This document provides an overview and introduction to the Confidential Assets implementation in Elements. A list of relevant RPCs is provided as well as a list of references providing further information.
A working knowledge of Bitcoin and Elements, familiarity with Elements Remote Procedure Calls (RPCs), and some knowledge of the cryptography used in Bitcoin are assumed.
Using Elements, the sender of a transaction can hide the amounts and types of assets in a transaction’s outputs, in such a way that:
- Only the sender and receiver of the transaction can see the actual amounts and types of assets.
- A verifier can prove that all assets coming out of a transaction went into it.
- The amounts and assets of the outputs may be revealed to a third party, by the receiver or by the sender.
This feature is called Confidential Assets. To create a confidential assets transaction, the recipient generates a Confidential Address and provides it to the sender. The sender uses the address to create a Confidential Transaction, and "blinds" its outputs, obscuring the type and quantity of assets while guaranteeing the integrity of the transaction. The recipient may later "unblind" and spend the transaction's outputs at will. The unblinding process is called "rewinding", or "rewinding the range proof", and requires the private blinding key of the confidential address. Either the sender or a receiver may also share a blinding key with a third party, enabling them to view, but not to spend, the transaction's outputs.
Confidential Assets transactions do not conceal the transaction ids or indexes on the inputs (the transaction graph is public, as it is with Bitcoin).
A Confidential Transaction must also include an explicit (unblinded) fee output, paid in the sidechain's default asset (L-BTC for Liquid).
Elements' Confidential Assets capability is built on four cryptographic primitives implemented in the secpk25k1-zkp C++ library [^1]:
In a confidential transaction, asset types and amounts are blinded using Pedersen commitments and mapped to "asset commitments" and "value commitments". These commitments obscure both the asset types and amounts, while allowing a verifier to prove that the sum of a transaction's inputs is equal to the sum of its outputs plus the transaction fee, without needing to see the actual amounts.
When constructing a confidential transaction using Pedersen commitments, it is possible to introduce an output with a negative value, thereby causing coin inflation in other outputs. To prevent this, a cryptographic range proof is included in each blinded transaction output. A range proof allows a verifier to prove that an output's value lies within a specified non-negative range, without having to see the actual value.
When assets are blinded in a transaction, a verifier cannot see which input assets are sent to which outputs. A Surjection proof [^2] allows a verifier to prove that an output’s asset appears in at least one input, without revealing the actual asset type. In other words, the mapping from input assets to the output asset must be an "onto" function, or a "surjection". Every blinded transaction output has a Surjection proof.
The receiver and the sender need a shared secret to blind and unblind
the outputs and to create the range proof. ECDH key exchange is used
to create the shared secret. When generating a Confidential Address,
the receiver creates a private blinding key and embeds the public key
in the address. The sender generates an ephemeral keypair, and with
the address' public blinding key, uses ECDH to generate a shared
secret. The sender's key for the ECDH key exchange is written into the
ConfidentialNonce
field of the transaction output.
See the "Confidential Assets" paper [^3] for more detailed information and references.
The layout and fields of a Confidential Transaction are described in elements-tx-format.md.
Confidential Transactions are significantly larger than non-confidential transactions due to the addition of the range and surjection proofs. Each blinded output adds about 4k bytes to the transaction. As the proofs are stored in output witness fields, each blinded output adds about 1k vBytes to the virtual size of the transaction (please see https://en.bitcoin.it/wiki/Weight_units for information about transaction size measurements).
A range proof proves that a value lies within a specific range. The
exact range is determined by the parameters of the range proof. A
tradeoff exists between the size of the range, and the size of the
proof’s serialization in the transaction output. A larger range
requires a larger proof. In Elements, the default size of the range
proof is 52 bits, allowing a value up to 2^52-1 satoshis to be
represented. In Elements, it is possible to create more than 21
million satoshis of an asset over multiple issuances, but it is not
possible to send more than 21 million in any one unconfidential output. The range
proof parameters are not part of consensus, and may be overridden and
adjusted using the ct_bits
elements configuration parameter. Reducing
the number of bits will reduce the size of a transaction, and also
reduce the maximum provable value of any output.
An Elements range proof is a Borromean ring signature [^4] over possible values of each digit in the base 4 representation of an output value. Each digit requires the storage of 4+1 elliptic curve points. Not including a fixed size header for the range proof, the space required for a range proof in Elements is approximately 80 bytes per bit of precision (default 52).
The secp256k1-zkp range proof implementation supports a maximum range of 64 bits. A range proof supporting a 64 bit range has a size of 5134 bytes.
Bulletproofs are another type of range proof that can be used for confidential transactions. Bulletproofs are much smaller than the range proofs currently used in Elements. See [^5] for more information about Bulletproofs.
A surjection proof is a ring signature over the differences between input and output commitments. The space required for a surjection proof in Elements is proportional to the number of transaction inputs. Each input adds approximately 40 bytes to the surjection proof. In the Elements implementation, a maximum of 256 inputs is supported. See [^2] for more information on serialization size, and [^4] for more information on ring signatures.
A confidential address combines a segwit address and a public blinding key into a single checksummed string. This address format is called "blech32" and is based on the "bech32" format that was introduced for segwit. Liquid production addresses use the prefix "lq1". Liquid regtest (elementsregtest) addresses use the prefix "el1".
By default, the Elements RPC getnewaddress
will return a
confidential address. A non-confidential segwit address and a public
blinding key may be combined with the RPC createblindedaddress
to
create a confidential address.
See the python script ../test/functional/test_framework/liquid_addr.py for a reference implementation of blech32 addresses.
The steps for manually creating a confidential transaction using Elements RPCs are as follows:
createrawtransaction
– adds inputs and outputs to an empty transaction. Any outputs using confidential addresses will be blinded.fundrawtransaction
– adds inputs to pay for the transaction, and an output for change.blindrawtransaction
- blinds the outputs of a transaction, using inputs from the wallet.rawblindrawtransaction
- can be used to blind a transaction whose inputs are not from the local wallet; input blinding factors must be provided.signrawtransaction
- signs inputs using the wallet. At this point, the transaction may be broadcasted, usingsendrawtransaction
.
In the construction of a Pedersen Commitment, random blinding factors
are used in every transaction output value commitment except the
last. The last blinding factor is chosen so that the blinding factors
(and therefore the commitments) sum to zero. As a consequence, when
blinding a transaction, there must be at least one output available to
"balance the blinding factors". The Elements RPCs
blindrawtransaction
and blindrawtransaction
may add an additional
zero-valued output.
An asset issuance creates a non-zero amount of a new asset, and zero or more reissuance token that may be used to create more of the same asset at a later time. Reissuance tokens are also called "inflation keys".
In Elements, there are four types of transaction inputs:
- "typical" inputs that spend UTXOs
- coinbase inputs
- peg-ins
- asset issuances/reissuances.
An asset issuance input defines the ID of a new asset, some non-zero amount of the asset to be issued, and zero or more reissuance tokens. While the ID of the asset must be explicit (it is a property derived from the issuance itself), the amount of the asset issued and the number of reissuance tokens may be blinded in the input.
An asset reissuance input issues an additional amount of an existing asset. The ID of the asset being reissued cannot be blinded, but the amount of additional asset being created can be blinded in the input.
The range proofs for an input's issuance and reissuance amounts are stored in the input witness.
The non-fee outputs of an issuance transaction, as in any transaction in Elements, may be blinded. There will be at least one output for the new asset, an explicit (unblinded) output for the transaction fee, an optional change output, and optionally at least one output for reissuance tokens.
See the elements transaction format document elements-tx-format.md for more information.
The private key used to blind the amount of an issuance or reissuance
input may be revealed or imported into an Elements wallet, using the
RPCs dumpissuanceblindingkey
or importissuanceblindingkey
,
respectively.
In summary, the id of an issued or reissued asset is always explicit, but the issued amounts and destinations may be blinded and kept confidential.
Partially Signed Bitcoin Transactions (PSBT) is a document standard that allows multiple parties to construct and sign a bitcoin transaction offline, before broadcasting it. Elements expands on PSBT to provide support for assets and confidential transactions, with Partially Signed Elements Transactions (PSET).
Several Elements RPCs provide support for working with PSETs. Note that the PSET RPCs in Elements retain "psbt" in their names of RPCs adapted from Bitcoin core.
A description of PSET is outside the scope of this document. Please see pset.mediawiki for more information.
RPCs that are directly related to Confidential Transactions are listed here in the groups listed in the Elements help text. Note that some raw transaction RPCs appear in the Wallet section. See the Elements RPC help for details on parameters and invocation.
createblindedaddress
Creates a confidential address by combining an unconfidential address
and a public blinding key.
analyzepsbt
Provides information about the blinding status of a PSET's outputs.
createpsbt
All blinding parameters may be set explicitly when defining a PSET.
decoderawtransaction
Displays a JSON representation of a raw transaction. Useful for
inspecting confidential parameters. When the wallet is able to unblind
a parameter, the unblinded value will replace the blinded value. The
"ConfidentialNonce" parameter will be displayed as "commitmentnonce".
gettransaction
For a transaction in this wallet, displays properties of the
transaction, including some confidential parameters such as asset and
value blinders ("blinding factors").
getrawtransaction <txid> true
With verbose=true
and txindex=1
, getrawtransaction
will display
public details of a transaction's confidential parameters. Use
unblindrawtransaction
followed by decoderawtransaction
to unblind and
decode the transaction's private parameters.
blindrawtransaction
Blinds the outputs of a raw transaction (as might be created by
createrawtransaction
), using only wallet inputs.
rawblindrawtransaction
Blinds the outputs of a raw transaction (as might be created by
createrawtransaction
). This RPC requires that all blinding factors be
provided explicitly.
getnewaddress
By default, generates a confidential address encoded as blech32 (see
"Confidential Addresses" section above). The public key is embedded in
the address along with the ScriptPubKey. A confidential address is a
tuple (confidential_key, unconfidential address).
getaddressinfo
Displays the (public) confidential and unconfidential properties of an address.
dumpblindingkey
Reveals the private blinding key for a confidential address. A
third-party will need this key to unblind transactions (see
"Third-party Unblinding" below).
dumpissuanceblindingkey
Reveals the private blinding key that was used to blind the amounts on
an issuance input. This key is required when using reissuance tokens.
dumpmasterblindingkey
Reveals the wallet's master blinding key from which all blinding keys for generated addresses are derived. See SLIP-007.
importaddress
A confidential address may be imported at any time. However, in order
to unblind outputs for a confidential address, it is necessary to also
import the blinding key for that address' public blinding key
(called "confidential_key" in the RPC help).
See the importblindingkey
RPC.
importblindingkey
Imports the private blinding key associated with an address.
importissuanceblindingkey
Imports a private blinding key that may be used to unblind the amounts
on an issuance input or to reissue additional amounts of an asset (using
reissuance tokens).
importmasterblindingkey
Use with caution! Importing a master blinding key into a wallet will
erase the existing master blinding key, potentially making all
addresses owned by that wallet unspendable.
blindrawtransaction
Blinds the outputs of a raw transaction (as might be created by createrawtransaction
)
unblindrawtransaction
Unblinds a raw transaction, using addresses and blinding keys known to
the wallet (they must have been previously imported, see "Third-party
Unblinding" below).
A third-party may be granted the ability to unblind the amounts and assets in a confidential transaction, without being able to spend the transaction’s UTXOs. Using Elements, the third-party would create a "watch-only wallet" for the addresses in question, and import the private blinding keys for those addresses.
Let's suppose that Alice has sent a confidential transaction to Bob. Bob wants Victor to be able to see what and how much was sent.
Victor, with Bob’s help, creates a watch-only wallet in Elements:
-
for each address of interest A, Victor runs the RPC
importaddress <A>
-
Bob exports the blinding key for address A using
dumpblindingkey <A>
and shares it with Victor. -
Victor imports the blinding key for the address using
importblindingkey <A> <blinding key>
Once the blinding key is imported, the Elements wallet will treat the Confidential address address as watch-only, and its outputs will be visible in transaction details and in the wallet balance.
Please note that if Bob reuses an address A, Victor will also be able to see the amounts and values in any transaction sending to A.
Alternatively, a watch-only wallet may import the master blinding key of another wallet. The watch-only wallet would then be able to view the UTXOs for any confidential address created by the original wallet.
Anyone with the blinding key for an output's confidential address can rewind the rangeproof for the output, and reveal the blinding factors and actual amounts and assets that were committed to.
Please see the Elements Project tutorial for examples of how to unblind with Elements.
An Elements wallet has a "master blinding key", from which all
blinding keys for that wallet are deterministically derived. A
blinding key for an address is generated as HMAC_SHA256(master blinding key, <address ScriptPubKey>)
. See SLIP-0077 [^6].
Each confidential address has an associated confidential_key, which is a public key embedded in the address and used by the sender to create a nonce. The private key for the confidential_key is called "blinding key" for the address.
See contrib/assets_tutorial/assets_tutorial.py for examples of using confidential transactions with assets.
See test/functional/feature_confidential_transactions.py for functional tests that exercise the confidential assets feature of Elements.
- Mimblewimble
- Monero
- Tari
-
Blockstream Research. libsecp256k1-zkp https://github.com/BlockstreamResearch/secp256k1-zkp Retrieved 2023-03-08
-
Andrew Poelstra. Surjection Proof Module https://github.com/BlockstreamResearch/secp256k1-zkp/blob/master/src/modules/surjection/surjection.md Retrieved 2023-03-08
-
Andrew Poelstra, Adam Back, Mark Friedenbach, Gregory Maxwell, Pieter Wuille. Confidential Assets. https://blockstream.com/bitcoin17-final41.pdf Retrieved 2023-03-08.
-
Gregory Maxwell, Andrew Poelstra. Borromean Ring Signatures https://www.semanticscholar.org/paper/Borromean-Ring-Signatures-%E2%88%97-Maxwell-Poelstra/4160470c7f6cf05ffc81a98e8fd67fb0c84836ea?p2df
-
Benedikt Bunz, Jonathan Bootle, Dan Boneh, Andrew Poelstra, Pieter Wuille, Greg Maxwell. Bulletproofs: Short Proofs for Confidential Transactions and More. https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=8418611 Retrieved 2023-03-08.
-
SLIP-077 Proposal for wallet blinding key derivation. https://github.com/satoshilabs/slips/blob/master/slip-0077.md
-
The Elements Project. Confidential Transactions. https://elementsproject.org/features/confidential-transactions Retrieved 2023-03-08
-
Tari Labs University. Confidential Assets https://tlu.tarilabs.com/digital-assets/confidential-assets Retrieved 2023-03-08
-
Adam Gibson. An investigation into Confidential Transactions. https://github.com/AdamISZ/ConfidentialTransactionsDoc/blob/master/essayonCT.pdf. Retrieved 2023-03-08. This document is a highly readable introduction to CT as they were implemented in the Elements Alpha. Note that it does not describe how asset blinding, surjection proofs, and the method of blinding key derivation described is different than Elements’ current algorithm.
-
Elements Project Developers. Liquid Confidential Transaction Walkthrough with Libwally. https://wally.readthedocs.io/en/release_0.8.8/Liquid/