The entity that owns and operates a gateway within a network. * Application Context-ID: The relevant identifier used by originator's application and the beneficiary's application to identify the context of the asset transfer at the gateway level. The context identifier may also be used to bind the application to selected gateway for the given transfer instance, identified by a Session-ID. * Gateway Session-ID: This the identifier used between the sender gateway and the recipient gateway to identify the specific transfer instance. The Session-ID must be included in all messages between the gateways. 3. Assumptions and Principles The following assumptions and principles underlie the design of the current gateway architecture, and correspond to the design principles of the Internet architecture. 3.1. Design Principles * Opaque network resources: The interior resources of each network is assumed to be opaque to (hidden from) external entities. Any resources to be made accessible to an external entity must be made explicitly accessible by a gateway with proper authorization. * Externalization of value: The asset transfer protocol is agnostic (oblivious) to the economic or monetary value (if any) of the digital asset being transferred. Hardjono, et al. Expires 8 July 2024 [Page 4] Internet-Draft SAT Architecture January 2024 The opaque resources principle permits the architecture to be applied in cases where one (or both) networks are private (closed membership). It is the analog of the autonomous systems principle in IP networking [Clar88], where interior routes in local subnets are not visible to other external networks. The value-externalization principle permits an asset transfer protocol to be designed for efficiency, security and reliability -- independent of the changes in the perceived economic value of the digital asset. It is the analog of the end-to-end principle in the Internet architecture [SRC84], where contextual information is placed at the endpoints of the transfer. 3.2. Operational Assumptions The following conditions are assumed to have occurred, leading to the invocation of the asset transfer protocol between two gateways: * Application level context establishment: The transfer request from an Originator utilizing an application (App1) in the origin network is assumed to have occurred, and that some context- identifier has subsequently been derived by the respective applications (App1 and App2). Furthermore, this context- identifier is assumed to have been delivered by the each application to its corresponding gateway, permiting each gateway to internally bind the transfer session-identifier to that context-identifier. * Identification of asset to be transferred: The applications at the originator and the beneficiary are assumed to have identified the digital asset to be transferred. * Identification of originator and beneficiary: The originator and beneficiary are assumed to have been identified and that consent has been obtained from both parties regarding the asset transfer. * Identification of origin and destination asset networks: The origin and destination networks is assumed to have been identified. * Selection of gateway: The two corresponding gateways at the origin and destination networks is assumed to have been identified and selected. 3.3. Assumptions Regarding Gateway Operators The following conditions are assumed to have occurred, leading to the invocation of the asset transfer protocol between two gateways: Hardjono, et al. Expires 8 July 2024 [Page 5] Internet-Draft SAT Architecture January 2024 * Identification of gateway-owners: The owners of the two corresponding gateways are assumed to have been identified and their ownership status verified. * Gateway liabilities: Gateways and gateway-operators are assumed to take on legal and financial liability for their transactions, and gateways are assumed to operate under a well-defined legal framework (e.g. contractual relationship). Furthermore, the legal framework is assumed to be supported by compatible legislation in the relevant jurisdictions where the gateways are operating. * Gateway message signatures: All messages between gateways are assumed to be signed and verified (e.g. X.509). * Transitory ownership of asset by gateway: Assets being transferred via SAT will be technically be owned by gateway in transit and gateways are liable for them while they have ownership. * Network data: Gateways are assumed to have mechanisms in place to trust data returned from their local networks. This will depend on the technical architecture and capabilities of each specific network. * Gateways are trusted: The gateways are assumed to be trusted to carry-out all the stages of the protocol described in this architecture. 4. Gateway Interoperability Modes The current interoperability architecture based on gateways recognizes several types of transfer flows: * Asset transfer: This refers to the transfer of a digital asset from the origin network to a destination network, where a successful asset transfer causes the asset to be extinguished in the origin network and be created (generated) at the destination network. * Data transfer: This refers to the transfer of data only under authorization, in such a way that the data can be verified by a third party. The data transfer mode addresses the use-cases where the state update in one network or system depends on the existence of state information recorded in a different network or system. Hardjono, et al. Expires 8 July 2024 [Page 6] Internet-Draft SAT Architecture January 2024 * Asset exchange (swap): This refers to the case where two users are present in two networks, and they perform concurrent and atomic swaps of two assets in the two corresponding networks, without transferring the assets outside the networks. The gateways aid in coordinating the messages pertaining to the swap. The remainder of this architecture document will focus on the asset transfer flows. 5. Architecture 5.1. Goal of Architecture The goal of the interoperability architecture is to permit two (2) gateways belonging to distinct networks to conduct a transfer of digital assets transfer between them, in a secure, atomic and verifiable manner. The asset as understood by the two gateways is expressed in an standard digital format in a way meaningful to the gateway syntactically and semantically. The architecture recognizes that there are different networks currently in operation and evolving, and that in many cases the interior technical constructs in these networks maybe incompatible with one another. The architecture therefore assumes that in addition to implementing the bilateral secure asset transfer protocol, a gateway has the role of making opaque (i.e. hiding) the constructs that are local and specific to its network. Overall this approach ensures a high degree of interoperability across these networks, where each network can operate as a true autonomous system. Additionally, this approach permits each network to evolve its interior technology implementations without affecting other (external) networks. The current architecture focuses on unidirectional asset transfers, although the building blocks in this architecture can be used to support protocols for bidirectional transfers. For simplicity the current architecture employs two (2) gateways per transfer as the basic building block, with one gateway in the origin and destination networks respectively. However, the architecture seeks to be extensible to address future cases involving multiple gateways at both sides. Hardjono, et al. Expires 8 July 2024 [Page 7] Internet-Draft SAT Architecture January 2024 5.2. Overview of Asset Transfer An asset transfer between two networks is performed using a secure asset transfer protocol implemented by the gateways in the respective networks. The two gateways implement the protocol in a direct interaction (unmediated). A successful transfer results in the asset being extinguished (burned) or marked on the origin network, and for the asset to be regenerated (minted) at the destination network. The secure asset transfer protocol provides a coordination between the two gateways through the various message flows in the protocol that is communicated over a secure channel. The protocol implements a commitment mechanism between the two gateways to ensure that the relevant properties atomicity, consistency, isolation, and durability (ACID) are achieved in the transfer. The mechanism to extinguish (burn) or regenerate (mint) an asset from/into a network by its gateway is dependent on the specific network and is outside the scope of the current architecture. As part of the commitment mechanism, the sender gateway in the origin network must deliver a signed assertion to the receiver gateway at the destination network which states that asset in question has been extinguished (burned) from the origin network. Similarly, the receiver gateway at the destination network must in return deliver a signed assertion to the sender gateway at the origin network which states that the asset has been regenerated (minted) in the destination network. These two tasks must be performed in a synchronized fashion between the two gateways, and the commitment mechanism must provide sufficent evidence of the asset transfer that is verifiable by an authorized third party. 5.3. Desirable Properties of Asset Transfer The desirable features of asset transfers between two gateway include, but not limited, to the following: * Atomicity: A transfer must either commit or entirely fail (failure means no change to asset state). Hardjono, et al. Expires 8 July 2024 [Page 8] Internet-Draft SAT Architecture January 2024 * Consistency: A transfer (commit or fail) always leaves the networks in a consistent state (i.e. the asset is located in one network only at any time). * Isolation: While the transfer is occurring, the asset state cannot be modified in the origin network. * Durability: Once a transfer has been committed by both gateways, it must remain so regardless of subsequent gateway crashes. * Verifiable by authorized third parties: The proof that the asset has been extinguished in the origin network, and the proof that the asset has been generated in the destination network must be verifiable by an authorized third party. An implementation of the asset transfer protocol should satisfy these properties, independent of whether the implementation employs stateful messaging or stateless messaging between the two gateways. Effecting an asset transfer safely and securely is not simply a matter of communicating desire or intent between two systems represented by gateways, though such communication is a necessary part of asset transfer. The systems, or at least their gateway proxies, must be interoperable in order to transfer assets among themselves, but such interoperability imposes strictly more demands on systems managing digital assets, especially systems that are built on distributed ledgers, than conventional communication interoperability does. Communication interoperability, which is concerned with syntax and semantics of information geared towards producing a common understanding (or knowledge reconciliation) among systems, is insufficient to fulfill an asset transfer that requires systems to carry out state updates in concert with each other. But communication, or messaging standards, play a necessary and complementary role to asset transfer protocols. An exemplar of this is ISO 20022, which is a comprehensive global standard for financial messaging that specifies message syntax for common actions occurring in financial business processes, including payments, credit card transactions, securities settlements, funds, and trade [ISO20022]. This standard provides the tools to model business processes from basic logical building blocks and schemas to construct messages using common formats like XML, JSON, and ASN.1. As we will see later in this document, such messaging standards are useful to communicate information about the states of processes and digital assets across systems, to make requests, and to convey intent. They therefore play a necessary and complementary role in Hardjono, et al. Expires 8 July 2024 [Page 9] Internet-Draft SAT Architecture January 2024 asset transfer protocols. However they are by themselves insufficient to ensure the ACID and verifiability properties described earlier. Another way to think about the relationship between messaging standards like ISO 20022 and asset transfer protocols is that the former is concerned with the "what" of cross- system interoperability whereas the latter is concerned with the "how". Both kinds of protocols treat systems as black boxes, but asset transfer protocols must place some responsibility, and depend, on systems to drive a protocol instance to successful conclusion. 5.4. Event log-data, crash recovery and backup gateways Implementations of a gateway should maintain event logs and checkpoints for the purpose of gateway crash recovery. The log-data generated by a gateway should be considered as an interior resource accessible to other authorized gateways within the same network. The mechanism used to provide gateway crash-recovery is dependent on the specific network. For interoperability purposes the information contained in the log and the format of the log-data should be standardized. The resumption of an interrupted transfer session (e.g. due to gateway crash, network failure, etc.) should take into consideration the aspects of secure channel establishment and the aspects of the transfer protocol resumption. In some cases, a new secure channel (e.g. TLS session) may need to be established between the two gateways, before a resumption of the transfer can begin. The log-data collected by a gateway acts also as a checkpoint mechanism to assist the recovered (or backup) gateway in continuing the transfer. The point at which to re-start the transfer protocol flow is dependent on the implementation of the gateway recovery strategy. 5.5. Overview of the Stages in Asset Transfer The interaction between two gateways in the secure asset transfer protocol is summarized in Figure 1, where the origin network is NW1 and the destination network is NW2. T he gateways are denoted as G1 and G2 respectively. Hardjono, et al. Expires 8 July 2024 [Page 10] Internet-Draft SAT Architecture January 2024 Originator Beneficiary | | +-------------+ +-------------+ | Client | | Client | | Application | | Application | | (App1) | | (App2) | +-------------+ +-------------+ | | | (Stages) | V V +-------------+ |<-----(1)----->| +-------------+ | Network | +----+ +----+ | Network | | NW1 | |Gate| |Gate| | NW2 | | |--|way |<-----(2)----->|way |--| | | +---------+ | | G1 | | G2 | | +---------+ | | | State | | +----+ +----+ | | State | | | | Data DB1| | +----+ +----+ | | Data DB2| | | +---------+ | |<-----(3)----->| | +---------+ | +-------------+ +-------------+ Figure 1 The stages are summarized as follows. * Stage 0: Pre-transfer Verification and Context Establishment. The two applications utilized by the originator and beneficiary is assumed to interact as part of the asset transfer. In this stage, the applications App1 and App2 may establish some shared transfer context information (e.g. Context-ID) at the application level that will be made available to their respective gateways G1 and G2. The legal verification of the identities of the Originator and Beneficiary may occur in this stages [FATF]. This stage is outside the scope of the current architecture. * Stage 1: Transfer Initiation Claims negotiations. In this stage gateways G1 and G2 must exchange information (claims) regarding the asset to be transferred, the identity information of the Originator and Beneficiary and other information regarding relevant actors (e.g. gateway owner/operator). * Additionally, the gateways must exchange information regarding the gateway and network characteristics that are unique to G1, G2, NW1 and NW2 for this particular transfer instance. Hardjono, et al. Expires 8 July 2024 [Page 11] Internet-Draft SAT Architecture January 2024 * Stage 2: Lock Assertion and Receipt. In this stage, gateway G1 must provide gateway G2 with a signed assertion that the asset in NW1 has been immobilized and under the control on G1. A signed assertion is needed because NW1 may be a private or closed network, and therefore the state-database (ledger) in NW1 is no readable by external entities including by G2. Gateway G1 must therefore make this signed assertion explicitly. Note that the owner/operator of G1 takes on liability in signing this assertion. * Stage 3: Commitment Preparation and Finalization. In this stage gateways G1 and G2 commit to the unidirectional asset transfer using a 3PC (3-phase commit) subprotocol. These transfer stages will be further discussed below. 6. Transfer Initiation Claims negotiations (Stage-1) The purpose of this stage is for the sender gateway (G1) and the receiver gateway (G2) to agree on the asset instance to be transferred from the origin network NW1 to the destination network NW2. In addition, the gateways must exchange validated information or artifacts regarding the originator and beneficiary of the asset transfer, and exchange gateway-specific and network-specific parameters. These artifacts are contained in the Transfer Initiation Claims set that is sent from gateway G1 to G2. The set of claims may be negotiated between GH1 and G2 in multi-round set of messages. Hardjono, et al. Expires 8 July 2024 [Page 12] Internet-Draft SAT Architecture January 2024 App1 DB1 G1 G2 DB2 App2 | | | | | | | | | | | | |<------------ (transfer context establishment) ------------>| | | | | | | |---request------->| |<------request----| | | | | | | ..|.....|............|......................|............|.....|.. | | | Stage 1 | | | | | | | | | | | (1.1)|<---Proposal Claims-->| | | | | | | | | | | | | | | | | (1.2)|<--Proposal Receipt-->| | | | | | | | | | | | | | | | | (1.3)|<--Transf. Commence-->| | | | | | | | | | | | | | | | | (1.4)|<--- ACK Commence --->| | | | | | | | | ..|.....|............|......................|............|.....|.. | | | | | | Figure 2 This stage starts with the assumption that in network NW1 the gateway who processes the asset transfer has been selected (namely gateway G1). It also assumes that the destination network NW2 has been identified where the beneficiary is located, and that gateway G2 in network NW2 has been identified. The first message (Transfer Proposal Claims) maybe multi-round in the sense there is a negotiation of the claims between G1 and G2. Once G2 accepts the agreed claims, G2 must send a signed receipt carrying the hash of the claims agreed. There are several steps that may occur in Stage 1: * Secure channel establishment between G1 and G2: This includes the mutual verification of the gateway device identities and the exchange of the relevant parameters for secure channel establishment. In cases where device attestation [RATS] is required, the mutual attestation protocol must occur between G1 and G2 prior to proceeding to the next stage. Hardjono, et al. Expires 8 July 2024 [Page 13] Internet-Draft SAT Architecture January 2024 * Mutual device attestations: In cases where device attestation [RATS] is required, each gateway must yield attestation evidence to the other regarding its configuration. A gateway may take on the role as a attestation verifier, or it may rely on an external verifier to appraise the received evidence. * Validation of the gateway ownership: There must be a means for gateway G1 and G2 to verify their respective ownerships (i.e. entities owning G1 and G2 respectively). Examples of ownership verification mechanism include X.509 certificates, directories of gateways and owners, and others. * Validation of owner status: In some jurisdictions, limitations may be placed for regulated asset service providers to transact only with other similarly regulated service providers. Examples of mechanisms used to validate legal status of service providers include directories, Extended Validation (EV) X.509 certificates, and others. * Identification and validation of type/asset profile: Both gateways must agree on the type of asset being transferred based on the published profile of the asset. Gateway G1 must communicate the asset-profile identification to gateway G2, who in turn must validate both the legal status of the asset as well as the technical capability of its network to accept the type of asset. The policies governing network NW2 with regards to permissible incoming assets must be enforced by G2. * Exchange of Travel Rule information and validation: In jurisdictions where the Travel Rule policies regarding originator and beneficiary information is enforced [FATF], the owners of gateways G1 and G2 must comply to the Travel Rule. Mechanisms must be used to permit gateways G1 and G2 to make available originator/beneficiary information to one another in such a away that the Travel Rule information can be logged as part of the asset transfer history. * Negotiation of asset transfer protocol parameters: Gateway G1 and G2 must agree on the parameters to be employed within the asset transfer protocol. Examples include endpoints definitions for resources, type of commitment flows (e.g. 2PC or 3PC), lock-time durations, and others [SAT]. Hardjono, et al. Expires 8 July 2024 [Page 14] Internet-Draft SAT Architecture January 2024 We do not need to invent new standards for several of these steps. Instead, we can rely on existing messaging standards like ISO 20022 [ISO20022] or ITIN [ITIN] for gateway ownership validation, owner status validation, asset profile identification, and communication of travel rule and transfer context information. For identification of digital assets maintained by distributed ledgers or blockchain systems, we can also rely on standards like ITIN [ITIN]. Once gateways G1 and G2 agree on the claims related to the asset transfer, the two gateways can proceed by G1 sending the Transfer Commence message, which must be explicitly acknowledged by gateway G2. 7. Asset Lock Assertion and Receipt (Stage 2) In this stage, gateway G1 must issue a signed assertion that the asset in origin network NW1 has been immobilized and under the control of G1. The steps of Stage 2 are summarized in Figure 4, and broadly consists of the following: * G1 lock/escrow asset (2.1): Gateway G1 proceeds to establish a lock or escrow the asset belonging to the originator. This prevents other local transactions in NW1 from changing the state of the asset until such time the lock by G1 is finalized or released. A time-lock or escrow may also be employed. * Lock Assertion (2.2): Gateway G1 sends a digitally signed assertion regarding the locked (escrowed or immobilized) state on the asset in network NW1. The signature by G1 is performed using its entity public-key pair. This signature signifies that G1 (i.e. its owner/operator) is legally standing behind its statement regarding the locked/escrowed state on the asset. The mechanism to lock or immobilize the asset is outside the scope of SATP. * G2 Logs and Broadcasts lock-assertion information (2.3): Gateway G2 logs a copy of the signed lock-assertion message received in Step 2.4 to its local state data DB2. G2 may also broadcast the fasts of the lock-assertion to all members of network NW2. The mechanism to log and to broadcast is out of scope for SATP. Hardjono, et al. Expires 8 July 2024 [Page 15] Internet-Draft SAT Architecture January 2024 * Lock-Assertion Receipt (2.4): If gateway G2 accepts the signed assertion from G1, then G2 responds with a digitally signed receipt message which includes a hash of the previous lock- assertion message. The signature by G2 is performed using its entity public-key pair. Otherwise, if G2 declines accepting the assertion then G2 can simply ignore the transfer and let the session time-out (i.e. transfer attempt has failed). Orig DB1 G1 G2 DB2 Benef | | | (Stage 1) | | | | | | | | | ..|.....|............|....................|..............|.....|.. | | | Stage 2 | | | | | | | | | | |<---Lock----|(2.1) | | | | | | | | | | | (2.2)|--Lock-Assertion--->| | | | | | | | | | | | (2.3)|--Broadcast-->| | | | | | | | | | | | | | | | |<-----Receipt-------|(2.4) | | | | | | | | ..|.....|............|....................|..............|.....|.. | | | | | | Figure 3 The purpose of the signed lock-assertion is for dispute resolution between G1 and G2 (i.e. the entities who own and operate G1 and G2 respectively) in the case that asset state inconsistencies in NW1 and NW2 are discovered later. The gateway G2 must return a digitally signed receipt to G1 regarding the earlier signed lock-assertion in order to cover G1 (exculpatory proof) in the case of later denial by G2. 8. Commitment Preparation and Finalization (Stage 3) In Stage 3 the gateways G1 and G2 finalizes to the asset transfer by performing a commitment protocol (e.g. 2PC or 3PC) as a process (sub- protocol) embedded within the overall SATP asset transfer protocol. Hardjono, et al. Expires 8 July 2024 [Page 16] Internet-Draft SAT Architecture January 2024 Upon receiving the signed receipt message from G2 in the previous stage, G1 begins the commitment (see Figure 5): * Commit-prepare (3.1): Gateway G1 indicates to G2 to prepare for the commitment of the transfer. This message must include a hash of the previous messages (message 2.5 and 2.6). * Temporary asset mint (3.2): Gateway G2 creates (mints) an equivalent asset in NW2 assigned to itself as the owner. This step can be reversed (i.e. asset destroyed) in the case of the failure in the commitment steps because G2 is still the owner of the asset in NW2. * Commit-ready (3.3): Gateway G2 sends a commit-ready message to G1 indicating that it is ready to carry-out the last steps of the commitment subprotocol. Note that that the entire asset transfer session can be aborted before this step without affecting the asset state in the respective networks. * Asset burn (3.4): Gateway G1 extinguishes (burns) the asset in network NW1 which it has locked since Step 2.3. * Commit-final assertion (3.5): Gateway G1 indicates to G2 that G1 has performed the extinguishment of the asset in NW1. This message must be digitally signed by G1. * Asset-assignment (3.6): Gateway G2 assigns the minted asset (which it has been self-holding since Step 3.2) to the Beneficiary. * ACK-final receipt (3.7): Gateway G2 sends a signed assertion that it has assigned the asset to the intended Beneficiary. * Receipt broadcast (3.8) Gateway G1 logs a copy of the signed receipt message to its local state data DB2. G1 may also broadcast the fasts of the signed receipt to all members of network NW1. The mechanism to log and to broadcast is out of scope for SATP. * Transfer complete (3.9): Gateway G1 must explicitly close the asset transfer session with gateway G2. This allows both sides to close down the secure channel established earlier in Stage 1. Hardjono, et al. Expires 8 July 2024 [Page 17] Internet-Draft SAT Architecture January 2024 Orig DB1 G1 G2 DB2 Benef | | | (Stage 2) | | | | | | | | | ..|.....|.............|....................|............|.....|.. | | | Stage 3 | | | | | | | | | | | (3.1)|--Commit Prepare--->| | | | | | | | | | | | (3.2)|----Mint--->| | | | | | | | | | |<--Commit Ready ----|(3.3) | | | | | | | | | | | | | | | |<---Burn-----|(3.4) | | | | | | | | | | | (3.5)|-Commit Final Asrt->| | | | | | | | | | | | | | | | | | (3.6)|---Assign-->| | | | | | | | | | |<-----ACK Final-----|(3.7) | | | | | | | | | | | | | | | |<-Broadcast--|(3.8) | | | | | | | | | | | (3.9)|-----Completed----->| | | | | | | | | ..|.....|.............|....................|............|.....|.. | | | | | | Figure 4 9. The Commitment sub-protocol Within Stage 3, the gateways must implement one (or more) transactional commitment sub-protocols that permit the coordination between two gateways, and the final commitment of the asset transfer. In the case that there are multiple commitment subprotocols supported by the gateways, the choice of the sub-protocol (type/version) and the corresponding commitment evidence must be negotiated between the gateways during Stage 1. For example, in Stage 2 and Stage 3 discussed above the gateways G1 and G2 may implement the classic 2-Phase or 3-Phase Commit (2PC or 3PC) sub-protocol [Gray81] as a means to ensure efficient and non- disputable commitments to the asset transfer. Hardjono, et al. Expires 8 July 2024 [Page 18] Internet-Draft SAT Architecture January 2024 Historically, transactional commitment protocols employ locking mechanisms to prevent update conflicts on the data item in question. When used within the context of digital asset transfers across networks, the fact that an asset has been locked in NW1 must be communicated via an assertion to G2 (as the 3PC participant) in an indisputable manner. Similarly, G2 must return a signed assertion to G1 that the asset has been regenerated (minted) in NW2. These signed assertions must be verifiable by an authorized third party, in the case that disputes occur (post event) or where legal audit is required on the asset transfer. The precise form of these assertions must be standardized (for the given transactional commitment protocol) to eliminate any ambiguity. 10. Security Considerations As an asset network holds an increasing number of digital assets, it may become attractive to attackers seeking to compromise the cryptographic keys of the entities, services and its end-users. Gateways are of particular interest to attackers because they enable the transferal of digital assets to external networks, which may or may not be regulated. As such, hardening technologies and tamper- resistant crypto-processors (e.g. TPM, SGX) should be used for implementations of gateways [HS19]. 11. Policy Considerations Digital asset transfers must be policy-driven in the sense that it must observe and enforce the policies defined for the network. Resources that make-up a network are owned and operated by entities (e.g. legal persons or organizations), and these entities typically operate within regulatory jurisdictions [FATF]. It is the responsibility of these entities to translate regulatory policies into functions on networks that comply to the relevant regulatory policies. At the application layer, asset transfers must take into consideration the legal status of assets and incorporate relevant asset-related policies into their business logic. These policies must permeate down to the gateways that implement the functions of asset transaction processing. Hardjono, et al. Expires 8 July 2024 [Page 19] Internet-Draft SAT Architecture January 2024 12. Threat Model Considerations Prior to commencing the transfer over the secure channel (TLS), gateways and the operators are assumed to have verified the actor identities, asset artifacts and gateway device-level identification. Relevant assertions are signed by gateways, thereby placing legal and financial liabilities on the operator. However, while adhering to the protocol steps, gateways may purposely take actions that are legitimate from the protocol design perspective, but which may either delay a transfer or cause multiple aborts of a transfer. Delays and aborts in the protocol flow that are late (i.e., stage 3) cause time and resource loss (e.g., network fees). For example, gateway G1 could intentionally delay a transfer by pausing (or simply not continuing) after it receives the commit-ready message (3.3) from G2. If gateway G2 reaches timeout, it may decide to abort the transfer altogether. This would entail G2 reversing the temporary asset in N2 that it established in step 3.2. This reversal could cause some financial loss if network N2 has high transaction fees. Similarly, one or both gateways may purposely fail to send the appropriate signed receipt, such as the commit-final assertion (3.5) from G1 or the ACK-final receipt (3.7) from G2. However, in this case both sides have sufficient evidence to dispute the validity of the transfer. 13. Compatibility Considerations As the asset transfer protocol must be completely agnostic to the anatomy of a digital asset and to the type of ledger technology underlying a system maintaining digital assets, it must be compatible with different asset identification standards like ISO 20022 and ITIN, and with standards for communicating information about business processes (like ISO 20022). Keeping the Stage-0 specification open and not tied to a specific messaging or identification standard allows the Secure Asset Transfer architecture to be flexible and inclusive, and thereby meet compatibility goals. 14. References 14.1. Normative References [FATF] FATF, "International Standards on Combating Money Laundering and the Financing of Terrorism and Proliferation - FATF Revision of Recommendation 15 Hardjono, et al. Expires 8 July 2024 [Page 20] Internet-Draft SAT Architecture January 2024 (Updated June 2021)", October 2018, . [ISO] ISO, "Blockchain and distributed ledger technologies- Vocabulary (ISO:22739:2020)", July 2020, . [ISO20022] ISO, "Universal Financial Industry Message Scheme (ISO 20022).", July 2023, . [ITIN] ITSA, "International Token Identification Number.", July 2023, . [NIST] Yaga, D., Mell, P., Roby, N., and K. Scarfone, "NIST Blockchain Technology Overview (NISTR-8202)", October 2018, . [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, DOI 10.17487/RFC2119, March 1997, . [SAT] Hargreaves, M., Hardjono, T., and R. Belchior, "IETF Secure Asset Transfer Protocol (SATP)", 9 July 2023, .>. 14.2. Informative References [ABCH20] Ankenbrand, T., Bieri, D., Cortivo, R., Hoehener, J., and T. Hardjono, "Proposal for a Comprehensive Crypto Asset Taxonomy", May 2020, . [Abebe19] Abebe, E., Behl, D., Govindarajan, C., Hu, Y., Karunamoorthy, D., Novotny, P., Pandit, V., Ramakrishna, V., and C. Vecchiola, "Enabling Enterprise Blockchain Interoperability with Trusted Data Transfer (Middleware 2019, Industry Track)", December 2019, . [Abebe21] Abebe, E., Hu, Y., Irvin, A., Karunamoorthy, D., Pandit, V., Ramakrishna, V., and J. Yu, "Verifiable Observation of Permissioned Ledgers (ICBC2021)", May 2021, . Hardjono, et al. Expires 8 July 2024 [Page 21] Internet-Draft SAT Architecture January 2024 [BCH21] Belchior, R., Correia, M., and T. Hardjono, "DLT Gateway Crash Recovery Mechanism, IETF, draft-belchior-gateway- recovery-01.", March 2021, . [BVGC20] Belchior, R., Vasconcelos, A., Guerreiro, S., and M. Correia, "A Survey on Blockchain Interoperability: Past, Present, and Future Trends", May 2020, . [Clar88] Clark, D., "The Design Philosophy of the DARPA Internet Protocols, ACM Computer Communication Review, Proc SIGCOMM 88, vol. 18, no. 4, pp. 106-114", August 1988. [DLVIEW] Ramakrishna, V., Pandit, V., Nishad, S., Narayanam, K., and D. Vinayagamurthy, "Views and View Addresses for Blockchain/DLT Interoperability, IETF Draft", November 2021. [Gray81] Gray, J., "The Transaction Concept: Virtues and Limitations, in VLDB Proceedings of the 7th International Conference, Cannes, France, September 1981, pp. 144-154", September 1981. [Herl19] Herlihy, M., "Blockchains From a Distributed Computing Perspective, Communications of the ACM, vol. 62, no. 2, pp. 78-85", February 2019, . [HLP19] Hardjono, T., Lipton, A., and A. Pentland, "Towards and Interoperability Architecture for Blockchain Autonomous Systems, IEEE Transactions on Engineering Management", June 2019, . [HS2019] Hardjono, T. and N. Smith, "Decentralized Trusted Computing Base for Blockchain Infrastructure Security, Frontiers Journal, Special Issue on Blockchain Technology, Vol. 2, No. 24", December 2019, . [IDevID] Richardson, M. and J. Yang, "A Taxonomy of operational security of manufacturer installed keys and anchors. IETF draft-richardson-t2trg-idevid-considerations-01", August 2020, . Hardjono, et al. Expires 8 July 2024 [Page 22] Internet-Draft SAT Architecture January 2024 [SRC84] Saltzer, J., Reed, D., and D. Clark, "End-to-End Arguments in System Design, ACM Transactions on Computer Systems, vol. 2, no. 4, pp. 277-288", November 1984. Authors' Addresses Thomas Hardjono MIT Email: hardjono@mit.edu Martin Hargreaves Quant Network Email: martin.hargreaves@quant.network Ned Smith Intel Email: ned.smith@intel.com Venkatraman Ramakrishna IBM Email: vramakr2@in.ibm.com Hardjono, et al. Expires 8 July 2024 [Page 23]