Zerocoin Anonymous Distributed ECash from Bitcoin
This page was last modified on 20 December 2015, at 15:46.

Abstract: Bitcoin is the first ecash system to see widespread
adoption. While Bitcoin offers the potential for new types of
financial interaction, it has significant limitations regarding
privacy. Specifically, because the Bitcoin transaction log is
completely public, users’ privacy is protected only through the
use of pseudonyms. In this paper we propose Zerocoin, a crypto
graphic extension to Bitcoin that augments the protocol to allow
for fully anonymous currency transactions. Our system uses
standard cryptographic assumptions and does not introduce
new trusted parties or otherwise change the security model of
Bitcoin. We detail Zerocoin’s cryptographic construction, its
integration into Bitcoin, and examine its performance both in
terms of computation and impact on the Bitcoin protocol.
Contents
 1 Introduction
 2 Overview of Bitcoin
 3 Decentralized ecash
 4 Decentralized ecash from strong RSA
 5 Integrating with Bitcoin
 6 Limited Anonymity and Forward Security
 7 Real world security and parameter choice
 8 Performance
 9 Previous work
 10 Conclusion and future work
 11 Contact authors
 12 References
Introduction
Digital currencies have a long academic pedigree. As of yet, however, no system from the academic literature has seen widespread use. Bitcoin, on the other hand, is a viable digital currency with a market capitalization valued at more than $100 million ^{[1]} and between $2 and $5 million USD in transactions a day ^{[2]}. Unlike many proposed digital currencies, Bitcoin is fully decentralized and requires no central bank or authority. Instead, its security depends on a distributed architecture and two assumptions: that a majority of its nodes are honest and that a substantive proofof work can deter Sybil attacks. As a consequence, Bitcoin requires neither legal mechanisms to detect and punish double spending nor trusted parties to be chosen, monitored, or policed. This decentralized design is likely responsible for Bitcoin’s success, but it comes at a price: all transactions are public and conducted between cryptographically binding pseudonyms.
While relatively few academic works have considered the privacy implications of Bitcoin’s design ^{[2]}, ^{[3]}, the preliminary results are not encouraging. In one example, researchers were able to trace the spending of 25,000 bitcoins that were allegedly stolen in 2011 ^{[3]}, ^{[4]}. Although tracking stolen coins may seem harmless, we note that similar techniques could also be applied to trace sensitive transactions, thus violating users’ privacy. Moreover, there is reason to believe that sophisticated results from other domains (e.g., efforts to de anonymize social network data using network topology ^{[5]}) will soon be applied to Bitcoin transaction graph.
Since all Bitcoin transactions are public, anonymous transactions are necessary to avoid tracking by third parties even if we do not wish to provide the absolute anonymity typically associated with ecash schemes. On top of such transactions, one could build mechanisms to partially or explicitly identify participants to authorized parties (e.g., law enforcement). However, to limit this information to authorized parties, we must first anonymize the underlying public transactions.
The Bitcoin community generally acknowledges the privacy weaknesses of the currency. Unfortunately, the available mitigations are quite limited. The most common recommendation is to employ a laundry service which exchanges different users’ bitcoins. Several of these are in commercial operation today ^{[6]}, ^{[7]}. These services, however, have severe limitations: operators can steal funds, track coins, or simply go out of business, taking users’ funds with them. Perhaps in recognition of these risks, many services offer short laundering periods, which lead to minimal transaction volumes and hence to limited anonymity.
Our contribution. In this paper we describe Zerocoin, a distributed ecash system that uses cryptographic techniques to break the link between individual Bitcoin transactions without adding trusted parties. To do this, we first define the abstract functionality and security requirements of a new primitive that we call a decentralized ecash scheme. We next propose a concrete instantiation and prove it secure under standard cryptographic assumptions. Finally, we describe the specific extensions required to integrate our protocol into the Bitcoin system and evaluate the performance of a prototype implementation derived from the original open source bitcoind client.
We are not the first to propose ecash techniques for solving Bitcoin’s privacy problems. However, a common problem with many ecash protocols is that they rely fundamentally on a trusted currency issuer or “bank,” who creates electronic “coins” using a blind signature scheme. One solution (attempted unsuccessfully with Bitcoin ^{[8]}) is to simply appoint such a party. Alternatively, one can distribute the responsibility among a quorum of nodes using threshold cryptography. Unfortunately, both of these solutions introduce points of failure and seem inconsistent with the Bitcoin network model, which consists of many untrusted nodes that routinely enter and exit the network. Moreover, the problem of choosing longterm trusted parties, especially in the legal and regulatory grey area Bitcoin operates in, seems like a major impediment to adoption. Zerocoin eliminates the need for such coin issuers by allowing individual Bitcoin clients to generate their own coins — provided that they have sufficient classical bitcoins to do so.
Intuition behind our construction. To understand the intuition behind Zerocoin, consider the following “pencil and paper” protocol example. Imagine that all users share access to a physical bulletin board. To mint a zerocoin of ﬁxed denomination $1, a user Alice ﬁrst generates a random coin serial number S, then commits to S using a secure digital commitment scheme. The resulting commitment is a coin, denoted C, which can only be opened by a random number r to reveal the serial number S. Alice pins C to the public bulletin board, along with $1 of physical currency. All users will accept C provided it is correctly structured and carries the correct sum of currency.
To redeem her coin C, Alice ﬁrst scans the bulletin board to obtain the set of valid commitments that have thus far been posted by all users in the system. She next produces a noninteractive zeroknowledge proof for the following two statements: (1) she knows a and (2) she knows a hidden value r such that the commitment C opens to S. In full view of the others, Alice, using a disguise to hide her identity, posts a “spend” transaction containing The remaining users verify the proof and check that S has not previously appeared in any other spend transaction. If these conditions are met, the users allow Alice to collect $1 from any location on the bulletin board; otherwise they reject her transaction and prevent her from collecting the currency.
This simple protocol achieves some important aims. First, Alice’s minted coin cannot be linked to her retrieved funds: in order to link the coin C to the serial number S used in her withdrawal, one must either know r or directly know which coin Alice proved knowledge of, neither of which are revealed by the proof. Thus, even if the original dollar bill is recognizably tainted (e.g., it was used in a controversial transaction), it cannot be linked to Alice’s new dollar bill. At the same time, if the commitment and zeroknowledge proof are secure, then Alice cannot doublespend any coin without reusing the serial number S and thus being detected by the network participants.
Of course, the above protocol is not workable: bulletin boards are a poor place to store money and critical information. Currency might be stolen or serial numbers removed to allow double spends. More importantly, to conduct this protocol over a network, Alice requires a distributed digital backing currency.
The ﬁrst and most basic contribution of our work is to recognize that Bitcoin answers all of these concerns, providing us with a backing currency, a bulletin board, and a conditional currency redemption mechanism. Indeed, the core of the Bitcoin protocol is the decentralized calculation of a block chain which acts as a trusted, appendonly bulletin board that can both store information and process ﬁnancial transactions. Alice can add her commitments and escrow funds by placing them in the block chain while being assured that strict protocol conditions (and not her colleagues’ scruples) determine when her committed funds may be accessed.
Of course, even when integrated with the Bitcoin block chain, the protocol above has another practical challenge. Speciﬁcally, it is difﬁcult to efﬁciently prove that a commitment C is in the set . The naive solution is to prove the disjunction Unfortunately such “OR proofs” have size O(N), which renders them impractical for all but small values of N.
Our second contribution is to solve this problem, producing a new construction with proofs that do not grow linearly as N increases. Rather than specifying an expensive OR proof, we employ a “public” oneway accumulator to reduce the size of this proof. Oneway accumulators ^{[9]}, ^{[10]}, ^{[11]}, ^{[12]}, ^{[13]}, ﬁrst proposed by Benaloh and de Mare ^{[9]}, allow parties to combine many elements into a constantsized data structure, while efﬁciently proving that one speciﬁc value is contained within the set. In our construction, the Bitcoin network computes an accumulator A over the commitments , along with the appropriate membership witnesses for each item in the set. The spender need only prove knowledge of one such witness. In practice, this can reduce the cost of the spender’s proof to or even constant size.
Our application requires speciﬁc properties from the accumulator. With no trusted parties, the accumulator and its associated witnesses must be publicly computable and veriﬁable (though we are willing to relax this requirement to include a single, trusted setup phase in which parameters are generated). Moreover, the accumulator must bind even the computing party to the values in the set. Lastly, the accumulator must support an efﬁcient noninteractive witnessindistinguishable or zeroknowledge proof of set membership. Fortunately such accumulators do exist. In our concrete proposal of Section IV we use a construction based on the Strong RSA accumulator of Camenisch and Lysyanskaya ^{[11]}, which is in turn based on an accumulator of Baric and Pﬁtzmann ^{[10]} and Benaloh and de Mare ^{[9]}.
Outline of this work. The rest of this paper proceeds as follows. In Section II we provide a brief technical overview of the Bitcoin protocol. In Section III we formally deﬁne the notion of decentralized ecash and provide correctness and security requirements for such a system. In Section IV we give a concrete realization of our scheme based on standard cryptographic hardness assumptions including the Discrete Logarithm problem and Strong RSA. Finally, in Sections V, VI, and VII, we describe how we integrate our ecash construction into the Bitcoin protocol, discuss the security and anonymity provided, and detail experimental results showing that our solution is practical.
Overview of Bitcoin
In this section we provide a short overview of the Bitcoin protocol. For a more detailed explanation, we refer the reader to the original speciﬁcation of Nakamoto ^{[14]} or to the summary of Barber et al. ^{[2]}.
The Bitcoin network. Bitcoin is a peertopeer network of nodes that distribute and record transactions, and clients used to interact with the network. The heart of Bitcoin is the block chain, which serves as an appendonly bulletin board maintained in a distributed fashion by the Bitcoin peers. The block chain consists of a series of blocks connected in a hash chain. Every Bitcoin block memorializes a set of transactions that are collected from the Bitcoin broadcast network.
Bitcoin peers compete to determine which node will generate the next canonical block. This competition requires each node to solve a proof of work based on identifying speciﬁc SHA256 preimages, speciﬁcally a block B such that SHA256(SHA256(B)) = . The value is selected by a periodic network vote to ensure that on average a block is created every 10 minutes. When a peer generates a valid solution, a process known as mining, it broadcasts the new block to all nodes in the system. If the block is valid (i.e., all transactions validate and a valid proof of work links the block to the chain thus far), then the new block is accepted as the head of the block chain. The process then repeats.
Bitcoin provides two separate incentives to peers that mine new blocks. First, successfully mining a new block (which requires a nontrivial computational investment) entitles the creator to a reward, currently set at 25 BTC.5 Second, nodes who mine blocks are entitled to collect transaction fees from every transaction they include. The fee paid by a given transaction is determined by its author (though miners may exclude transactions with insufﬁcient fees or prioritize high fee transactions).
Bitcoin transactions. A Bitcoin transaction consists of a set of outputs and inputs. Each output is described by the tuple (a, V) where a is the amount, denominated in Satoshi (one bitcoin = Satoshi), and V is a speciﬁcation of who is authorized to spend that output. This speciﬁcation, denoted scriptPubKey, is given in Bitcoin script, a stackbased nonTuringcomplete language similar to Forth.
Transaction inputs are simply a reference to a previous transaction output, as well as a second script, scriptSig, with code and data that when combined with scriptPubKey evaluates to true. Coinbase transactions, which start off every block and pay its creator, do not include a transaction input.
To send d bitcoins to Bob, Alice embeds the hash of Bob’s ECDSA public key , the amount d, and some script instructions in scriptPubKey as one output of a transaction whose referenced inputs total at least d bitcoins (see Figure 2). Since any excess input is paid as a transaction fee to the node who includes it in a block, Alice typically adds a second output paying the surplus change back to herself. Once the transaction is broadcasted to the network and included in a block, the bitcoins belong to Bob. However, Bob should only consider the coins his once at least ﬁve subsequent blocks reference this block. Bob can spend these coins in a transaction by referencing it as an input and including in scriptSig a signature on the claiming transaction under and the public key .
Anonymity. Anonymity was not one of the design goals of Bitcoin ^{[3]}, ^{[14]}, ^{[15]}. Bitcoin provides only pseudonymity through the use of Bitcoin identities (public keys or their hashes), of which a Bitcoin user can generate an unlimited number. Indeed, many Bitcoin clients routinely generate new identities in an effort to preserve the user’s privacy.
Regardless of Bitcoin design goals, Bitcoin’s user base seems willing to go through considerable effort to maintain their anonymity — including risking their money and paying transaction fees. One illustration of this is the existence of laundries that (for a fee) will mix together different users’ funds in the hopes that shufﬂing makes them difﬁcult to trace ^{[2]}, ^{[6]}, ^{[7]}. Because such systems require the users to trust the laundry to both (a) not record how the mixing is done and (b) give the users back the money they put in to the pot, use of these systems involves a fair amount of risk.
Decentralized ecash
Our approach to anonymizing the Bitcoin network uses a form of cryptographic ecash. Since our construction does not require a central coin issuer, we refer to it as a decentralized ecash scheme. In this section we deﬁne the algorithms that make up a decentralized ecash scheme and describe the correctness and security properties required of such a system.
Notation. Let λ represent an adjustable security parameter, let poly(·) represent some polynomial function, and let ν(·) represent a negligible function. We use C to indicate the set of allowable coin values.
Deﬁnition 3.1 (Decentralized ECash Scheme): A decentralized ecash scheme consists of a tuple of possibly randomized algorithms (Setup, Mint, Spend, Verify).
 . On input a security parameter, output a set of global public parameters params and a description of the set C.
 . On input parameters params, output a coin , as well as a trapdoor skc.
 . Given params, a coin c, its trapdoor skc, some transaction string , and an arbitrary set of coins , output a coin spend transaction consisting of a proof and serial number S if . Otherwise output .
 . Given params, a proof π, a serial number S, transaction information R, and a set of coins , output 1 if and is valid. Otherwise output 0.
We note that the Setup routine may be executed by a trusted party. Since this setup occurs only once and does not produce any corresponding secret values, we believe that this relaxation is acceptable for realworld applications. Some concrete instantiations may use different assumptions.
Each coin is generated using a randomized minting algorithm. The serial number S is a unique value released during the spending of a coin and is designed to prevent any user from spending the same coin twice. We will now formalize the correctness and security properties of a decentralized ecash scheme. Each call to the Spend algorithm can include an arbitrary string R, which is intended to store transactionspeciﬁc information (e.g., the identity of a transaction recipient).
Correctness. Every decentralized ecash scheme must satisfy the following correctness requirement. Let and . Let be any valid set of coins, where , and assign . The scheme is correct if, over all , and random coins used in the above algorithms, the following equality holds with probability :
Security. The security of a decentralized ecash system is deﬁned by the following two games: Anonymity and Balance. We ﬁrst describe the Anonymity experiment, which ensures that the adversary cannot link a given coin spend transaction to the coin associated with it, even when the attacker provides many of the coins used in generating the spend transaction.
Deﬁnition 3.2 (Anonymity): A decentralized ecash scheme satisﬁes the Anonymity requirement if every probabilistic polynomialtime (p.p.t.) adversary has negligible advantage in the following experiment.
Anonymity Setup() For : () ← Mint() Spend() Output:
We deﬁne ’s advantage in the above game as [b = b0] − 1/2.
The Balance property requires more consideration. Intuitively, we wish to ensure that an attacker cannot spend more coins than she mints, even when she has access to coins and spend transactions produced by honest parties. Note that to strengthen our deﬁnition, we also capture the property that an attacker might alter valid coins, e.g., by modifying their transaction information string R.
Our deﬁnition is reminiscent of the “onemore forgery” deﬁnition commonly used for blind signatures. We provide the attacker with a collection of valid coins and an oracle that she may use to spend any of them. Ultimately must produce m coins and m+1 valid spend transactions such that no transaction duplicates a serial number or modiﬁes a transaction produced by the honest oracle.
Deﬁnition 3.3 (Balance): A decentralized ecash scheme Π = (Setup, Mint, Spend, Verify) satisﬁes the Balance property if ∀N ≤ poly(λ) every p.p.t. adversary has negligible advantage in the following experiment.
Balance() Setup() For Mint() Output:
The oracle operates as follows: on the query , the oracle outputs if . Otherwise it returns Spend() to and records in the set .
We say that wins (i.e., she produces more spends than minted coins) if where :
 Verify(.
 .
 .
 appears in only one tuple from .
We deﬁne ’s advantage as the probability that wins the above game.
Decentralized ecash from strong RSA
In this section we describe a concrete instantiation of a decentralized ecash scheme. We ﬁrst deﬁne the necessary cryptographic ingredients.
Cryptographic Building Blocks
Zeroknowledge proofs and signatures of knowledge. Our protocols use zeroknowledge proofs that can be instantiated using the technique of Schnorr ^{[16]}, with extensions due to e.g., ^{[17]}, ^{[18]}, ^{[19]}, ^{[20]}. We convert these into noninteractive proofs by applying the FiatShamir heuristic ^{[21]}. In the latter case, we refer to the resulting noninteractive proofs as signatures of knowledge as deﬁned in ^{[22]}.
When referring to these proofs we will use the notation of Camenisch and Stadler ^{[23]}. For instance, denotes a noninteractive zeroknowledge proof of knowledge of the elements and that satisfy both and . All values not enclosed in ()’s are assumed to be known to the veriﬁer. Similarly, the extension indicates a signature of knowledge on message m.
Accumulators.Our construction uses an accumulator based on the Strong RSA assumption. The accumulator we use was ﬁrst proposed by Benaloh and de Mare [10] and later improved by Baric and Pﬁtzmann [11] and Camenisch and Lysyanskaya [12]. We describe the accumulator using the following algorithms:
 . On input a security parameter, sample primes (with polynomial dependence on the security parameter), compute , and sample a seed value . Output as .
 . On input params and a set of prime numbers compute the accumulator as .
 . On input params , a set of prime numbers as described above, and a value , the witness is the accumulation of all the values in besides , i.e., .
 . On input params , an element , and witness , compute and output 1 if and only if , is prime, and as deﬁned previously.
For simplicity, the description above uses the full calculation of A. Camenisch and Lysyanskaya ^{[11]} observe that the accumulator may also be incrementally updated, i.e., given an existing accumulator it is possible to add an element and produce a new accumulator value by computing . We make extensive use of this optimization in our practical implementation.
Camenisch and Lysyanskaya ^{[11]} show that the accumulator satisfies a strong collisionresistance property if the Strong RSA assumption is hard. Informally, this ensures that no p.p.t. adversary can produce a pair such that and yet AccVerify is satisfied. Additionally, they describe an efficient zeroknowledge proof of knowledge that a committed value is in an accumulator. We convert this into a noninteractive proof using the FiatShamir transform and refer to the resulting proof using the following notation:
Our Construction
We now describe a concrete decentralized ecash scheme. Our scheme is secure assuming the hardness of the Strong RSA and Discrete Logarithm assumptions, and the existence of a zeroknowledge proof system.
We now describe the algorithms:
 . On input a security parameter,
run to obtain the values . Next generate primes such that for .Select random generators such that and is a subgroup of . Output .
 . Select and compute such that . Set and output .
 . If output . Compute and . Output where comprises the following signature of knowledge:
 . Given a proof , a serial number , and a set of coins , first compute . Next verify that is the aforementioned signature of knowledge on using the known public values. If the proof verifies successfully, output 1, otherwise output 0.
Our protocol assumes a trusted setup process for generating the parameters. We stress that the accumulator trapdoor is not used subsequent to the Setup procedure and can therefore be destroyed immediately after the parameters are generated. Alternatively, implementers can use the technique of Sander for generating socalled RSA UFOs for accumulator parameters without a trapdoor ^{[24]}.
Security Analysis
We now consider the security of our construction.
Theorem 4.1: If the zeroknowledge signature of knowledge is computationally zeroknowledge in the random oracle model, then satisfies the Anonymity property.
We provide a proof sketch for Theorem 4.1 in Appendix A. Intuitively, the security of our construction stems from the fact that the coin commitment is a perfectlyhiding commitment and the signature proof is at least computationally zeroknowledge. These two facts ensure that the adversary has at most negligible advantage in guessing which coin was spent.
Theorem 4.2: If the signature proof is sound in the random oracle model, the Strong RSA problem is hard, and the Discrete Logarithm problem is hard in , then satisfies the Balance property.
A proof of Theorem 4.1 is included in Appendix A. Briefly, this proof relies on the binding properties of the coin commitment, as well as the soundness and unforgeability of the ZKSoK and collisionresistance of the accumulator. We show that an adversary who wins the Balance game with nonnegligible advantage can be used to either find a collision in the commitment scheme (allowing us to solve the Discrete Logarithm problem) or find a collision in the accumulator (which leads to a solution for Strong RSA).
Integrating with Bitcoin
While the construction of the previous section gives an overview of our approach, we have yet to describe how our techniques integrate with Bitcoin. In this section we address the specific challenges that come up when we combine a decentralized ecash scheme with the Bitcoin protocol.
The general overview of our approach is straightforward. To mint a zerocoin of denomination , Alice runs and stores securely. She then embeds c in the output of a Bitcoin transaction that spends classical bitcoins. Once a mint transaction has been accepted into the block chain, is included in the global accumulator A, and the currency cannot be accessed except through a Zerocoin spend, i.e., it is essentially placed into escrow.
To spend c with Bob, Alice first constructs a partial transaction ptx that references an unclaimed mint transaction as input and includes Bob’s public key as output. She then traverses all valid mint transactions in the block chain, assembles the set of minted coins , and runs . Finally, she completes the transaction by embedding in the scriptSig of the input of . The output of this transaction could also be a further Zerocoin mint transaction — a feature that may be useful to transfer value between multiple Zerocoin instances (i.e., of different denomination) running in the same block chain.
When this transaction appears on the network, nodes check that and check that does not appear in any previous transaction. If these condition hold and the referenced mint transaction is not claimed as an input into a different transaction, the network accepts the spend as valid and allows Alice to redeem bitcoins.
Computing the accumulator. A naive implementation of the construction in section "Decentralized ecash from strong RSA " requires that the verifier recompute the accumulator with each call to . In practice, the cost can be substantially reduced.
First, recall that the accumulator in our construction can be computed incrementally, hence nodes can add new coins to the accumulation when they arrive. To exploit this, we require any node mining a new block to add the zerocoins in that block to the previous block’s accumulator and store the resulting new accumulator value in the coinbase transaction at the start of the new block. We call this an accumulator checkpoint. Peer nodes validate this computation before accepting the new block into the blockchain. Provided that this verification occurs routinely when blocks are added to the chain, some clients may choose to trust the accumulator in older (confirmed) blocks rather than recompute it from scratch.
With this optimization, Alice need no longer compute the accumulator A and the full witness for . Instead she can merely reference the current block’s accumulator checkpoint and compute the witness starting from the checkpoint preceding her mint (instead of starting at ), since computing the witness is equivalent to accumulating .
New transaction types. Bitcoin transactions use a flexible scripting language to determine the validity of each transaction. Unfortunately, Bitcoin script is (by design) not Turingcomplete. Moreover, large segments of the alreadylimited script functionality have been disabled in the Bitcoin production network due to security concerns. Hence, the existing script language cannot be used for sophisticated calculations such as verifying zeroknowledge proofs. Fortunately for our purposes, the Bitcoin designers chose to reserve several script operations for future expansion.
We extend Bitcoin by adding a new instruction: ZEROCOIN_MINT. Minting a zerocoin constructs a transaction with an output whose scriptPubKey contains this instruction and a coin . Nodes who receive this transaction should validate that is a wellformed coin. To spend a zerocoin, Alice constructs a new transaction that claims as input some Zerocoin mint transaction and has a scriptSig field containing and a reference to the block containing the accumulator used in . A verifier extracts the accumulator from the referenced block and, using it, validates the spend as described earlier
Finally, we note that transactions must be signed to prevent an attacker from simply changing who the transaction is payed to. Normal Bitcoin transactions include an ECDSA signature by the key specified in the scriptPubKey of the referenced input. However, for a spend transaction on an arbitrary zerocoin, there is no ECDSA public key. Instead, we use the ZKSoK to sign the transaction hash that normally would be signed using ECDSA.
Statekeeping and side effects. Validating a zerocoin changes Bitcoin’s semantics: currently, Bitcoin’s persistent state is defined solely in terms of transactions and blocks of transactions. Furthermore, access to this state is done via explicit reference by hash. Zerocoin, on the other hand, because of its strong anonymity requirement, deals with existentials: the coin is in the set of thusfarminted coins and its serial number is not yet in the set of spent serial numbers. To enable these type of qualifiers, we introduce side effects into Bitcoin transaction handling. Processing a mint transaction causes a coin to be accumulated as a side effect. Processing a spend transaction causes the coin serial number to be added to a list of spent serial numbers held by the client.
For coin serial numbers, we have little choice but to keep a full list of them per client and incur the (small) overhead of storing that list and the larger engineering overhead of handling all possible ways a transaction can enter a client. The accumulator state is maintained within the accumulator checkpoints, which the client verifies for each received block.
Proof optimizations. For reasonable parameter sizes, the proofs produced by exceed Bitcoin’s 10KB transaction size limits. Although we can simply increase this limit, doing so has two drawbacks: (1) it drastically increases the storage requirements for Bitcoin since current transactions are between 1 and 2 KB and (2) it may increase memory pressure on clients that store transactions in memory.
In our prototype implementation we store our proofs in a separate, wellknown location (a simple server). A full implementation could use a Distributed Hash Table or non blockchain backed storage in Bitcoin. While we recommend storing proofs in the block chain, these alternatives do not increase the storage required for the block chain.
Suggestions for Optimizing Proof Verification
The complexity of the proofs will also lead to longer verification times than expected with a standard Bitcoin transaction. This is magnified by the fact that a Bitcoin transaction is verified once when it is included by a block and again by every node when that block is accepted into the block chain. Although the former cost can be accounted for by charging transaction fees, it would obviously be ideal for these costs to be as low as possible.
One approach is to distribute the cost of verification over the entire network and not make each node verify the entire proof. Because the ZKSoK we use utilizes cutandchoose techniques, it essentially consists of repeated iterations of the same proof (reducing the probability of forgery to roughly ). We can simply have nodes randomly select which iterations of the proofs they verify. By distributing this process across the network, we should achieve approximately the same security with less duplication of effort.
This optimization involves a timespace tradeoff, since the existing proof is verified by computing a series of (at a minimum) 1024 bit values and hashing the result. A naive implementation would require us to send fully computed — greatly increasing the size of the proof – since the client will only compute some of them but needs all of them to verify the hash. We can avoid this issue by replacing the standard hash with a Merkel tree where the leaves are the hashed Ti values and the root is the challenge hash used in the proof. We can then send the 160 bit or 256 bit intermediate nodes instead of the 1024 bit values, allowing the verifier to compute only a subset of the values and yet still validate the proof against the challenge without drastically increasing the proof size.
Limited Anonymity and Forward Security
A serious concern in the Bitcoin community is the loss of wallets due to poor endpoint security. In traditional Bitcoin, this results in the theft of coins ^{[4]}. However, in the Zerocoin setting it may also allow an attacker to deanonymize Zerocoin transactions using the stored . The obvious solution is to securely delete immediately after a coin is spent. Unfortunately, this provides no protection if is stolen at some earlier point.
One solution is to generate the spend transaction immediately (or shortly after) the coin is minted, possibly using an earlier checkpoint for calculating . This greatly reduces the user’s anonymity by decreasing the number of coins in and leaking some information about when the coin was minted. However, no attacker who compromises the wallet can link any zerocoins in it to their mint transactions.
Code Changes
For our implementation, we chose to modify bitcoind, the original opensource Bitcoin C++ client. This required several modifications. First, we added instructions to the Bitcoin script for minting and spending zerocoins. Next, we added transaction types and code for handling these new instructions, as well as maintaining the list of spent serial numbers and the accumulator. We used the Charm cryptographic framework ^{[25]} to implement the cryptographic constructions in Python, and we used Boost’s Python utilities to call that code from within bitcoind. This introduces some performance overhead, but it allowed us to rapidly prototype and leave room for implementing future constructions as well.
Incremental Deployment
As described above, Zerocoin requires changes to the Bitcoin protocol that must happen globally: while transactions containing the new instructions will be validated by updated servers, they will fail validation on older nodes, potentially causing the network to split when a block is produced that validates for some, but not all, nodes. Although this is not the first time Bitcoin has faced this problem, and there is precedent for a flag day type upgrade strategy ^{[26]}, it is not clear how willing the Bitcoin community is to repeat it. As such, we consider the possibility of an incremental deployment.
One way to accomplish this is to embed the above protocol as comments in standard Bitcoin scripts. For non Zerocoin aware nodes, this data is effectively inert, and we can use Bitcoin’s of signature support to specify that such comment embedded zerocoins are valid only if signed by some subset of the Zerocoin processing nodes. Such Zerocoin aware nodes can parse the comments and charge transaction fees for validation according to the proofs embedded in the comments, thus providing an incentive for more nodes to provide such services. Since this only changes the validation mechanism for Zerocoin, the Anonymity property holds as does the Balance property if no more than Zerocoin nodes are malicious.
Some care must be taken when electing these nodes to prevent a Sybil attack. Thankfully, if we require that such a node also produce blocks in the Bitcoin block chain, we have a decent deterrent. Furthermore, because any malfeasance of these nodes is readily detectable (since they signed an invalid Zerocoin transaction), third parties can audit these nodes and potentially hold funds in escrow to deter fraud.
Real world security and parameter choice
Anonymity of Zerocoin
Definition 3.2 states that given two Zerocoin mints and one spend, one cannot do much better than guess which minted coin was spent. Put differently, an attacker learns no more from our scheme than they would from observing the mints and spends of some ideal scheme. However, even an ideal scheme imposes limitations. For example, consider a case where coins are minted, then all coins are subsequently spent. If another coin is minted after this point, the size of the anonymity set for the next spend is , not , since it is clear to all observers that the previous coins have been used. We also stress that — as in many anonymity systems — privacy may be compromised by an attacker who mints a large fraction of the active coins. Hence, a lower bound on the anonymity provided is the number of coins minted by honest parties between a coin’s mint and its spend. An upper bound is the total set of minted coins.
We also note that Zerocoin reveals the number of minted and spent coins to all users of the system, which provides a potential source of information to attackers. This is in contrast to many previous ecash schemes which reveal this information primarily to merchants and the bank. However, we believe this may be an advantage rather than a loss, since the bank is generally considered an adversarial party in most ecash security models. The public model of Zerocoin actually removes an information asymmetry by allowing users to determine when such conditions might pose a problem.
Lastly, Zerocoin does not hide the denominations used in a transaction. In practice, this problem can be avoided by simply fixing one or a small set of coin denominations and exchanging coins until one has those denominations, or by simply using Zerocoin to anonymize bitcoins.
Parameters
Generally, cryptographers specify security in terms of a single, adjustable security parameter . Indeed, we have used this notation throughout the previous sections. In reality, however, there are three distinct security choices for Zerocoin which affect either the system’s anonymity, its resilience to counterfeiting, or both. These are:
 The size of the Schnorr group used in the coin commitments.
 The size of the RSA modulus used in the accumulator.
 , the security of the zeroknowledge proofs.
Commitments. Because Pedersen commitments are information theoretically hiding for any Schnorr group whose order is large enough to fit the committed values, the size of the group used does not affect the long term anonymity of Zerocoin. The security of the commitment scheme does, however, affect counterfeiting: an attacker who can break the binding property of the commitment scheme can mint a zerocoin that opens to at least two different serial numbers, resulting in a double spend. As a result, the Schnorr group must be large enough that such an attack cannot be feasibly mounted in the lifetime of a coin. On the other hand, the size of the signature of knowledge used in coin spends increases linearly with the size of the Schnorr group.
One solution is to minimize the group size by announcing fresh parameters for the commitment scheme periodically and forcing old zerocoins to expire unless exchanged for new zerocoins minted under the fresh parameters. Since all coins being spent on the network at time are spent with the current parameters and all previous coins can be converted to fresh ones, this does not decrease the anonymity of the system. It does, however, require users to convert old zerocoins to fresh ones before the old parameters expire. For our prototype implementation, we chose to use 1024 bit parameters on the assumption that commitment parameters could be regenerated periodically. We explore the possibility of extensions to Zerocoin that might enable smaller groups in section "Conclusion and future work".
Accumulator RSA key. Because generating a new accumulator requires either a new trusted setup phase or generating a new RSA UFO ^{[24]}, we cannot rekey very frequently. As a result, the accumulator is long lived, and thus we truly need long term security. Therefore we currently propose an RSA key of at least 3072 bits. We note that this does not greatly affect the size of the coins themselves, and, because the proof of accumulator membership is efficient, this does not have a large adverse effect on the overall coin spend proof size. Moreover, although rekeying the accumulator is expensive, it need not reduce the anonymity of the system since the new parameters can be used to reaccumulate the existing coin set and hence anonymize spends over that whole history.
Zeroknowledge proof security . This parameter affects the anonymity and security of the zeroknowledge proof. It also greatly affects the size of the spend proof. Thankfully, since each proof is independent, it applies per proof and therefore per spend. As such, a dishonest party would have to expend roughly effort to forge a single coin or could link a single coin mint to a spend with probability roughly . As such we pick bits.
Performance
To validate our results, we conducted several experiments using the modified bitcoind implementation described in Section V. We ran our experiments with three different parameter sizes, where each corresponds to a length of the RSA modulus : 1024 bits, 2048 bits, and 3072 bits.
We conducted two types of experiments: (1) microbenchmarks that measure the performance of our cryptographic constructions and (2) tests of our whole modified Bitcoin client measuring the time to verify Zerocoin carrying blocks. The former gives us a reasonable estimate of the cost of minting a single zerocoin, spending it, and verifying the resulting transaction. The latter gives us an estimate of Zerocoin’s impact on the existing Bitcoin network and the computational cost that will be born by each node that verifies Zerocoin transactions.
All of our experiments were conducted on an Intel Xeon E31270 V2 (3.50GHz quadcore processor with hyperthreading) with 16GB of RAM, running 64bit Ubuntu Server 11.04 with Linux kernel 2.6.38.
Microbenchmarks
To evaluate the performance of our Mint, Spend, and Verify algorithms in isolation, we conducted a series of microbenchmarks using the Charm (Python) implementation. Our goal in these experiments was to provide a direct estimate of the performance of our cryptographic primitives.
Experimental setup. One challenge in conducting our microbenchmarks is the accumulation of coins in for the witness in or for the global accumulator in both and . This is problematic for two reasons. First, we do not know how large will be in practice. Second, in our implementation accumulations are incremental. To address these issues we chose to break our microbenchmarks into two separate experiments. The first experiment simply computes the accumulator for a number of possible sizes of , ranging from 1 to 50,000 elements. The second experiment measures the runtime of the and routines with a precomputed accumulator and witness .
We conducted our experiments on a single thread of the processor, using all three parameter sizes. All experiments were performed 500 times, and the results given represent the average of these times. Figure 3a shows the measured times for computing the coin operations, Figure 3b shows the resulting proof sizes for each security parameter, and Figure 3c shows the resulting times for computing the accumulator. We stress that accumulation in our system is incremental, typically over at most the 200−500 transactions in a block (which takes at worst eight seconds), and hence the cost of computing the global accumulator is therefore amortized. The only time one might accumulate 50,000 coins at one time would be when generating the witness for a very old zerocoin.
Block Verification
How Zerocoin affects network transaction processing determines its practicality and scalability. Like all transactions, Zerocoin spends must be verified first by the miner to make sure he is not including invalid transactions in a block and then again by the network to make sure it is not including an invalid block in the block chain. In both cases, this entails checking that for each Zerocoin transaction and computing the accumulator checkpoint.
We need to know the impact of this for two reasons. First, the Bitcoin protocol specifies that a new block should be created on average once every 10 minutes. If verification takes longer than 10 minutes for blocks with a reasonable number of zerocoins, then the network cannot function. Second, while the cost of generating these blocks and verifying their transactions can be offset by transaction fees and coin mining, the cost of verifying blocks prior to appending them to the block chain is only offset for mining nodes (who can view it as part of the cost of mining a new block). This leaves anyone else verifying the block chain with an uncompensated computational cost.
Experimental setup. To measure the effect of Zerocoin on block verification time, we measure how long it takes our modified bitcoind client to verify externally loaded test blocks containing 200, 400, and 800 transactions where 0, 10, 25, 75, or 100 percent of the transactions are Zerocoin transactions (half of which are mints and half are spends). We repeat this experiment for all three security parameters.
Our test data consists of two blocks. The first contains Zerocoin mints that must exist for any spends to occur. The second block is our actual test vector. It contains, in a random order, Zerocoin spends of the coins in the previous block, Zerocoin mints, and standard Bitcoin sendToAddress transactions. We measure how long the processblock call of the bitcoind client takes to verify the second block containing the mix of Zerocoin and classical Bitcoin transactions. For accuracy, we repeat these measurements 100 times and average the results. The results are presented in Figure 3d.
Discussion
Our results show that Zerocoin scales beyond current Bitcoin transaction volumes. Though we require significant computational effort, verification does not fundamentally threaten the operation of the network: even with a block containing 800 Zerocoin transactions — roughly double the average size of a Bitcoin block currently — verification takes less than five minutes. This is under the unreasonable assumption that all Bitcoin transactions are supplanted by Zerocoin transactions. In fact, we can scale well beyond Bitcoin’s current average of between 200 and 400 transactions per block ^{[27]} if Zerocoin transactions are not the majority of transactions on the network. If, as the graph suggests, we assume that verification scales linearly, then we can support a 50% transaction mix out to 350 transactions per minute (3,500 transactions per block) and a 10% mixture out to 800 transactions per minute (8,000 per block).
One remaining question is at what point we start running a risk of coin serial number collisions causing erroneous double spends. Even for our smallest serial numbers — 160 bits — the collision probability is small, and for the 256 bit serial numbers used with the 3072 bit accumulator, our collision probability is at worst equal to the odds of a collision on a normal Bitcoin transaction which uses SHA256 hashes.
We stress several caveats about the above data. First, our prototype system does not exploit any parallelism either for verifying multiple Zerocoin transactions or in validating an individual proof. Since the only serial dependency for either of these tasks is the (fast) duplicate serial number check, this offers the opportunity for substantial improvement.
Second, the above data is not an accurate estimate of the financial cost of Zerocoin for the network: (a) it is an overestimate of a mining node’s extra effort when verifying proposed blocks since in practice many transactions in a received block will already have been received and validated by the node as it attempts to construct its own contribution to the block chain; (b) execution time is a poor metric in the context of Bitcoin, since miners are concerned with actual monetary operating cost; (c) since mining is typically performed using GPUs and to a lesser extent FPGAs and ASICs, which are far more efficient at computing hash collisions, the CPU cost measured here is likely insignificant.
Finally, our experiment neglects the load on a node both from processing incoming transactions and from solving the proof of work. Again, we contend that most nodes will probably use GPUs for mining, and as such the latter is not an issue. The former, however, remains an unknown. At the very least it seems unlikely to disproportionately affect Zerocoin performance.
Previous work
ECash and Bitcoin
Electronic cash has long been a research topic for cryptographers. Many cryptographic ecash systems focus on user privacy and typically assume the existence of a semitrusted coin issuer or bank. Ecash schemes largely break down into online schemes where users have contact with a bank or registry and offline schemes where spending can occur even without a network connection. Chaum introduced the first online cryptographic ecash system ^{[28]} based on RSA signatures, later extending this work to the offline setting ^{[29]} by deanonymizing users who doublespent. Many subsequent works improved upon these techniques while maintaining the requirement of a trusted bank: for example, by making coins divisible ^{[30]}, ^{[31]} and reducing wallet size ^{[32]}. One exception to the rule above comes from Sander and TaShma ^{[33]} who presciently developed an alternative model that is reminiscent of our proposal: the central bank is replaced with a hash chain and signatures with accumulators. Unfortunately the accumulator was not practical, a central party was still required, and no realworld system existed to compute the chain.
Bitcoin’s primary goal, on the other hand, is not anonymity. It has its roots in a nonacademic proposal by Wei Dai for a distributed currency based on solving computational problems ^{[34]}. In Dai’s original proposal anyone could create currency, but all transactions had to be broadcast to all clients. A second variant limited currency generation and transaction broadcast to a set of servers, which is effectively the approach Bitcoin takes. This is a marked distinction from most, if not all, other ecash systems since there is no need to select one or more trusted parties. There is a general assumption that a majority of the Bitcoin nodes are honest, but anyone can join a node to the Bitcoin network, and anyone can get the entire transaction graph. An overview of Bitcoin and some of its shortcomings was presented by Barber et. al. in ^{[2]}.
Anonymity
Numerous works have shown that “pseudonymized” graphs can be reidentified even under passive analysis. Narayanan and Shmatikov ^{[5]} showed that real world social networks can be passively deanonymized. Similarly, Backstrom et al. ^{[35]} constructed targeted attacks against anonymized social networks to test for relationships between vertices. Previously, Narayanan and Shmatikov deanonymized users in the Netflix prize data set by correlating data from IMDB ^{[36]}.
Bitcoin itself came into existence in 2009 and is now beginning to receive scrutiny from privacy researchers. Deanonymization techniques were applied effectively to Bitcoin even at its relatively small 2011 size by Reid and Harrigan ^{[3]}. Ron and Shamir examined the general structure of the Bitcoin network graph ^{[1]} after its nearly 3fold expansion. Finally, we have been made privately aware of two other earlystage efforts to examine Bitcoin anonymity.
Conclusion and future work
Zerocoin is a distributed ecash scheme that provides strong user anonymity and coin security under the assumption that there is a distributed, online, appendonly transaction store. We use Bitcoin to provide such a store and the backing currency for our scheme. After providing general definitions, we proposed a concrete realization based on RSA accumulators and noninteractive zeroknowledge signatures of knowledge. Finally, we integrated our construction into Bitcoin and measured its performance.
Our work leaves several open problems. First, although our scheme is workable, the need for a doublediscrete logarithm proof leads to large proof sizes and verification times. We would prefer a scheme with both smaller proofs and greater speed. This is particularly important when it comes to reducing the cost of thirdparty verification of Zerocoin transactions. There are several promising constructions in the cryptographic literature, e.g., bilinear accumulators, mercurial commitments ^{[12]}, ^{[37]}. While we were not able to find an analogue of our scheme using alternative components, it is possible that further research will lead to other solutions. Ideally such an improvement could produce a dropin replacement for our existing implementation.
Second, Zerocoin currently derives both its anonymity and security against counterfeiting from strong cryptographic assumptions at the cost of substantially increased computational complexity and size. As discussed in section VIB, anonymity is relatively cheap, and this cost is principally driven by the anticounterfeiting requirement, manifesting itself through the size of the coins and the proofs used.
In Bitcoin, counterfeiting a coin is not computationally prohibitive, it is merely computationally costly, requiring the user to obtain control of at least 51% of the network. This provides a possible alternative to our standard cryptographic assumptions: rather than the strong assumption that computing discrete logs is infeasible, we might construct our scheme on the weak assumption that there is no financial incentive to break our construction as the cost of computing a discrete log exceeds the value of the resulting counterfeit coins.
For example, if we require spends to prove that fresh and random bases were used in the commitments for the corresponding mint transaction (e.g., by selecting the bases for the commitment from the hash of the coin serial number and proving that the serial number is fresh), then it appears that an attacker can only forge a single zerocoin per discrete log computation. Provided the cost of computing such a discrete log is greater than the value of a zerocoin, forging a coin is not profitable. How small this allows us to make the coins is an open question. There is relatively little work comparing the asymptotic difficulty of solving multiple distinct discrete logs in a fixed group, and it is not clear how theory translates into practice. We leave these questions, along with the security of the above proposed construction, as issues for future work.
Finally, we believe that further research could lead to different tradeoffs between security, accountability, and anonymity. A common objection to Bitcoin is that it can facilitate money laundering by circumventing legally binding financial reporting requirements. We propose that additional protocol modifications (e.g., the use of anonymous credentials ^{[38]}) might allow users to maintain their anonymity while demonstrating compliance with reporting requirements.
Acknowledgements. We thank Stephen Checkoway, George Danezis, and the anonymous reviewers for their helpful comments. The research in this paper was supported in part by the Office of Naval Research under contract N0001411 10470, and DARPA and the Air Force Research Laboratory (AFRL) under contract FA87501120211.
Contact authors
Cite error: Invalid <references>
tag;
parameter "group" is allowed only.
<references />
, or <references group="..." />
References
Cite error: Invalid <references>
tag;
parameter "group" is allowed only.
<references />
, or <references group="..." />
Cite error:
<ref>
tags exist for a group named "@:", but no corresponding <references group="@:"/>
tag was found, or a closing </ref>
is missing ↑ ^{1.0} ^{1.1} D. Ron and A. Shamir, “Quantitative Analysis of the Full Bitcoin Transaction Graph,” Cryptology ePrint Archive, Report 2012/584, 2012, http://eprint.iacr.org/.
 ↑ ^{2.0} ^{2.1} ^{2.2} ^{2.3} ^{2.4} S. Barber, X. Boyen, E. Shi, and E. Uzun, “Bitter to better – how to make bitcoin a better currency,” in Financial Cryptography 2012, vol. 7397 of LNCS, 2012, pp. 399–414.
 ↑ ^{3.0} ^{3.1} ^{3.2} ^{3.3} F. Reid and M. Harrigan, “An analysis of anonymity in the Bitcoin system,” in Privacy, security, risk and trust (PASSAT), 2011 IEEE Third Internatiojn Conference on Social Computing (SOCIALCOM). IEEE, 2011, pp. 1318–1326.
 ↑ ^{4.0} ^{4.1} T. B. Lee, “A risky currency? Alleged $500,000 Bitcoin heist raises questions,” Available at http://arstechnica.com/, June 2011.
 ↑ ^{5.0} ^{5.1} A. Narayanan and V. Shmatikov, “Deanonymizing social net works,” in Security and Privacy, 2009 30th IEEE Symposium on. IEEE, 2009, pp. 173–187.
 ↑ ^{6.0} ^{6.1} “Bitcoin fog company,” http://www.bitcoinfog.com/.
 ↑ ^{7.0} ^{7.1} “The Bitcoin Laundry,” http://www.bitcoinlaundry.com/.
 ↑ “Blind Bitcoin,” Information at https://en.bitcoin.it/wiki/Blind Bitcoin Transfers.
 ↑ ^{9.0} ^{9.1} ^{9.2} J. Benaloh and M. de Mare, “Oneway accumulators: a decentralized alternative to digital signatures,” in EUROCRYPT ’93, vol. 765 of LNCS, 1994, pp. 274–285.
 ↑ ^{10.0} ^{10.1} N. Bari ́ c and B. Pfitzmann, “Collisionfree accumulators and failstop signature schemes without trees,” in EUROCRYPT ’97, vol. 1233 of LNCS, 1997, pp. 480–494.
 ↑ ^{11.0} ^{11.1} ^{11.2} ^{11.3} J. Camenisch and A. Lysyanskaya, “Dynamic accumulators and application to efficient revocation of anonymous creden tials,” in CRYPTO ’02, 2002, pp. 61–76.
 ↑ ^{12.0} ^{12.1} L. Nguyen, “Accumulators from bilinear pairings and appli cations,” in Topics in Cryptology – CTRSA 2005, 2005, vol. 3376 LNCS, pp. 275–292.
 ↑ J. Camenisch, M. Kohlweiss, and C. Soriente, “An accumulator based on bilinear maps and efficient revocation for anonymous credentials,” in PKC ’09, vol. 5443 of LNCS, 2009, pp. 481– 500.
 ↑ ^{14.0} ^{14.1} S. Nakamoto, “Bitcoin: A peertopeer electronic cash system, 2009,” 2012. [Online]. Available: http://www.bitcoin.org/ bitcoin.pdf
 ↑ European Central Bank, “Virtual currency schemes,” Available at http://www.ecb.europa.eu/pub/pdf/other/ virtualcurrencyschemes201210en.pdf, October 2012.
 ↑ C.P. Schnorr, “Efficient signature generation for smart cards,” Journal of Cryptology, vol. 4, no. 3, pp. 239–252, 1991.
 ↑ R. Cramer, I. Damg ̊ ard, and B. Schoenmakers, “Proofs of partial knowledge and simplified design of witness hiding protocols,” in CRYPTO ’94, vol. 839 of LNCS, 1994, pp. 174–187.
 ↑ J. Camenisch and M. Michels, “Proving in zeroknowledge that a number n is the product of two safe primes,” in EUROCRYPT ’99, vol. 1592 of LNCS, 1999, pp. 107–122.
 ↑ J. L. Camenisch, “Group signature schemes and payment systems based on the discrete logarithm problem,” Ph.D. dissertation, ETH Z ̈ urich, 1998.
 ↑ S. Brands, “Rapid demonstration of linear relations connected by boolean operators,” in EUROCRYPT ’97, vol. 1233 of LNCS, 1997, pp. 318–333.
 ↑ A. Fiat and A. Shamir, “How to prove yourself: Practical solutions to identification and signature problems,” in CRYPTO ’86, vol. 263 of LNCS, 1986, pp. 186–194.
 ↑ M. Chase and A. Lysyanskaya, “On signatures of knowledge,” in CRYPTO’06, vol. 4117 of LNCS, 2006, pp. 78–96.
 ↑ J. Camenisch and M. Stadler, “Efficient group signature schemes for large groups,” in CRYPTO ’97, vol. 1296 of LNCS, 1997, pp. 410–424.
 ↑ ^{24.0} ^{24.1} T. Sander, “Efficient accumulators without trapdoor extended abstract,” in Information and Communication Security, vol. 1726 of LNCS, 1999, pp. 252–262.
 ↑ J. A. Akinyele, C. Garman, I. Miers, M. W. Pagano, M. Rushanan, M. Green, and A. D. Rubin, “Charm: A framework for rapidly prototyping cryptosystems,” To appear, Journal of Cryptographic Engineering, 2013. [Online]. Available: http://dx.doi.org/10.1007/s1338901300573
 ↑ [Online]. Available: https://en.bitcoin.it/wiki/BIP 0016
 ↑ [Online]. Available: http://blockchain.info/charts/n transactionsperblock
 ↑ D. Chaum, “Blind signatures for untraceable payments,” in CRYPTO ’82. Plenum Press, 1982, pp. 199–203.
 ↑ D. Chaum, A. Fiat, and M. Naor, “Untraceable electronic cash,” in CRYPTO 88, 1990, vol. 403 of LNCS, pp. 319–327.
 ↑ T. Okamoto and K. Ohta, “Universal electronic cash,” in CRYPTO 91, 1992, vol. 576 of LNCS, pp. 324–337.
 ↑ T. Okamoto, “An efficient divisible electronic cash scheme,” in Crypt ’95, 1995, vol. 963 of LNCS, pp. 438–451.
 ↑ J. Camenisch, S. Hohenberger, and A. Lysyanskaya, “Compact ecash,” in EUROCRYPT ’05, 2005, vol. 3494 of LNCS, pp. 566–566.
 ↑ T. Sander and A. TaShma, “Auditable, anonymous electronic cash (extended abstract),” in CRYPTO ’99, vol. 1666 of LNCS, 1999, pp. 555–572.
 ↑ W. Dai. Bmoney proposal. [Online]. Available: http: //www.weidai.com/bmoney.txt
 ↑ L. Backstrom, C. Dwork, and J. Kleinberg, “Wherefore art thou r3579x?: Anonymized social networks, hidden patterns, and structural steganography,” in Proceedings of the 16th international conference on World Wide Web, ser. WWW ’07. New York, NY, USA: ACM, 2007, pp. 181–190.
 ↑ A. Narayanan and V. Shmatikov, “Robust deanonymization of large sparse datasets,” in IEEE Symposium on Security and Privacy. IEEE, 2008, pp. 111–125.
 ↑ M. Chase, A. Healy, A. Lysyanskaya, T. Malkin, and L. Reyzin, “Mercurial commitments with applications to zeroknowledge sets,” in EUROCRYPT ’05, vol. 3494, 2005, pp. 422–439.
 ↑ J. Camenisch and A. Lysyanskaya, “An efficient system for nontransferable anonymous credentials with optional anonymity revocation,” in EUROCRYPT ’01, vol. 2045 of LCNS, 2001, pp. 93–118.
Присоединяйся к команде
ISSN:
Следуй за Полисом
Оставайся в курсе последних событий
License
Except as otherwise noted, the content of this page is licensed under the Creative Commons Creative Commons «AttributionNonCommercialNoDerivatives» 4.0 License, and code samples are licensed under the Apache 2.0 License. See Terms of Use for details.