Master’s Thesis

Designing a cryptocurrency and a

detailed analysis on anonymous
schemes
Chair of Economic Theory
Universität Basel

Supervised by:
Professor Dr. Aleksander Berentsen

Author:
Tobias Petri
Submission Date: January 11, 2018

Abstract

The goal of the following thesis will be to design a digital currency with
preferably a private ledger without losing the effciency of a decentrali-
zed approach and the implied blockchain technology. Furthermore, an
analysis of various anonymity and privacy procedures from an economi-
cal perspective within decentralized schemes will be shown. To preceed
this, there will be a presentation of the vast bulk of promised anonymous
structures from existing cryptocurrencies. After that there will be one
possible implementation of the presented anonymous schemes for the ne-
wly designed cryptocurrency.

1
Contents

1 Introduction 1

2 Designing a Cryptocurrency 3
2.1 Reasons for a New Cryptocurrency . . . . . . . . . . . . . . . . . . . 4

2.2 The Blockchain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

2.2.1 Public Blockchain . . . . . . . . . . . . . . . . . . . . . . . . . 7

2.2.2 Private Blockchain . . . . . . . . . . . . . . . . . . . . . . . . 8

2.3 The Consensus Mechanism . . . . . . . . . . . . . . . . . . . . . . . 10

2.3.1 Permissionless Consenus Mechanism . . . . . . . . . . . . . . 11

2.3.2 Permissioned Consensus Mechanism . . . . . . . . . . . . . . . 14

2.4 Creating the New Cryptocurrency . . . . . . . . . . . . . . . . . . . . 16

2.4.1 The Terminal Overview . . . . . . . . . . . . . . . . . . . . . 16

2.4.2 The Blockchain and Transactions . . . . . . . . . . . . . . . . 17

2.4.3 The Consensus Mechanism . . . . . . . . . . . . . . . . . . . . 23

2.4.4 Further Properties of the Created Cryptocurrency . . . . . . . 28

3 Anonymity Schemes from an Economic Point of View 31

3.1 Mixing the Funds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

3.2 CryptoNote and Ring Signatures . . . . . . . . . . . . . . . . . . . . 35

3.3 Zero Knowledge Proofs . . . . . . . . . . . . . . . . . . . . . . . . . . 37

3.4 Zero Knowledge Succinct Non-interactive Argument of Knowledge

(Zk snarks) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

3.5 Further anonymity schemes . . . . . . . . . . . . . . . . . . . . . . . 42

3.6 ChoiceCoin's Approach to Anonymity . . . . . . . . . . . . . . . . . . 43

4 Conclusion 45

A Appendices i
 A.1 Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i

A.2 Source Codes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv

References xiv

List of Figures

1 Verifying a Transaction Procedure . . . . . . . . . . . . . . . . . . . . 3

2 Dierent Forms of Ledger Levels (based on Baran(1962) and Brennan

et al. (2016) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

3 Example of a Blockchain: Bitcoin provided by Zheng et al. (2016) . . 7

4 The Concept of Hyperledger . . . . . . . . . . . . . . . . . . . . . . . 9

5 Architecture of the Cryptocurrency . . . . . . . . . . . . . . . . . . . 16

6 Code makeTransaction . . . . . . . . . . . . . . . . . . . . . . . . . . 19

7 Code txnRawList . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

8 Code nodeLeader . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

9 Code block creation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

10 Code block size limit . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

11 Code create nal block . . . . . . . . . . . . . . . . . . . . . . . . . . 21

12 Code output rst block . . . . . . . . . . . . . . . . . . . . . . . . . . 21

13 Code state . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

14 Code complete Blockchain: Part1 . . . . . . . . . . . . . . . . . . . . 22

15 Code complete Blockchain: Part2 . . . . . . . . . . . . . . . . . . . . 22

16 Code output general . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

17 24 pre-dened Nodes and 4 UNLs in a Network System . . . . . . . . 24

18 Transaction Process of 24 pre-dened Nodes and four UNLs in a Net-

work System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

19 Process of CoinJoin . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
 20 Process of CryptoNote . . . . . . . . . . . . . . . . . . . . . . . . . . 36

21 Process of Zerocoin by Green et al. (2013) . . . . . . . . . . . . . . . 38

22 Process of Zerocash by Ben-Sasson et al. (2014) . . . . . . . . . . . . 41

23 Process of optional Anonymity . . . . . . . . . . . . . . . . . . . . . . 44

Plagiatserklärung

Ich bezeuge mit meiner Unterschrift, dass meine Angaben über die bei der Abfas-

sung meiner Arbeit benutzten Hilfsmittel sowie über die mir zuteil gewordene Hilfe

in jeder Hinsicht der Wahrheit entsprechen und vollständig sind.

Ich habe das Merkblatt zu Plagiat und Betrug vom 22. Februar 2011 gelesen und

bin mir der Konsequenzen eines solchen Handelns bewusst.

Basel, January 11, 2018

Tobias Petri
1 Introduction

A currency is a form of money that includes coins and paper notes. Generally

speaking each country has its own ocial currency which are typically issued by

each country's central bank, but there are several notable exceptions to this model.

One exception is the Euro which represents the legal tender for multiple European

countries. Moreover, the currency can be used as a medium of exchange for goods

and services, as a unit of account and as a store of value. The denition of a

cryptocurrency is somewhat dierent to a at currency. A cryptocurrency is a type

of digital or a virtual currency that does not possess any physical form and it is

only available in the underlying network. Furthermore, cryptography is used in

order to validate secure transactions. Cryptocurrencies do not need to be issued,

regulated, and controlled by sovereigns and its responsible authorities, meaning the

value of cryptocurrencies is set by the market. In addition, cryptocurrencies have no

intrinsic value, can be seen as a medium of exchange, and most often have entirely

decentralized networks with independent verication procedures.

The world of digital currencies is embedded in a rapidly evolving environment. The

most famous digital currency is Bitcoin. It was introduced in 2009 by Satoshi

Nakamoto and since then has captured the public's attention. Bitcoin does not only

face competition by traditional existing at currencies, backed by central banks, but

also from other cryptocurrencies. Additionally, Bitcoin has a rst-mover advantage

in the cryptocurrency market. It could be an advantage or might be a drawback

in which new digital currencies could prot from a second-mover advantage and

suppress Bitcoin's market dominance in the future. Currently, Bitcoin is still the

digital currency with the highest price and the biggest market capitalization. As of

January 11, 2018 its market price is trading at $13,600 with a market capitalization

of $ 228 billion. The other cryptocurrencies in the top ve are: Ethereum with

a price of $ 1,200 and a market capitalization of $ 117 billion, Ripple with $ 1.70

and market capitalization of $ 67 billion, BitcoinCash with $ 2,500 and market

capitalization of $ 42 billion, and nally Cardano with a price of $ 0.69 and the

market capitalization adds up to $ 18 billion. Ethereum, Ripple, BitcoinCash and

Cardano have a combined market capitalizaition of 36 percent, while Bitcoin has 34

1
percentage points and is dominating the cryptocurrency market.

As some digital currencies have been accepted as a means of payments predominantly

online, central banks are paying increased attention to digital currency issues. For

instance, the European Central Bank is following their impact on monetary policy

and price stability(European Central Bank, Feb. 2015), the US Federal Reserve

1 www.coinmarketcap.com

1
is implementing their ability to deliver a faster, more secure and more ecient

payment system(Bernanke, Nov. 2013), and the UK Treasury is inclined to support

innovation(Johnson et al., 2014)in this area. The Bank for International settlements

(BIS) is also reviewing the intersection between cryptocurrency and central banks

more thoroughly (Quarterly Review, Sep. 2017). These developments do not come

as a surprise due to the current nancial systematic settlements in charge of central

banks. For example, the Clearing House Automated System (CHAPS), Target 2

and Fedwire are cost intensive and have a reduced possibility to innovate.

The motivation to design a new cryptocurrency draws from the existing ones. Nu-

merous digital currencies are created in a short period of time. However, there is no

detailed plan what has to be considered when building a new cryptocurrency from

scratch. Furthermore, the invented currency will have specic properties from dif-

ferent available coins and alternative details that could provide new ways of thinking

and give additional inputs to the cryptocurrency environment.

A further aspect in this paper will be based on the anonymity of cryptocurrencies.

According to Back et. al (2001) the anonymity of a system as a whole is determined

by an ecient, reliable and usable approach including the costs for security issues.

All these factors inuence the size of the user network and the possible degree of

an anonymous achievement. Anonymity and privacy have been a central goal of

cryptography for a long time. David L. Chaum's paper from 1985 called Secu-

rity without identication: Transaction Systems to Make Big Brother Obsolete

discusses anonymous digital cash using blind signatures and the double spend prob-

lem. The title of Chaum's paper indicates what current blockchains should enable:

"a path of regulation without sacricing privacy" (Diedrich (2016), p.254). As an

example with a decentralized structure, Bitcoin does not possess true anonymity.

Their transactions have pseudonymous addresses with hash keys in which there can

be easy linkages between users if a user reuses a Bitcoin address and the transactions

are stored on a public blockchain. In practice, Bitcoin oers only weak anonymous

schemes. For these reasons, the approach of the following thesis will introduce fur-

ther anonymity structures from already existing cryptocurrencies.

To summarize, the thesis will include the following points. Firstly, there will be an

introduction of cryptocurrencies and a provision of motivation for a newly designed

one. After that, an explanation of the new invented digital currency will be provided

including the blockchain with code credentials and a consensus mechanism. In

the next section, the economics of anonymity will be highlighted, followed by the

anonymous structure of various cryptocurrencies and their implementations, and a

possible recommendation of anonymity for the self-invented digital currency. In the

2
concluding remarks, a brief summary of the ndings in the thesis will be presented.

2 Designing a Cryptocurrency

The cryptocurrency Bitcoin solved one big issue, the double-spending problem that

could not be solved with digital cash. Before that, there had to be an intermediary

which conrmed the transactions. However, the blockchain cryptography inhibits

people from spending the same digital cash multiple times by having a decentral-

ized ledger whereon transactions are conrmed. This ledger removes the necessity

of having a central authority in place. The blockchain of Bitcoin can approach

the double-spending problem with a combination of a peer-to-peer structure(P2P)

technology and the private-to-public key cryptography making a new piece of dig-

ital cash. The coin owners and the transactions are stored in the public ledger.

The conrmation of those transactions and the coin ownership is determined by

cryptographic protocols and the miners.

The above example of Bitcoin as a cryptocurrency demonstrates that basically each

cryptocurrency consists of three main components: a blockchain, a currency and a

protocol. Every form of digital cash normally has both a currency and a protocol. In

addition to having a currency and a protocol, cryptocurrencies utilize blockchains,

sometimes their own unique blockchain or others modelled o of existing blockchains

from other ones, such as Bitcoin or Ethereum. Naming an example, Litecoin has its

own currency and protocol, and also relies on the Litecoin blockchain.

Figure 1: Verifying a Transaction Procedure

Figure 1 displays how a transaction in the cryptocurrency sphere can be realized

in general. First, person A gives a notication that she wants to make a transac-

tion from her account to the account of person B. In two and three, the requested

transaction is distributed to a decentralized peer to peer network that consists of

nodes mostly in a computing environment. The nodes in this ledger are trying to

validate the transaction by using dierent consensus algorithms such as proof-of-

work or proof-of-stake approaches. In the fourth step, when the algoritm consensus

has been solved successfully, the transaction can be seen as veried and it can be

combined with transactions from other members. The combination of all these

3
transactions will form a new block of data for the ledger. After that the block can

be attached to the existing blockchain. This blockchain, either visible or restricted

to the public, can hardly be changed and it will stay permanently on the ledger.

Finally, person A and B can be informed that the transaction has been executed

succesfully.

2.1 Reasons for a New Cryptocurrency

Nevertheless, existing digital currencies such as Bitcoin also have drawbacks. Kosba

et al. (2016) are having issues with Bitcoins' privacy and are suggesting a new ap-

proach with its Hawk model. Limited privacy can only be ensured with pseudony-

mous public key implications and the reduced privacy issues are discussed among

others by Androulaki et al. (2013) and Bonneau et al. (2014). Thus, Green et al.

(2013) also suggest improving the anonymity of a user by presenting a new cryp-

tocurrency named Zerocoin. Moreover, Barber et al. (2012) are showing the loss

of Bitcoin coins for instance through malware incidents as an additional disadvan-

tage of Bitcoin. One further weakness involves the reduced transaction scalability.

Whereas PayPal can transfer over 100 transactions per second, Bitcoin is only able

to handle at maximum seven transactions per second and is having diculties in

increasing this rate to higher levels.

Since 2010, there was a one megabyte restriction for the block size to limit the

amount of transactions to seven per second. However, the scalability issue has been

a highly controversial topic for several years in the Bitcoin community itself. Due

to that on the 1st of August 2017, a hard fork took place in which a new Bitcoin

version called BitcoinCash emerged. The original Bitcoin version wanted to increase

its block size to two megabytes (MB) in November 2017. However, there were some

controversies and on December 28, 2017 another hard fork took place introducing

an increase of block size to four MB by Segwit2x when the entire wallets and users

change to Segregated Witness addresses and transactions. BitcoinCash does not

support Segregated Witness. Nevertheless, its block size rises to eight MB meaning

the transactions will increase by eight times per second over today.

The expense of computational energy by dealing with proof-of-work mechanisms

and the condence on broadcast are one component for the limited scalability. The

computational energy to handle the transaction ledger and mitigate double spending

will have the capacity of large power plants or more by some estimation providers.

For instance, in March 2016 the daily energy consumption for the entire mining

network was around 350 Kilowatt (KW) that is approximately the energy consump-

4
tion of 280,000 US households. Deetman (2016) estimates that, at actual growth

rates in computing technology, the Bitcoin ecosystem could consume around 15 Gi-

gawatt (GW) by 2020. This amount is similar to the consumption usage of Denmark

in 2014. There are some existing alternatives such as Permacoin or Litecoin that

can limit the computational energy aspects; however, it cannot reduce the costs

eciently.

Another key limitation of Bitcoin refers to the monetary supply. This involves no

control mechanism over monetary supply after its creation. For instance, Bitcoin will

provide around 21 million coins in total in 2140 while it is not possible to manipulate

the supply which means there is full to almost complete rigid macroeconomic policy

and no exibility. Moreover, the predened total value of 21 million coins in the

algorithm bears the risk of a deation within the Bitcoin system.

Additionally, the value of Bitcoin and other digital currencies can be extremely

volatile as it has been experienced for Bitcoin at the most recent development of its

currency performance with respect to the dollar. For instance the highest price of

Bitcoin was reached so far in December 2017 with a price above $ 20,000. Just a

2
year ago the price was only around $ 780. The daily market volatility is also quite

high meaning that some users might not swap their coins for goods and services and

the means of payment is thereby questionable.

The reasons aforementioned imply that existing cryptocurrencies such as Bitcoin

as an explicit show case do not satisfy the demand in terms of fast transaction

scalability and anonymity properly. However, it is also worth mentioning that one

cryptocurrency cannot satisfy all desired key properties, yet. The newly invented

cryptocurrency should possess a fast transaction scalability and a exible network

system. Thus, the presented cryptocurrency will have the following key properties:

it will use a private blockchain approach, will be highly scalable, provide an op-

tional higher degree of anonymity and a central authorithy will be responsible for

an adequate implementation of the new coins, the sound control of the process, and

validation of its transaction system.

2.2 The Blockchain

Classically, the verication of transaction processes is dominated by intermediaries.

But according to Gomber et. al (2017) dependence on those intermediaries means

that the whole procedure takes a long time and is expensive. It also involves a

credit -and reputation risk if an intermediary is not able to maintain the verication

2 www.coinmarketcap.com

5
process. With the invention of the blockchain technology a user does not have to

rely on intermediation for the verication of transactions any longer. The blockchain

solves all the aforementioned issues with a change from trusting people to trusting

math (Antonopoulos, (2014) p.1) because human interference is no longer needed

for the verication process. In other words, a blockchain is trustless and the user

only has to trust the system and its technology but no other party or intermediary.

A blockchain consists of four basic elements: a replicated ledger, cryptography,

consensus, and business logic. The replicated ledger contains a complete history

of carried out transactions, it has unalterable past modications, it is distributed

and it can be replicated. The integrity of the ledger is one characteristic of the

cryptographic element. Further points include the authenticity of transactions, a

possible degree of anonymity and privacy of transactions and the identity of partic-

ipants. The consensus refers to the decentralized protocol, validated transactions

and a shared control that respects possible disruptions. Finally, the business logic

presents the logic of the dened ledger and gives out the execution together with the

transaction. The logic itself can vary broadly from coins to smart contracts. Based

on these four elements, a blockchain can dier immensely due to the implemented

specications and the inherent variability. For these reasons, only blockchains re-

garding to the cryptocurrency environment, will be considered and their dierent

forms of ledger possibilities will be presented. Following Brennan et al. (2016) three

ledger properties can determine the degree of a blockchain. The rst property is

the amount of copies, the second characteristic is the accessibility of the reader and

nally the write access.

Figure 2: Dierent Forms of Ledger Levels (based on Baran(1962) and Brennan et

al. (2016)

6
In the cryptocurrency sphere, two basic blockchain types can be distinguished: the

private blockchain and the public blockchain.

2.2.1 Public Blockchain

Brennan et al. (2016) dierentiate the public ledger into two parts. For the per-

missionless public ledger any person can participate in the network and view the

transaction history and the consensus mechanism. Furthermore, anyone can also

join the consensus process in which the transactions get veried and to nd out the

actual state of the chains. On the other side, in the permissioned public ledger,

only permissioned entities are allowed to participate in the consensus mechanism

controlling a pre-selected set of nodes. But anyone of them can view and make

transactions upon the blockchain. By comparing both ledgers, the permissionless

public ledger results in increased transparency and accountability.

The public ledger in general is completely distributed and it has a complete trustless

integrity. Swanson (2015) mentions an obvious advantage of a public blockchain is

that someone does not have to prove her identity to the blockchain network. Fur-

thermore, there are no entry barriers for the users and miners except the needed

technology. Morevover, any miner who respects the rules can participate in solving

the consensus mechanism and verify the block for getting the mining reward. A

further positive argument involves the openness of public ledgers. The open envi-

ronment permits many participants to use the system, resulting in gaining network

eects, for instance in the improvement of the blockchain system. Finally, the public

ledger gives users a protection from the developers because one developer does not

have the authorithy to change anything within the system. Two points encourage

this statement. Firstly, it will strengthen trust and ensure increased interactions

within the network. Secondly, the pressure of other entities on oneself will be di-

minished because nobody can force a user to change anything because one user does

not have the authority to proceed. In fewer words, a public blockchain is censorship

resistant. Bitcoin -and Ethereum blockchains are representative examples of a per-

missionless public ledger. Ripple is also an example of a public blockchain, but only

permissioned users can take part in the consensus mechanism, while anyone is able

to make and see transactions on its blockhain.

Figure 3: Example of a Blockchain: Bitcoin provided by Zheng et al. (2016)

7
The typical Bitcoin blockchain example can be seen in Figure 3. In this case the

blockchain has a chain of data blocks in which one block consists of several trans-

1
actions (TX -TX n). The respective blockchain is increased by every additional

block and shows an entire list of the transaction history. In the specic case of

Bitcoin, the miners create the validated blocks that are rewarded by bitcoin coins

and transfer fees. The blocks' validation is done by the network via cryptography.

Moreover, each block has a timestamp, the hash value of the last veried block and

a nonce. The rst block of the blockchain is the so called genesis block and the

following procedure assures the integrity of the complete blockchain including the

genesis block. The hash values only exist once, meaning fraud can be prevented.

The reason for the prevention of fraud is that the changes within the blocks would

immediately change the hash value in the chain. The proof-of-work is a consensus

mechanism in the Bitcoin blockchain which ensures that when the majority of nodes

agree to the validation of transactions within a block, the block can be added to

the existing chain. According to Swanson (2015) the consensus algorithm consists

of a set of rules and processes that enables the maintenance between several nodes.

Newly added transactions are not a part of the ledger automatically. The consensus

procedure gives the permission that the transactions can be part of the block after

a specic time. With Bitcoin, transactions can be conveyed to the ledger after ten

minutes. After the transfer of the block, altering the information in the blockchain

is exceedingly dicult, if not, near impossible.

2.2.2 Private Blockchain

The private ledger consists of one central authority or organization that can give

instructions to the ledger and check the consensus meaning only this authority has

write permissions. The responsible authority also cares about the identity of the

ledger. Read permissions can be public or restricted depending on the desired char-

acteristics and needs of the blockchain.

In general, the private ledger has multiple advantages compared to public legders.

One advantage is that a user has to be an approved entity for participating in the

consensus mechanism in the network. Following the consensus mechanism, public

ledgers use predominantly proof-of-work applications to reach consensus. Private

blockhains do not need to use computer power related mining applications to acquire

consensus because all entities and thereby nodes are known. In this case you can use

algorithms such as Raft or Paxos and more algorithms without any proof-of-work

mining. A detailed analysis of dierent consensus algorithms will be presented in

the next paragraph. As there is no requirement to use any of the mining procedures,

8
the 51% attack that is needed to manipulate the blockchain within the system from

possible miner collusion is not possible. Since the verication process will need

only a few nodes, the transactions are cheaper and more scalable referring to public

blockchains. A further obvious advantage involves the privacy element of private

blockchains. Within private ledgers solely the responsible organization or authority

can change the rights or rules of the blockchain, or see the transactions. In addition

to that, as long as read permissions will be under restriction, a greater form of privacy

will be enhanced which can allow restricted access to the transactions within the

blockchain. Existing examples of cryptocurrencies based on a private blockchain are

Hyperledger, Interledger or Multichain.

Figure 4: The Concept of Hyperledger

In the following, Hyperldeger has one organization and 3 nodes. The preparation of

a transaction is done by the Software Development Kit (SDK). This preparation is

called transaction proposal because this proposal will be forwarded to the nodes. It

can be sent to one or several nodes depending on policy details. All nodes are exe-

cuting a simulation on its respective ledgers. Meaning the key from the transaction

proposal will be updated to a new value set. The nodes sign it with cryptography

and send it as a so called endorsement response to the SDK. The endorsement re-

sponse consists of the cryptographic details of the node and the transaction, and the

updated value of the new set. The SDK collects all endorsement responses, signs

them with a key and transmits them as an invocation request to an orderer. The

orderer has the task to verify the chaincode and its implemented policy. A possible

policy could be that all nodes have to agree on the transaction proposal. However,

if for example only node one got the proposal and sent the endorsement response

to the SDK, the orderer cannot verify the transaction due to unfullled policy re-

9
quirements. Additionally, the orderer veries every endorsement response from all

nodes and the updated set of the new values. Those endorsement requests from all

participating nodes must be the same because the result will consist of nodes that

have the same ledger, the same data and the same chaincodes. If the policy and

the endorsement requests are valid, the orderer sends the updated set of data to all

the nodes as an invocation command with a valid transaction. The participating

nodes accept the new data set and update its respective ledger on the same stage

to keep all nodes in synchronization. Hyperledger is not restricted to only one or-

ganization. It can also have several organizations in parallel executing transactions

with a respective orderer and multiple nodes.

2.3 The Consensus Mechanism

The blockchain's main operation is that the network should agree on the ledger

properties collectively. However, the authority for maintaining the accounts is a

decentralized structure. In order to achieve decentralization on the blockchain, the

network keeps a consensus mechanism around the recorded information. Consensus

mechanisms enable a secure update of distributed states by using a fault tolerance

within the system. The updating process of the replicas in the system follows a

programmed transition rule in the network with the help of a state machine executing

it on every replica. This process ensures that the state will still be in the system if

some nodes crash. The state machine rule allows for the same execution of outputs

of every node, giving the same inputs and resulting in an agreement in the consensus

protocol. In other words, the state transition rules are the rules of the blockchain

protocol.

According to Baliga (2017), a consensus protocol consists of three key characteristics.

The rst property is safety which is when all nodes give out the same output and

the nodes are in line with the protocol. It also refers to the consistency of the

shared states. The second point involves the guarantee of liveness of the consensus

protocol when all valid nodes participate in the consensus producing a nal value.

Fault tolerance is the third property and a consensus protocol can be maintained

if it can live with a few faulty nodes that participate in the consensus process.

Fischer et al. (1985) show that for asynchronous systems all three properties cannot

be satised simultaneously by a deterministic consensus protocol. Fault tolerance

can elaborate in two dierent ways within distributed networks. The rst category,

called fail-stop faults, causes nodes to stop participate in the consensus protocol by

having hardware or software issues. The second form of faults are Byzantine faults.

Lamport et al. (1980) come across the "Byzantine General's problem". A Byzantine

10
node is able to lie, can give out ambiguous results or mislead other nodes which are

part of the consensus protocol. With a limited amount of Byzantine nodes in the

distributed system, the consensus protocol must reach consensus and the consensus

protocol is not allowed to operate incorrectly. As with dierent forms of blockchains,

the consensus protocol also works dierently in a permissionless and a permissioned

setup.

2.3.1 Permissionless Consenus Mechanism

In a permissionless network, the number of nodes is large and unknown. Any node

can join the network meaning the nodes are anonymous and trustless. For this

system, the consensus mechanism has to be responsible for malicious behavior, es-

pecially Sybil attacks. Bitcoin solved the problem with a proof-of-work (PoW)

approach. Other early adopters of cryptocurrencies such as LiteCoin or Monero also

rely on the PoW consensus.

The PoW mechanism

In the Bitcoin PoW mechanism each node has to prove it has done some work in

order to add blocks to the blockchain. Following Diedrich (2016) the work is a

non-ending puzzle challenge in which all nodes participate against each other in the

network. The node looks for a hash value given a correct nonce input which is not

greater than a specic number attached with a diculty level set by the system.

The diculty level depends on the Bitcoin protocol and the currently existing hash

power-the hash rate- of the nodes. On average, one block is produced in a ten

minute interval. To solve the PoW task, the node has to nd a matching hash value

and this process is called mining. The node that nds the wining hash value rst

gets the mining reward. Currently, this mining reward consists of 12.5 newly minted

bitcoins and an optionable transaction fee. The mining reward of the bitcoins halves

every four years and in the future there will be only a transaction fee left. However,

in the race to the wining hash value more than one node can nd a matching hash

value at the same time. This means every winning node informs the network about

its new added block to the blockchain which can result in a temporary fork in the

system. In this case by adding more blocks, the branch with the maximum size will

be part of the blockchain, eventually and the other branch will die out.

According to Baliga (2017) the PoW mechanisms have several weak elements. The

rst includes the possibility of 51% attacks in which the attacker can double-spend

his or her funds or can actively dictate which transactions to include on the block-

hain. Eyal et al. (2013) prove a further weakness of the PoW approach. It is

11
called selsh mining, in which honest mining entities are persuaded to join a 51%

attack. Additionally, other factors that inuence the Bitcoin PoW negatively are a

long duration of transaction conrmation resulting in a poor match of immediate

transaction nality and a high transaction rate. The waste of energy computation

for getting the hash values in the mining process is an additional drawback of Bit-

coin's PoW system. On the other hand, the scalability of the nodes participating in

the network is very good and the system is completely decentralized with open-end

participation.

The Bitcoin PoW is not the only PoW in the cryptocurrency environment. Ethereum

has a separate consensus model known as EthHash. EthHash is conrmed faster and

created on ASIC resistance to ght the 51% attacks that Bitcoin is vulnerable to.

Mining centralization is an additional weak point of Bitcoin and a further reason

that EthHash has been designed. The PoW of Ethereum uses two properties for

tackling mining centralization. The rst one is called memory hardness in which

the computer is able to shift data around in memory unlike pursuing calculations.

The second technique refers to the GHOST protocol that is a revised version of

the Bitcoin PoW. The headers of the recently discarded blocks are contained in the

technique of GHOST. The discarded blocks, called uncle blocks, were part of the

temporary forks and not on the main blockchain. The node that creates the uncle

block and the node that includes the created uncle blocks on the blockchain, receive

a smaller reward to nudge them to work on the current block in the Ethereum

blockchain again. As with the Bitcoin PoW, the EthHash is also looking to nd

a correct nonce input that can give out a hash value below a specic diculty

threshold. Ehereum's PoW also cares about a possible 51% attack. But with the

ASIC design, EthHash can develop a better resistance level in its network than

the Bitcoin PoW. In addition, Ethereum plans to move to a proof-of stake (PoS)

algorithm in the future.

The PoS mechanism

The main disadvantage of PoW consensus protocols involves the high waste of elec-

tricity consumption in the mining process. The PoS mechanism is trying to overcome

this problem by using a dierent approach. Instead of using a mining operation, a

user's stake or ownership of the cryptocurrency will play a part in the blockchain

system. The amount of coins of each participant will be used as a stake to acquire

a part in the cryptocurrency's validation process. This stake is proportional to the

transformation of a user to a validator and the chance of creating blocks in the

blockchain. For instance a user with 200 coins in the system will be two times more

likely to be selected as someone with only 100 coins.

12
Furthermore, the PoS procedure does not need any specialist hardware to mine

the blocks. The PoS algorithm cannot predict the next validator of block creation

as the selection is pseudo-randomized. Some weak PoS algorithms are confronted

with an issue known as "Nothing-at-Stake". In those cases, the nodes have no

incentive to vote on the correct block meaning nodes can vote on several blocks

with several forks in order to enhance their chances of getting remuneration. In

this circumstance a user with nothing to lose has no incentive to behave honestly.

One possible solution is to instruct a validator to store the coins in a form of digital

vault. In case the validator tries to double fork or vote in the network, his or her

coins will be destroyed. Peercoin implemented the PoS mechanism rstly. Other

cryptocurrencies that followed the approach are BitShares, NXT or Tendermint.

Ethereum will release an own PoS algorithm called Casper in the Serenity version

of Ethereum in the future. Casper uses the procedure of digital vaults and hopes to

achieve consensus in this way.

Further "proof-of" mechanisms

Beside the PoW and PoS mechanisms there are numerous consensus mechanisms

around today and it is still an ongoing process in the eld of permissionless se-

tups. An alternative consensus protocol is the proof of elapsed time (PoET). This

algorithm works almost as the PoW approach but with less waste of electricity.

Furthermore, no cryptographic puzzle needs to be solved because the consensus

mechanism relies on a trusted execution environment to assure that the creation of

blocks is done in a randomized lottery and without any work. The proof-of-activity

mechanism is another one that uses a combination of both PoW and PoS. In the rst

stage, there is a mining process until the winning block only includes a header and

the address of the miner's remuneration. In the second stage, the system enables

PoS and it is based on the headers' information while some validators are selected to

sign the newly created block. The ownership of coins of the validator determines the

probability of the validator's selection. The fees are divided between the miners and

the validators. The cryptocurrency Decred uses a variation of the proof-of-activity

algorithm. By sending coins to a non-reachable address in spite of investing money

for the computer hardware, the amount of coins permits the user to mine in the

system with a randomized selection process. This consensus mechanism is called

proof-of burn. The more coins the user burns by sending coins to a non-reachable

address, the greater their chances are to mine the new block. The cryptocurrency

Slimcoin uses a combination of PoW, PoS and the proof-of-burn mechanism. In the

proof-of-capacity environment the user will provide hard drive space. The more a

user will supply, the greater is the possibility to mine the newly created block and

receive the mining reward. Burstcoin implemented the proof-of-capacity approach in

13
its consensus mechanism. Additional variations of the proof-of-capacity algorithms

are proof-of-storage and proof-of-space mechanisms.

2.3.2 Permissioned Consensus Mechanism

In the permissioned case, the amount of nodes is restrictedly small and the identity

is known, meaning they can be semi-trusted in general. For these reasons, com-

pared to the permissionless network an alternative consensus mechanism without

any computer power based mining, such as the PoW, can be implemented. Per-

missioned platforms can rely on and adopt existing algorithms such as Paxos, Raft

or numerous "Byzantine Fault Tolerance" algorithms. Within distributed systems,

those consensus approaches have placed their focus on creating fault tolerance fac-

ing unreliable systems that provided fail-stop faults. Lamport (2001) presents one

form of such an algorithm to implement it in a fault-tolerant distributed environ-

ment called Paxos. This consensus algorithm consists of three roles executed by

three classes namely proposers, acceptors and learners. On the one hand, Paxos

is created to provide a fault-tolerant and consistent approach regarding completely

or temporary failed nodes. On the other hand, it is built to ensure a reliable net-

work in case of unreliably delivered messages. Paxos' consensus algorithm provides

a progressive and consistent procedure which can be reached with a state machine

process within a distributed system.

Ongaro et al. (2014) present a consensus algorithm called Raft which is based on

Paxos. Raft focuses on a replicated log and enhances the element of understand-

ability of its consensus algorithm compared to Paxos. In general, the eciency in

Paxos and Raft are the same unlike the structure to improve understandability and

building of practical systems. Furthermore, there is a separation of the leader elec-

tion, log replication, providing a safe system, and membership changes. The servers

in Raft can also be divided in one of the three states similar to Paxos: a leader, the

followers and the candidates. First, the client goes to all the followers, after that, the

log gets to the selected leader. Finally, all the candidates are used to choose a new

leader. The greatest dierence between Paxos and Raft is the leader selection. In

Raft, the leader selection plays an important part while it is not the case in Paxos'

algorithm. Paxos, Raft and further modications of them can order transactions in

distributed systems. These networks can organize generated requests from clients

and respective state transformations in a distributed environment with the use of

replicated state machines. In such systems 2f + 1 numbers of nodes are needed,

where f is the tolerance level with the amount of fail-stop failures.

Practical Byzantine Fault Tolerance (PBFT) algorithm

14
With the tolerance of Byzantine faults, the consensus protocol got more complex

with additional layers of messaging within the system. Castro et al. (1999) show a

"Practical Byzantine Fault Tolerance" (PBFT) algorithm that can provide consensus

with Byzantine faults. The PBFT approach uses the concept of a replicated state

machine and voting. The replicas can be distinguished into primary and secondary

replicas in which the secondary ones review the proper implementation of decisions

from the primary replicas and can be substituted with a new primary in case the

used one is compromised. The replicas can also be used to optimize the signature

and encryption of messages that have been swapped between replicas and clients. A

further optimization involves the reduction of the size and the amount of messages

exchanged for a smooth implementation in the system amid Byzantine faults. The

algorithm needs 3f +1 replicas to tolerate f faulty nodes. Regarding the performance

of the replicated service the PBFT algorithm institutes a low overhead. Following

experiments in the paper of Castro et al. (1999), a 3% overhead can be reported for

a reproduced network le system (NFS) service. Nevertheless, the PBFT algorithm

is limited to around 20 nodes in terms of scalability and current academic research

results. Above the 20 nodes threshold, the messaging overhead goes up tremendously

due to an increase of the number of replicas.

Ripple consensus algorithm

Following Schwartz et al. (2014) the consensus algorithm of Ripple is inspired by the

Byzantine Fault tolerance with modications of unlimited participation from users,

gateways and also market-makers. Each node needs a denition for a Unique Node

List (UNL). The UNL consists of other Ripple nodes with trust of the existing node

and no collusion against it. A consensus can be reached on the Ripple network when

each node calls other nodes in its UNL. Every UNL needs an interference of 40%

with other nodes in the system. The consensus can be achieved in several rounds

in which each node gathers transactions in a specic data form. The data structure

is known as candidate set and a node relays its candidate sets to other ones. The

validation of the transactions is ensured by the nodes and a voting system. In the

voting system, each node screens its candidate set and sends transactions to the

next round that are getting the largest amount of voting shares. If a candidate set

gets a superma jority, in Ripple's case more than 80% of the accumulated votes from

all the nodes in the UNL, the candidate set can be seen as a valid block or a ledger.

Finally, the ledger is called the Last Closed Ledger and it will be included on the

Ripple blockchain. The appended consensus round begins with new transactions

and transactions that have not been proceeded in the previous round of consensus.

Once each sub-network achieved consensus, the process of consensus in the Ripple

network is completed.

15
2.4 Creating the New Cryptocurrency

As stated before, the key properties of a cryptocurrency are a blockchain, a currency

and a protocol. The newly invented digital currency will be based on a protocol to

send transactions from dierent participants within the network. The peer-to-peer

structure in the validation process of the transactions is an additional protocol that

will be used as it ensures resilience and decentralization aspects for the consensus

mechanism. The currency will be called ChoiceCoin (CC) and its tokens are called

Choice. Furthermore, ChociceCoin's main characteristics are a private blockchain

backed by a central authority, cryptographic usage and optional anonymity which

will be described in the next paragraph.

Figure 5: Architecture of the Cryptocurrency

The architecture of ChoiceCoin will be based on three key modules: the terminal

overview, the blockchain and transaction module, and the consensus mechanism.

2.4.1 The Terminal Overview

As the new cryptocurrency will be based on a restricted validator network, the

central authority will be in charge of the so called terminal overview. One service

component of the terminal includes the registration procedure of the nodes in the

system. Those pre-dened nodes can determine the verication process of the trans-

actions and identify the transaction authority. Furthermore, the central entity will

be responsible for controlling and managing the authorizations of the participants in

the new network. In addition to that, the terminal is also authorized to provide the

disclosure of identities and roles of the participants of the cryptocurrency meaning it

will provide the overall maintenance of the network. In this perspective, the central

entity can support a fast and exible adoption of new rules and orders and facilitate

their access of it to control the environment.

16
To implement or change the existing rules in the network, the pre-dened nodes can

participate in a voting system and if a ma jority is reached to change anything on the

set of rules, the nodes can propose it to the central authority. However, it should also

be noticed that the exercising power of the central authority has to be limited and it

cannot play god in the system without any approval by the participating nodes. One

option of the central entity could be a veto right for inaugurating new rules voted

by the majority of nodes. The auditability property signies the terminal can oer

the provision to ensure authorized participants to handle and allow the transactions

of each user using the network. Furthermore, the central authority can also screen

the complete process of the network system in order to improve the eciency and

scalability. Eventually, the terminal overview will set up a deterministically ordered

transaction list which will be submitted to the node leader in the transaction process.

2.4.2 The Blockchain and Transactions

The transaction process

The transaction procedure in the system works as follows:

1. Any user can propose a transaction to one of the dened nodes that is authorized

in the network and in the acceptance of transactions. This node will save all its

receiving transaction details such as the hashed values and the amount on a separate

transaction list as well.

2. All transactions from the chosen nodes will be collected. After that the central

authority will take all transactions from the nodes that sent them to the authority

and order them numerically with a timestamp to a transaction list. The central

authority is also responsible to delete any double entries or wrong sets of transactions

and hands over the transaction list to the node leader.

3. The node leader will be determined with a random function in the beginning of

each new transaction round. More details will be provided in the code representation

of the blockchain implementation.

4. This node leader is the leader of the verication process and takes back the

transaction list from the central authority to submit it to all validator nodes in the

network.

5. The nodes verify the transactions with a consensus algorithm that will be similar

to Ripple's consensus algorithm with a voting system, an avalanche process and

the need of a supermajority that is more than 80% of the votes.

17
6. After several verication rounds and reaching the supermajority the transaction

list can be seen as valid and the list will be transformed to a new block called

"theAddedBlock" to the existing blockchain.

7. Following the conrmation of the blockchain by all subnodes, the transaction can

be seen as accomplished and a new transaction round can begin.

The mentioned transaction process only shows a proper execution of the network

system without any misbehavior of nodes so far. The treatment of malicious nodes

and possible attacks will be covered in the choice of a stable and reliable consensus

algorithm. Moreover, it should be mentioned that any transactions which did not

get into the rst round of verication for "theAddedBlock" might be a part of the

transaction list for the next verication round and the next block implementation.

The choice of an accurate blockchain type

A major reason why the private blockchain approach will be chosen relates to the

scalability issues of public blockchains. McConaghy et al. (2016) dierentiate the

scalability into three parts: latency, throughput, and capacity and network band-

width. In all mentioned parts, most public blockchain types such as Bitcoin's have

worse properties in those three parts than the permissioned counterparts. As the

blockchain should consist of a high throughput, a low latency and a smaller capacity

and network bandwidth, the private blockcchain is superior to the public ledger pos-

sibility in transferring transactions. Even though the security versus cost trade-o

equips the private case with a lower security than public blockchains, the coming

costs have a bigger impact in choosing the blockchain technology in this case. The

permissioned ledger opportunities do not have to rely on high cost -and unneeded

energy intensive consensus mechanisms.

The consensus mechanisms of public blockchains also aect the transaction scal-

ability negatively. In addition to that, only permissioned entities will be allowed

in the participation of the consensus mechanism in the blockchain. The consensus

mechanism will be presented in the next sub-section. Furthermore, the pre-selected

nodes can be semi-trusted while one central authority can organize and change the

blockchain instructions. This will result in a more exible and faster implementa-

tion of new rules and upcoming changes. Since the anonymous structure will be

introduced in the next section, until now anyone of the pre-dened nodes can view

and make transactions upon the blockchain.

General process of the blockchain

The created blockchain is based on code lines running on the Python programming

18
 3
language. In the following a few key properties will be highlighted to enhance the

understanding of the chosen blockchain approach.

Figure 6: Code makeTransaction

The code snippet "makeTransaction()" displays two randomly produced transaction

pairs from users A to B and C to D. Deposits will be indicated with positive numbers,

while withdrawals will have a negative sign. Each transaction will have a timestamp,

here shown as "timestamp1" and "timestamp2" based on milliseconds. Additionally,

there will be the amount of the transaction for instance "aPays" for the user A and an

individual hash value. However, each node should also save its received transactions

on an extra transaction list in case the transaction details will have to be reviewed

at a later point to prevent fraudulent use of the system.

Figure 7: Code txnRawList

The "txnRawList()" will generate several transactions for each pair of transactions.

In this case 20 transactions will be simulated. This list will be transferred to the

central authority afterwards. There, it will check the transactions in general and

rank it into a deterministically ordered list based on the respective timestamps on

a rst-in rst-out approach. To implement a smooth block creation, there will be a

few checks and set of rules for the users and its transactions. Firstly, coins cannot be

created or destroyed by default. In other words the sum of deposits and withdrawals

of each transaction must be zero. The second condition is that a user's account must

4
be covered with enough funds for withdrawals and mitigate overdraft issues. In case

this set of rules cannot be maintained, the transaction will be rejected.

3 The complete code of the simulated blockchain can be found in the appendix section.
4 Those validity checks are dened in the functions "updateState()" and "isValidTxn()" in the
code section in the appendix

19
 Figure 8: Code nodeLeader

The node leader is based on a randomized function. In this case there are 24 pre-

dened nodes in the network. Thus, the node leader will be a random number

between one and 24. The node leader will be the leader of the following verication

round and relays the transaction list to all other pre-dened nodes. By applying this

working process, it will be ensured that no node will manipulate any transaction

such as omitting transactions which the node does not prefer or pursuing malicious

behavior. In the next round a new node leader will be determined with the help of

the random function generator. After that the verication process begins. The con-

sensus mechanism will be explained in the next sub-section including the consensus

algorithm without any code.

Figure 9: Code block creation

After successfully executing the consensus mechanism, blocks will be built. The

blocks have the following components: a timestamp of the block creation, a block

number, the previous hash of the parent's block, the amount of transactions ("txns")

and the new block hash. The initial states of users can be found in the denition of

5
the genesis block. In this demonstration, the network assumes each user starts with

60 coins. This assumption will simplify the simulation process with the creation of

the blocks and the blockchain in the following case.

Figure 10: Code block size limit

In this demonstration the amount of transaction to form a block is set to four to

show the process in theory. However, in a more practical approach as can be seen

with Ripple, the number of transaction could be 1000 to 1500 transactions per block.

This number is also feasible in the shown approach. The number can be set by the

central authority in the beginning and the dened nodes can also vote to change

this number at a later point in time if desired.

5 see Appendix for source code: "# generating the states and the genesis block"

20
 Figure 11: Code create nal block

If a superma jority in the consensus algorithm for all the transactions within the

transaction list is achieved, the list will be transformed into a new block called

"theAddedBlock" and will be on the existing blockchain (see Figure 11).

Figure 12: Code output rst block

This code snippet shows the output of chain [1]. The block does have the number one

because the genesis block started with zero. The hash of the genesis block is shown

in the previous hash. Furthermore, all transactions of the users are shown under the

"txns" section. In total, there are four transactions ("txnCount") as we dened it in

the "blockSizeLimit" section before. Finally, the new block also has a timestamp and

a new hash value: "faea5f4b6eaf30d782b796e88145b0140954023382fad0fd28b5a78a632f1f7d".

Figure 13: Code state

It is also worth mentioning that with this approach and creating the rst block, the

account balances that are denoted, as "state" in the code section for each user, get

updated immediately with all four transactions for the respective user.

Before the dened nodes can send the block to the blockchain, several checks will be

made to verify the chain validity. The checks involve the hash value of the blocks,

6
the validity of the blocks, and the state of the chain. The hash value check should

return that the block component matches the hash. Verifying the validity of the

blocks makes sure each block is based on the previous block and the current state.

It should return an updated state given the block is valid or otherwise send an error

message. Finally, checking the chains ensures the complete chain is valid including

the genesis block. If this is true it will give back the system state and if not an error

message will be drawn. Those checks are necessary in a blockchain environment

due to the nodes. In this approach only full nodes will be accepted in the system.

6 seeAppendix for source codes of checks: "def checkingBlockHash(block):", "def checkingBlock-

Validity(block, parent, state):" and "def checkingChain(chain):"

21
This property is similar to the Bitcoin system. A full node means the complete

blockchain will be maintained including all transactions on the pre-selected nodes.

Those nodes verify and create the blocks independently starting with the genesis

block until the most recently added block in the network. By checking the validity

of the complete chain, a new node can be protected against misbehavior by others

including invalid transactions. Additionally, the participating nodes in the system

have to check the validity of the newly added blocks to keep an updated blockchain

structure.

The complete blockchain structure can be compressed to the following code lines:

Figure 14: Code complete Blockchain: Part1

The expression "nodeBlockchain" returns the respective chain and the "nodeBlock-

Txns" includes all necessary transactions. In this case the four transactions and the

chain represents a new block entry for a dened node in the network.

Figure 15: Code complete Blockchain: Part2

In the following the "newAddedBlock" is transferred to two nodes, node one and

node two. Those nodes check the block and will update the current state if the

validity of the "newAddedBlock" was successful.

Figure 16: Code output general

In this example, in the beginning the blockchain consists of six blocks on both nodes.

Following that node one and node two check the state of the "newAddedBlock". If

the check was successful, the result will be a new valid block. The new created block

is added to the blockchain thereby the total amount of the blockchain increased to

seven blocks on both nodes eventually.

22
2.4.3 The Consensus Mechanism

To select a reasonable consensus mechanism several properties are crucial. One

indicator is the blockchain type. As aforementioned it will be based on a private

blockchain structure. Thus, the PoW model and with some limitiations the PoS

model, that could also be used in a permissioned setup, will not be ideal for the

setting. The transaction nality is another key issue. A transaction can be seen as

nal as soon as it is added to a block on a blockchain. This procedure is pursued

faster on the consensus algorithms of private blockchains than on public ones. Es-

pecially, the PoW and PoET take a longer time based on the model construction

with the danger of several blocks being mined simultaneously. In other words those

consensus models have probabilistic transaction nality.

Consecutively on the transaction nality argument is the transaction rate. The fol-

lowing consensus should execute and conrm transactions fast. This can be achieved

on a consensus mechanism based on PBFT including the Ripple and Stellar proto-

col and PoS approaches. The cost versus security tradeo also plays a signicant

role to determine a reliable consensus model. Whereas PoW and PoS have a high

cost for the participation process in the consensus with high energy costs in the

PoW approach and high coin expenses for the PoS model, the costs are low for the

others based on a permissioned setup. The characteristic of scalability of the peer

network is higher in the unpermissioned consensus types. An increasing number

of nodes in the consensus system in the permissioned case increases the amount

of messages sent between the nodes exponentially. The consequence is a greater

amount of overhead in the system. Therefore, for a fast system the number of peers

should be around 20 optimally. The level of trust plays no role in the proof of

models in the unpermissioned case. There, the nodes can be untrusted. On the

other hand, in a permissioned environment nodes have to be known. Furthermore,

based on current research at least two thirds of the validating nodes should behave

correctly to maintain a valid consensus process and avoid Byzantine failures. Based

on all these characteristics, the Ripple protocol with slight modications provides

the best option to implement it as a meaningful consensus mechanism for the new

cryptocurrency.

The consensus algorithm

The Ripple protocol is based on the usage of semi-trusted sub-networks in a wider

system. The factor of trust is almost negligible and can be made smaller with an

appropriate selection of member nodes. A consensus mechanism with a low latency

is the consequence of the Ripple protocol with a robust setup in terms of Byzantine

failures and standard failures. Schwartz et al. (2014) dene three main goals that

23
should be satised with the Ripple algorithm: agreement, correctness and utility.

The components to reach consensus of the Ripple protocol are a server, a ledger,

the last-closed ledger, the open ledger, a Unique Node List (UNL) and a proposer.

The server is responsible to run the Ripple Server Software while only registered

user nodes can participate in the consensus procedure. The updated blockchain

with all valid transactions veried in the consensus process is called ledger in the

Ripple protocol. The last-closed ledger is the most current veried block and reects

the recent state of the system. The open ledger represents the current working

block of the nodes reaching consensus. A key specic characteristic of the protocol

involves the UNL. Every server s has a UNL consisting of servers that s asks when

verifying the consensus mechanism. Solely the nodes of s are respected meaning the
UNL can be seen as a sub-network that is trusted by s on a collective perspective.

However, not any participants of the UNL have to be trusted. The proposer element

guarantees only suggestions from servers on the UNL of a server s can be considered
by s itself.

A nonfaulty node behaves honestly in the system, while a faulty node can get an

error either though being honest and getting standard failures based on data issues

or the faulty node performs a malicious behavior with Byzantine errors. To for-

malize a correct transaction in the protocol, an individual validating node makes a

decision given the information on the binary value of zero, implying no success or,

one meaning success. According to Attiya et al. (1984) consensus can be dened

based on three axioms: First, each nonfaulty node decides nitely ( C1). Second, all

nonfaulty nodes get the identical decision value ( C2). The third axiom embodies

for every nonfaulty nodes, zero and one are both feasible values. ( C3).

Figure 17: 24 pre-dened Nodes and 4 UNLs in a Network System

ChoiceCoin's consensus algorithm

24
The following algorithm of ChoiceCoin is based on the Ripple Protocol consensus

algorithm (RPCA) with slight dierences. The RPCA runs every few seconds by all

registered nodes. Unlike in the Ripple algorithm, the validating nodes in ChoiceCoin

cannot be run by the central authority in order to boost the decentralized argument

in the verication process. In gure 17, the new cryptocurrency will start with 24

pre-dened nodes and four UNLs with six nodes each in total. It can be seen that

not all nodes need to be linked to each other to reach consensus consistently. There

will be several rounds to achieve consensus. At rst, every server prepares a list

with all unconrmed valid transactions. After checking the transaction list with the

help of the terminal overview, the node leader gets the transaction list back. The

node leader distributes the transaction list to each server on its UNL and starts the

verication procedure based on a voting mechanism. If a transaction gets more than

a minimum of yes votes, the transaction will be forwarded to the next round if one

is available. Transactions that do not receive sucient votes will not be processed

in this round or delayed for the next consensus process on the next block.

Figure 18: Transaction Process of 24 pre-dened Nodes and four UNLs in a Network

System

Figure 18 shows the verication process of one transaction in a setup of 24 nodes and

four UNLs. The black color signies unconrmed notications of the nodes, while

the white colored nodes veried the transaction as valid. At rst the node leader,

here number two, gets the transaction, veries it, and relays it to other pre-dened

nodes within the UNL environment. In the rst round, it requires 50% of accepted

votes to send the transaction to the next round. In the second consensus round,

the third number in the gure, the bound increments by 10%, thus 60% of positive

votes are required. In the nal round, consensus needs at least 80% of agreement of

a server's UNL meaning the transaction can be added to the "newAddedBlock".

Correctness implies that a distributed system should make a dierence between a

valid and a fraudulent transaction. Based on the algorithm a valid transaction

25
can be achieved as long as a UNL of n nodes within the system has the following

property:

f ≤ (n − 1)/5. (1)

The parameter f denes the number of Byzantine failures. As long as this property

holds, the consensus protocol can be maintained. Another main goal is the agree-

ment argument. In case there is only one possible solution in a decentralized system,

this can be referred to the agreement requirement. To reach the requirement, every

nonfaulty node has to achieve consensus on the same list of transactions irrespective

of its UNLs. In the original white paper of Ripple, a fork is an option as long as

the UNL is smaller than 0.2 ∗ n_total. The parameter n_total is the number of all

nodes in the system. To prove the agreement requirement, there is an upper bound:

| U N Li ∩ U N Lj |≥ 0.2 max(| U N Li |, | U N Lj |) ∨ i, j (2)

The upper bound makes sure two sets of UNLs cannot reach consensus on conicting

transactions because the 80% threshold cannot be passed. However, Armknecht et

al. (2015) investigate a possible fork scenario in the underlying Ripple algorithm

and proved that forks are not possible in the system if and only if:

| U N Li ∩ U N Lj |> 0.4 max(| U N Li |, | U N Lj |) ∨ i, j (3)

This means instead of the 20% upper bound as presented by Ripple developers,

two intersected nodes of UNLs needs to have more than 40% of shared connections

such that forks are not possible. The developers of Ripple agreed to the proved

statement of Armknecht et. al (2015) and in the demonstrated example of gure

17, the intersection amounts to 50% such that a fork is impossible in the algorithm

protocol of ChoiceCoin as well.

In this context utility can be dened as the latency of the network. The Ripple al-

gorithm dierentiates utility into convergence, and heuristics and procedures. The

convergence argument veries the consensus will terminate on a nite time schedule.

Convergence is the point when strong correctness is achieved. In other words, equa-

tion (1) and (3) have to be fullled meaning consensus is satised in nite time. The

crucial element for the algorithm termination is the communication latency between

peers. The time to reply between nodes is screened and nodes that have a larger

response latency than a preset threshold b are deleted from all UNLs. Adding to

the convergence component, there are more heuristics and procedures supporting

26
the utility of the Ripple algorithm. To build the initial transaction list on each

pre-dened node, there is a two second window to make a proposal in each round

of consensus. This will guarantee participation of nodes with solid latency. In the

beginning UNLs will be determined by the central entity. This default list can be

changed by the users at a later point in time if favored based on a voting decision.

Schwartz et al. (2014) mention that consensus can also be reached with solely

one round. However, it might lack utility improvements. Several rounds with an

increase of the needed minimum threshold of agreement in percent provide utility

gains because high-latency nodes can be identied more easily. By processing only

one round of consensus, it could be only some transactions satisfy the 80% bound

and also slow-acting nodes can survive leading to lower transaction rates for the

whole system.

The RPCA can handle transactions in a few seconds and it depends on the required

time to accomplish a round of consensus. The algorithm is not the strongest result

for Asynchronous Byzantine failures with only tolerating a default rate of 20%. Nev-

ertheless, it can provide a quick and cheap network with sound safety and reliability

properties.

Critique on RPCA
Nevertheless, the Ripple algorithm faces some weaker points. Todd (2015) mentions

the optimal choice of UNLs is not answered properly and the rst best option of a

node should be to stick permanently with the default starter UNL provided by the

central entity. Additionally, there is no stated incentive why a node should process

a validation service at all. The Ripple algorithm does not possess a compensation

mechanism with transaction fees or a mining process of new coins. Thus, to reduce

potential risk for legal issues or nancial crime, Todd (2015) proposes making the

validation process on the Ripple ledger private. Armknecht et al. (2015) address

the privacy and anonymity issue of the Ripple algorithm. Similar to Bitcoin, Ripple

relies on pseudonymous signatures of the users to provide a limited degree of pri-

vacy. In addition, the open payment system with publicly announced transactions

limits the privacy argument in the network and privacy could be improved in the

Ripple system. The centralized element of the Ripple system is a further issue.

While there has to be a central authority to provide a permissioned network and

maintainenance, a lot of validating servers are also managed by the central author-

ity called Ripple Labs and diminish the decentralized approach in the verication

process tremendously.

27
2.4.4 Further Properties of the Created Cryptocurrency

In the following a few key properties of ChoiceCoin will be scrutinized in greater

detail.

Cryptography
ChoiceCoin will use private-to-public key cryptography. First, Chaum (1983) comes

up with an early draft of cryptography in 1983. With this approach, cryptographic

elements can only be calculated in one way and are irreversible. Cryptography

ensures the creation of digital secrets and fraud resistant digital signatures. Any

user of the network will have a key pair. The key pair has a private key for each

user to sign spending transactions of its account balance. Based on the private key,

a unique public key will be generated which is used to receive the funds. However,

ChoiceCoin only uses signatures. The ingredients on the blockchain are not fully

encrypted, yet. The property of signature hashes is smilar to Bitcoin or Ethereum.

So far, this approach ensures only pseudonymous transaction operations but no full

anonymity of the transactions and its participants. The public key can be recognized

in the network and all participants are able to see it, while the real world identity

7
of the users cannot be traced back. The algorithm used for the private key hash

creation is based on the SHA256 hash algorithm. While for the public key creation

elliptic curve multiplication is used for cryptographic public keys and it is based on

the secp256k1 approach which can be found in Bitcoin as well.

Coin issuance and money supply

A further key property of ChoiceCoin consists of the handling of the coin issuance

and the money supply. Optimally, the new issued coin should satisfy two goals:

support inclusiveness and maintain a stable store of value. Even under ideal cir-

cumstances, most cryptocurrencies face two kinds of ination. The rst form of

ination refers to price ination meaning an increase of the general price of level of

goods and services in an economy. The second kind of ination can be related to the

money supply due to the implemented issuance process. Ali et al.(2014) emphasize

that many digital currency forms have pre-set money supply paths regulated by

protocols and a xed total supply in the end. They argue that this procedure can

have some problems from a macroeconomic viewpoint. Some examples following

the rigid money supply are deation of goods and services or increased volatility in

prices with possible welfare destruction.

For instance Bitcoin limits its coins to a 21 million xed value covered by its algo-

7 see Appendix for source code based on Python programming language of creating private and
public key hashes

28
rithm until around the year 2100 and reducing the coin mining reward approximately

every four years by half. This mechanism fosters deationary tendencies. Litecoin

will issue 84 million coins while pursuing the same approach of increasing coins in

the system as Bitcoin. In the beginning of cryptocurrency launches, many new coins

were pre-minted. For instance Ixcoin had 580,000 pre-minted coins, or SolidCoin

30,000 pre-minted coins meaning the developers minted the coins themselves for a

period of time before the coin went public for all to mint. However, the pre-mining

process cannot be seen as a fair procedure for all participants in the network and

new issued cryptocurrencies abandoned the approach quickly.

The total supply and the issuance properties of Ethereum's coins, Ether, is not -

nite. The creation of its tokens is restricted to 18 million per year. While the total

issuance is xed, the monetary ination component will decrease every year. The-

oretically, with this approach at some point in the future, the rate of new created

Ether will equalize the calculated loss of misuse such as lost keys, tokens sent to

wrong addresses or also the death of users. But as aforementioned, the Ethereum

network is going to change its consensus algorithm from a PoW to a PoS method

called Casper. The switch signies more eciency and a need for less mining capaci-

ties. With this change the issuance may be altered, but it is still under investigation

for future research.

Ripple followed a dierent method for the coin issuance and supply compared to

other cryptocurrencies. The token called XRP has a total supply of 100 billion coins

and was created instantly at its inception in 2013. There were a lot of discussions

regarding how to approach the issuance and supply issue. Ultimately, Ripple did

not issue all coins to the public at once to avoid market ooding and uncertainty

regarding the price volatility. In May 2017, Ripple still owned around 60% of XRP

tokens and announced that it will put more than half of its XRP supply, 55 billion,

into escrow by the end of the year. Furthermore, around 1 billion will be given out

to the users each month. This communication approach should restrict uncertainty

and provide long-term stability for the immediate future.

Referring to other cryptocurrencies' examples, there is no universal concept dictating

how to give out tokens the best way. Only the respective market capitalization of

the total amount of coins is known. ChoiceCoin will have 42 million coins in total

at the beginning. Out of these 42 million coins one third meaning around 14 million

will be given out in an initial coin oering to all registered participants and the

pre-dened nodes. The remaining 28 million coins will be given out such that the

ination can be assessed reasonably. ChoiceCoin will orientate at the procedure

of Peercoin. Peercoin gives out approximately 1% of coins per year to satisfy the

29
ination argument and uses the PoW meachnism to reward the miners and increase

thereby the coin supply. In ChoiceCoin this increasing supply procedure with the

help of miners is not possible since no PoW method and no miners are needed.

Nevertheless the approach will be based on a lottery system. On the one hand,

50% of the new issued coins will be handed over to the pre-dened nodes, while the

other remaining 50% will be given out to all registered users of the cryptocurrency

network. By using this method the pre-dened nodes also have an incentive to

process validation services honestly.

Those users in the network must have done a transaction before and should be in

the network for at least half a year, since the coin issuance will be based on a 6

month interval. In other words, the coin issuance will take place twice per year and

the users must represent a reliable constituent to avoid inationary account creation

on the platform and possible misuse of the coins. The total money supply will be

reached in the year 2128 if no modications will be pursued. The presented idea

with the cap of 42 million coins is only hypothetical so far and it can be changed

anytime because the coin issuance and the money supply are not hard encoded in

the algorithm and it can be modied if desired by the pre-dened nodes in a voting

proposal to the central authority to provide a maximum degree of exibility.

Transaction fees
Following the original transaction procedure in the blockchain explanation, there is

no transaction fee introduced in the transaction process for ChoiceCoin. However,

a transaction fee will be included. The main reason is to incentivize the pre-dened

nodes to validate transactions properly. Many digital currencies have an advantage

of lower transaction fees compared to existing electronic payment services with credit

cards or international transfer payment systems. Especially, cryptocurrencies with

a mining mechanism, which often times have a subsidy for the miners included.

Bitcoin is a representative example of that process. The subsidy of the process

results in new minted coins for the miner. The degree of the subsidy can have

two sources. The rst one is the actual price of the currency. The other reason

depends on the belief of the mining entity about the currencies' price in the future.

The additional revenue gives miners the possibility for accepting smaller transaction

fees- below the marginal cost of conrming a valid block of transactions. On a short-

term basis, the subsidy with the new created currency enables an incentive for the

miners to support and help making the currency more established. In the long term

the supply of money is xed and the mining process with the subsidy mechanism

will disappear. Those currencies will face competition with other payment networks

on the cost issue. In this case, higher marginal costs of cryptocurrencies could cause

30
a competitive disadvantage to centralized systems which can rely on economies of

scale.

In Solidcoin, the fee amounted to always one coin irrespective whether the transac-

tion was only worth one coin or several coins in total. In Bitcoin, the transaction fee

is calculated on the magnitude of the transaction in kilobytes, whereas in Choice-

Coin the transaction fee will depend proportionally on the value of the transaction.

For instance a bigger transaction will have a bigger transaction fee. The standard

fee will amount to 0.5 - 1% of the total transaction size at the beginning and can

be changed anytime if desired by the pre selected nodes. Currently, the average

credit card processing fees are between 1.5 to 2%, while Paypal charges around 2%

on average for its service. The reason for oering such a small fee is that Choice-

Coin can handle from 1000 to 1500 transactions per block in a short period of time.

With that amount of transaction handling, the nodes can rely on the economies

of scale generating revenue that is higher than the marginal costs for the nodes.

Nevertheless, the pre-dened nodes should not prioritize transactions with greater

fees. Thus, they should treat each transaction with the same degree of importance

without the inuence of the fees itself. The receiving node which gets the transac-

tion order will collect the fee before adding the transaction to the transaction list.

Amid pre-dened nodes, a second transaction fee method is possible such that each

pre-dened node will get a pre-determined fee for each consensus round. However,

this method can be introduced at a later point in time if some nodes want to switch

to this approach in the network.

3 Anonymity Schemes from an Economic Point of

View

Following Acquisti et al. (2003) an anonymous structure cannot be built by the

sender or recipient in a general system. A user is not able to decision the sending

of anonymous messages. The anonymity depends on the infrastructure and the

distribution of trust to the underlying network. A node will be running in a shared

network if the associated incentives are greater than the marginal costs of supporting

the reliability of the network. Nevertheless, the associated costs are immense for the

running nodes. Besides the bandwidth and processing power, the costs also include

the reputation risk of the nodes and the right choice of selecting valid transactions

from fraudulent ones. Hiding messages is a crucial part in anonymous networks. The

senders consume anonymity and provide the cover trac that builds the anonymous

structures for other participants. Users have an advantage on broad systems due to

31
the noise provided by others.

A strong anonymity system relies on high trac in the network, while the high

trac can imply a better performance. The better performance is explained by a

smooth process of messages without any delays that is required on a light trac

network to provide anonymous schemes. However, networks with high trac do not

create the best option to hide at all, because if the degree of trust is too centralized

an extensive system can be targeted by insiders and attackers. The attacks can be

various. For instance attackers can chase for the eciency or reliability of nodes

or raise the cost of the running nodes. Back et al. (2001) make clear that an

anonymous network has to take into account the security objectives that consist of

an ecient, reliable and usable approach with the costs included. The security aim

inuences the size of the users in the network and it makes a reference to the possible

degree of anonymity. For decentralized systems similar to existing cryptocurrencies'

approaches, Acquisti et al.(2003) nd out that providing a solid level of trust, those

systems can have unbearable coordination costs. Those costs could only be faced if

a central authority can be implemented. Nevertheless, a central authority could be

vulnerable to attacks exploiting the trust argument.

Privacy and anonymity are also two dierent components in this environment. Ishai

et al. (2006) make clear that anonymity specializes on hiding the responsible users

that performed the action. The privacy argument focuses on what action has been

performed in the network.

Cryptography does not enable anonymity and privacy automatically. Privacy has

been one goal of cryptography for a long time. Chaum (1985) discusses anonymous

digital cash using blind signatures. While many cryptocurrencies with a decen-

tralized structure, take Bitcoin for instance, do not possess true anonymity. Their

transactions are recorded on the public blockchain. However, the true identity of

the users is not available because only the pseudonymous addresses with hash keys

are used. Additionally, if a user transaction has been identied all of that user's

transactions could be unveiled if they do not use a new pseudonymous address ev-

ery time they make a transaction. In practice, Bitcoin oers only weak anonymous

schemes. In general, Blockchain anonymity and privacy are not easy to reach es-

pecially for public blockchains because those ledgers should make all transactions

visible and should generate the verication of the supply of coins. Anonymity and

privacy mechanisms have to present a solution to both issues that can end up in

conicts between the protection of privacy and the maintenance of public veria-

bility. For these reasons, this thesis will include a discussion regarding additional

anonymity and privacy structures from already existing cryptocurrencies.

32
3.1 Mixing the Funds

One possibility that can be used is to mix the funds with other participants. Imagine

there is one group of users and each user will give the same amount of tokens to a

bucket. After that the tokens in the bucket will be mixed and each user will get back

the same starting value of tokens. The incentive behind this approach is that mixing

tokens will make it more complex to assign the original ownership of the tokens.

Following that logic, this procedure will generate some level of privacy. Existing

cryptocurrencies relying on the mixing approach are Dash and PIVX. However,

PIVX announced it will change to the Zerocoin privacy protocol in the future.

A clear advantage of this method is in its implementation. The method can be

applied on top of various cryptocurrencies while there is no change of the consensus

mechanism required. In addition to that the method is easy to implement into

existing cryptocurrency ecosystems. On the other hand, one disadvantage involves

the permanent accessibility of the mixing entities meaning the mixers have to be

online all the time. Furthermore the rst implementations of this method needed to

rely on trust to the third party doing the mixing and not stealing the coins.

CoinJoin is a procedure to mix transactions and improve the anonymity. In Bitcoin,

a normal transaction consists of inputs and outputs and those transactions can be

seen on the public blockchain. Thus, this concept only provides a low degree of

anonymity(see Figure 19). With the help of the CoinJoin mechanism, it is possible

to mix all inputs and outputs and add them as one transaction to the block. There

are still inputs and outputs available from the sender to the receiver but there is not

really a transaction because in this case all participants made payments.

Figure 19: Process of CoinJoin

The concept of CoinJoin is an advanced form of the mixing approach because steal-

ing the tokens is much harder. Böhme et al. (2017) study the implementation of

CoinJoin into the Bitcoin system. According to them, money laundering can be

pursued on this platform using bitcoins and the authors claim it is the cheapest

form of money laundering. CoinJoin transactions consist of secure transactions for

33
honest users but lack the trust element regarding market hierarchies simultaneously.

The trust element conicts with the anonymous execution of transactions as these

are getting a design target next to security. But recent research by Goldfeder et

al. (2017) states that a users's wallet can be made transparent even though several

rounds of CoinJoin mixing have been performed. The reason is the browser cook-

ies of the users. The mixing procedure can disguise the transaction path between

addresses but it cannot break the link entirely.

Important to notice is that in Bitcoin the CoinJoin procedure is not part of the

protocol. In Dash, the CoinJoin element is a component of the underlying protocol

and it is built on the Bitcoin software. Its main function is to provide instantaneous

transactions with a high degree of anonymity. Furthermore, the digital currency has

a two tier network with so called "Masternodes". Dash implemented "Masternodes"

which are responsible to oer a decentralized governance system. Every node can

become a "Masternode" as soon as the node proves to be the owner of 1000 DASH

coins. The "Masternode" can vote on essential issues and new proposals and if

passed the new issues and proposals are introduced on the blockchain immediately.

A further feature of Dash is the InstantX function. This function assures that

Masternodes will send transactions in real time because they can reach consensus

internally whose inputs is reserved for the respective output. Moreover, the Dash

system is governed de-centrally by the blockchain via "Masternodes" and its voting

system introducing new proposals.

The core feature providing anonymity in Dash is known as the Darksend function.

"Darksend" is based on the CoinJoin concept with additional improvements. This

network does not need a trusted third party because the mixing service will be

provided within the currency system. "Darksend" needs at least three users to start

the mixing process.

A stronger level of anonymity can be assured if the transactions use denominations

of 0.1 DASH, 1 DASH, 10 DASH and 100 DASH. All users should state their inputs

and outputs in the proposed denominations while the collection of transaction fees

should be charged in a separate way. The mixing protocol has a limit of 1000 DASH

tokens per session. The intervals are pre-determined and a user will give out requests

to join other clients with the help of "Masternodes". Before the user's transaction

can be a part of the "Masternode", a queue object is published to the system. This

object will include all necessary details including the user's desired denomination.

Moreover, stronger anonymity can be achieved if multiple sessions are pursued. The

concept of providing multiple sessions can be seen as a chain approach in which

the transactions are transmitted to several "Masternodes" resulting in a relaying

34
network to increase the protection of the user's identity.

"Darksend" improves the privacy element of the Dash ecosystem and the fungibility

of its coins. On the other hand, the mixing protocol needs available users who are

willing to mix for the required denominations otherwise there will be no mix. In

addition to that, the anonymity argument has a limit to the amount of participants

in the mixing process. A typical mixing session needs at least three users with

repeatable processes to increase the anonymity.

3.2 CryptoNote and Ring Signatures

Another opportunity to provide anonymity is through CryptoNote and ring signa-

tures. A ring signature depends on a group of users. Inside this group someone

proved that they signed the transaction but it cannot be stated which user was the

signor. Anyone can verify the signature but it is only a random guess to acknowledge

which user really proved the signature initially. Van Saberhagen (2012) emphasizes

two main properties inspired by Okamoto and Ohta's (1991) description of ideal

electronic cash that the implementation of CryptoNote has to satisfy. It should be

untraceable and unlinkable regarding the transactions. Untraceability means that

for every incoming transaction all senders have an equal probability. Unlinkability

refers to outgoing transactions and the proof of sending it to the same user is not

possible meaning if there are two transactions with receivers A and B, it is not

possible to identify them if A = B.

The process of CryptoNote is also based on ring signatures. A user can make a

transaction and take the outputs of available similar transactions on the blockchain

to create the inputs to a transaction based on ring signatures. With the ring signa-

ture, the input verication of one user can be made non-linkable to the user which

is performing the transaction. This process is done automatically in the protocol

without any extra notications to the users. The prevention of double spends can

be achieved through a traceable ring signature. It does not allow the owner of a

token to give a signature on two ring signatures with an equal public key and no

notications on the blockchain.

Existing cryptocurrencies that base their anonymity argument on CryptoNote or

ring signatures are Aeon, ByteCoin or Monero. ByteCoin was the rst one that in-

troduced the CryptoNote protocol in its implementation in 2012. Since the launch,

the protocol has gotten updated many times, for example a multiple signature trans-

action feature has been added. A further approach and the most famous digital

currency with CryptoNote technology can be investigated with Monero that im-

35
 Figure 20: Process of CryptoNote

plemented the Ring Condential Transactions (RingCT). Aeon was introduced in

2014 as a revised version based on Monero and uses the so called CryptoNight-Lite

algorithm in its protocol.

On the positive side, the technology does not require a mixing entity. An additional

advantage is the well established privacy of the technology with proven records.

Only inappropriate implementations have lead to attacks to unveil the anonymity

within the system. A further advantage is the increase of anonymity with more

time elapsed in the ecosystem. The reason is that the outputs will result in the

new inputs of the next mixing section. A higher transaction size results in a clear

disadvantage of using the CryptoNote technology due to lower transaction scalabil-

ity on the blockchain. There is also an increased danger of deanonymization with

incorrect implementations of the technology. For instance, a digital currency known

as Shadowcash did not implement the technology properly and its blockchain could

be de-anonymized completely. A further risk factor is the fact that those approaches

cannot be integrated directly on current cryptocurrency networks and need a sepa-

rate ecosystem to run. Furthermore, the ring signatures have a limit in their ring

size in most cases. Thus, the anonymity is curbed by the number of users in the

ring.

Monero, implemented in 2014, is a decentralized cryptocurrency and is not based

on the Bitcoin protocol but on CryptoNote protocol. Privacy, fungibility and un-

traceability are additional properties of Monero. Following Noether (2017), the

CryptoNote process is vulnerable in particular to two forms of attacks which can

unveil anonymity. The specic amount on a given transaction can increase the pos-

sibility of de-anonymizing the sender of that transaction because it will be easier to

nd out which user paid that amount at a given time. The second attack involves

the needed transaction properties. A pair ( P ,A) consisting of a public key P and an

amount A is required in ring signatures, while other public keys can have the same

36
amount. Therefore, regarding privacy, Monero relies on ring signatures, RingCT

and stealth addresses.

Those three technologies should ensure that the sending  and receiving part and the

amount within transactions are hidden. The transaction is private by default on the

Monero ecosystem. The RingCT is based on so called Condential transactions"

(CT) and it is a mandatory feature for all transactions since September 2017. The

prevention of double spending is assured via a "Multilayered Linkable Spontaneous

Anonymous Group Signature" (MLSAG). The combination of CT and MLSAG with

ring signatures ensures multiple inputs and outputs, anonymity, and the prevention

of double spending. Untraceable amounts, origins, and destinations are provided for

transactions on the RingCT protocol. Moreover, the coin creation can be pursued on

a PoW mechanism and it is thereby trustless with a veriable security. On the other

side, the ring size of Monero is limited to four to curtail the size of the transaction.

Thus, the anonymity depends on the number of users in the ring and it might be

possible to calculate the probability of the linked transactions. Furthermore, hiding

transactions can have negative eects on the verication of new coin creation and

the knowledge of the existing coin supply. If anyone can get access to the discrete

logarithm of the RingCT protocol that person would be able to create as many coins

as desired. However, as of yet no such attack has been noticed.

3.3 Zero Knowledge Proofs

Goldwasser et al. (1985) are some of the rst to come across zero knowledge proofs.

Originally, these proofs can be seen as a convention of two computer programs. One

constituent is known as the Prover, while the other part is called the Verier.

The "Prover" tries to convince the "Verier" that a statement is true. Furthermore,

there are three properties a zero knowledge proof has to satisfy. The rst property is

completeness meaning that an honest "Prover" will convince a "Verier". Soundness

is the second property and only given a true statement, a "Prover" is able to convince

a "Verier". Finally, zero-knowledge is required. In other words a "Verier" does

not get any additional information than that the statement is true. In this context a

zero knowledge proof is a proof that a user has knowledge about something without

giving out extra information regarding how the knowledge has been acquired. The

cryptocurrency Zerocoin and its token Zcoin is built on the concept of zero knowledge

proofs. As stated before, PIVX will also apply the Zerocoin method in the future.

An obvious advantage of the following approach is that the procedure does not need

any mixing entity. Moreover, a high level of anonymity can be guaranteed. The

37
anonymity does not depend on the number of users or the ring size. Every user can

use the anonymity set of a used denomination and apply it achieving much higher

scalability properties regarding transactions. All that is needed is just one min-

ing and spending process, while the transaction linkage between addresses breaks.

Additionally, the concept of zero knowledge proofs can rely on a well established

research history. Maintaining supply auditability is a further positive property.

On the contrary, the proof sizes are quite large and can increase the verication

period by several seconds aecting the transaction scalability negatively. Besides

the transaction scalibilty, the question is also where to store the proofs. The zero

knowledge proof approach needs a trusted setup to implement initial parameters.

If those parameters are unveiled, attackers can inltrate and leak into the system

performing actions such as creating extra coins. Another drawback has to do with

the deployment of the zero knowledge method. The support of the new mining

-and spending functionality requires adjustments on the existing protocols in the

cryptocurrency ecosystem.

Green et al. (2013) describe Zerocoin as a distributed e-cash system that extends

cryptographic techniques to existing coin protocols, such as Bitcoin's. Zerocoins can

be exchanged one-to-one with bitcoins. Any user is able to buy a Zcoin in exchange

for the correct amount of bitcoins. This is done via a new Zerocoin mining transac-

tion on the blockchain. It is important to note that the link of a Zerocoin mining

transaction and a Zerocoin spending transaction is completely broken through zero

knowledge proofs. When re-acquiring a Zcoin, one will get a totally dierent bitcoin

than the one that was used to purchase the Zcoin initially. It is dicult to determine

at which place the user took out the conversion due to the mixing and creation of

Zcoins with all other users in the system.

Figure 21: Process of Zerocoin by Green et al. (2013)

Zerocoin uses three concepts: digital commitments, a one-way accumulator and zero

knowledge proofs. The structure includes four core randomized algorithms. Those

algorithms can be grouped into Setup, Mint, Spend and Verify.

38
In the following λ indicates an adjustable security parameter and C represents a
λ
set of allowable token values. For the Setup algorithm: Setup (1 ) ⇒ params.

For the input of a security parameter, the output consists of global public param-

eters params and an explanation of the set C. Zerocoin's setup routine is a one-

time strong RSA accumulator by Bernaloh and de Mare (1994) and Camenisch and

Lysyanskaya (2001) with N = p ∗ q, u ∈ QRN (u 6= 1). Its parameters p and q consist

of two large prime numbers and are based on the RSA-2048 parameters. They were

created and destroyed immediately and are at present resistant even to quantum

computing power.

A core algorithm is the Mint algorithm: (params) ⇒ (c, skc). The input
Mint

parameter is params, while the output is a coin c ∈ C and a trapdoor value skc. For

mining one Zerocoin, the user has to spend a base coin. Each Zerocoin is committed

to a serial number S. The commitment can be seen almost as an encryption process

because the newly created coin does not show the serial number, it is secret, and

the coin refers to the chosen amount simultaneously. Furthermore, there will be a

randomness factor which will be kept secret all the time. The serial number and the

randomness factor form a hash value of the new Zerocoin and those tokens will get

some value as soon as the hash value can be found on the blockchain.

The next key algorithm is the Spend algorithm: (params, c, skc, R, C) ⇒

Spend

(π, S). For the spending algorithm the inputs are params, a coin c, the trapdoor skc,
some transaction string R ∈ (0, 1) and an arbitrary amount of coins C . The output

is a spending transaction including a proof π and a serial number S if c ∈ C ⊆ C .

Otherwise the output is nothing. In a spend transaction, the user has to include

the created serial number. The spending transaction's main message is its zero

knowledge proof. The proof includes the following statements: the mentioned and

posted valid Zcoin is on the blockchain and secondly this Zcoin has the serial number

S in accordance with the users' transaction.

Finally, the Verify algorithm: (params, π, S, R, C) ⇒ (0, 1). The input

Verify

properties include the params, the proof π , a serial number S , the transaction
information R, and a setting of coins C . The output will be 1 if C ⊆ C and

(π, S, R) is valid. If it is invalid the output will be 0. The zero knowledge proof is

crucial. Anyone in the system can verify a zero knowledge proof because any user is

convinced that the respective person minted a Zcoin and it included the respective

serial number S. The verier can also check the existing blockchain if that serial

number has been spent before. In addition to that zero knowledge proofs allow that

the veriers have no knowledge about the spending amount of Zcoins. It only nds

one other Zcoin and converts it back to the base coin on the blockchain, it must not

39
be the same as it was created. If only one user would apply this approach, there

would be no anonymity. The anonymity comes from the zero knowledge proofs based

on Zerocoins and pursuing multiple transactions from numerous users resulting in

sucient noise for the system.

Zerocoin provides strong anonymity with no transaction linkages but it needs a

trusted setup, more resources to store on the blockchain, and more computational

resources for the verication process.

3.4 Zero Knowledge Succinct Non-interactive Argument of

Knowledge (Zk snarks)

This form builds also on the knowledge of zero knowledge proof schemes with some

additional features. The attribute succinct refers to the property that the proof can

be veried in a few milliseconds and has a transaction size of only a few hundred

bytes. With the structure of non-interactive components, the proof only has one

message sent from the proving element to the verifying element. To date, the only

way of creating zk snark proofs is achieved through an initial setup that creates

a common reference string. The common reference string, also called the public

parameters in most networks, is the sharing point of the prover and the verier. The

cryptocurrency Zerocash considers the zk snark technology in its implementation.

The idea of Zerocash is to improve the ground work of Zerocoin's protocol. Zerocoin

can only use pre-determined denominations, while the transaction sizes with around

25kb are quite large. In ZCash and its usage of zk snark technology, the proof

sizes are small and allow a fast verication. Another advantage is that it is a very

good anonymity set to break the transactional linkages of addresses. Furthermore,

the transaction amount cannot be shown and there is no requirement to change

the tokens from a base coin to an anonymous one because the coins are exchanged

directly.

On the other hand, a trusted setup has to be used as was the case with Zerocoin.

However, this setup is structured more complexly. An incorrect implementation of

the setup can result in a forgery of coins. A consecutive argument on that point

is the dicult detection of the forged coins in the system because of a non-visible

supply auditability. An additional shortcoming is the generation of private transac-

tions. It can take almost a minute or even more, aecting the transaction scalability

negatively. Finally, the zk snark technology is a relatively new method and lacks a

well established academic research background.

40
Ben-Sasson et al. (2014) present Zerocash to build a separate anonymous currency

next to a base currency. In Zerocash's system any user can exchange a base coin into

a Zerocash token called Zerocoins and make transactions by sending those coins to

other users. The Zerocash algorithm has the following main components: "Setup",

"CreateAddress", "Mint", "Pour", "VerifyTransaction" and "Receive".

Two forms of transactions are used in Zerocash's implementation. The rst trans-

action is called Mint" and gives the user the right to exchange a pre-determined

number of base coins for an equal amount of Zcash and a Zerocash address. The

transaction includes a cryptographic commitment to a new token and determines

its coin value, owner address and a serial number. The SHA-256 hash function is

used for the commitment procedure and can make the coin's value and owner ad-

dress invisible. The nodes in the Zerocash environment remain a Merkle tree for

the available token commitments. Because of the decommited values and being a

limited witness as a member in the tree itself, every participant can show that he or

she is the owner through a coin commitment. However, this ownership proof is not

private meaning there is a second transaction to maintain privacy of the user and

that they know the information in zero knowledge.

Figure 22: Process of Zerocash by Ben-Sasson et al. (2014)

A private transaction taking some coins from the user to create new coins is done

via a "Pour" transaction. The pour transaction is based on the zero knowledge

approach, and for instance, for two input tokens and two output tokens it contains

that the user has ownership of the two input coins. Furthermore, every input coin

was a member of a previous mint transaction or each output coin was part of a

previous pour transaction. Additionally, the total sum of input coins has to be

the same as the total sum of output coins. This type of transaction ensures that

input tokens only show the serial numbers but not any information about the value

of input tokens or output tokens, or ownership addresses. The verication process

for the mint transaction can be pursued by any user. As stated before, the pour

transaction can be veried via zero knowledge proofs, in Zerocash's case it is done

by a zk snark-approach.

Denition: An arithmetic circuit satisability issue of an F-arithmetic circuit C:

41
Fn x Fh → Fl is caught by the relationRC = {(x, a) ∈ Fn x Fh : C(x, a) = 0l }; while

its language is LC = {x ∈ Fn : ∃a ∈ Fh s.t. C(x, a) = 0l }.

The algorithm of the zk-snark technology for a given eld F-arithmetic circuit sat-

isability in Zerocash can be distinguished into KeyGen, Prove and Verify

λ
polynomial timed algorithms. The KeyGen: KeyGen( 1 , C) → (pv, vk). As in-

put components remain a security parameter λ and an F-arithmetic circuit C . The

KeyGen algorithm provides a proving key called pk and a verication key vk . Both
keys can be used innitely for the verication and proof process. Moreover those

keys are dened as public parameters in LC .

The second polynomial algorithm is called Prove: Prove ( pk, x, a) → π . As input

parameters there is a proving key pk and any (x, a) ∈ RC . As output, it will give
out a non-interactive proof π for x ∈ LC . Finally the last component is the Verify

algorithm: Verify (vk, x, π) → b. For the input attributes verication key vk , x and

a proof π , the verier results in b = 1 if the verier got a conviction that x ∈ LC .

Additionally, the zk snark technology should fulll the properties of completeness,

succinctness, proof of knowledge, and perfect zero knowledge.

3.5 Further anonymity schemes

As there is a constant development of new anonymity structures for digital currencies

a few more schemes will be presented that do not t into one of the aforementioned

concepts. NAV Coin, created in 2014, consists of a mixture of the Bitcoin blockchain

and its own sub-chain. The transactions are encrypted making the sending amount

of coins untraceable. The technology of NAV Coin uses splitting transactions gen-

erated on a random number. Additionally, the divided transactions are based on

several blocks with time delays to provide a maximum level of anonymity. Finally,

the transaction parts will be reconnected and separated one more time to create

numerous identical transactions on the system.

The cryptocurrency Verge has its origin in the digital currency named Dodge-

CoinDark. It changed its name to Verge in 2016. Verge builds upon the Bitcoin

blockchain and wants to improve the privacy by adding central anonymity networks

with Tor and i2P technologies. Thereby, the user's transactions are routed through

several global servers. Every server in the ecosystem deletes the information of a

previous server such that the last exit node server does not know from which des-

tination the information originally came. Following this procedure, the users can

continue with their transactions as normal while Verge makes the currency trail

42
untraceable.

Showing the fast development in the cryptocurrency sphere regarding privacy and

anonymity, Zencash can be presented as a dignied example. In May 2017 Zencash

was established by a hard fork from ZClassic, while ZClassic originated from Zero-

cash. Zencash's main feature regarding anonymity and privacy is built on the zk

snark technology as in Zerocash. However, the developers are also considering a few

properties from the cryptocurrency Dash such as a decentralized governance model

and an integrated voting system. Similar to the "Masternodes" in Dash, Zencash

will implement "Secure Nodes" which are responsible to run the system and will get

a reward of 3-5% from the mining process.

3.6 ChoiceCoin's Approach to Anonymity

As aforementioned the anonymity is built mainly on the infrastructure and the dis-

tribution of trust to the underlying network. But anonymousness depends also on

the user's behavior and motivation. The dierent examples of anonymity schemes

show that its way of implementation has numerous variants and each structure

presents advantages and drawbacks depending on the environment and on its main

goal. Referring to ChoiceCoin and its modeled structure, several forms of anonymity

schemes are thinkable. Mixing the funds is one possibility, whereby the transaction

linkage cannot be broken entirely. The ring signature approach is also an oppor-

tunity, however, only the dierent pre-dened nodes could be part of the process

limiting the anonymity argument. The zk snark method is also feasible, but the

technology is quite new and lacks a well established research history, meaning it

could have undiscovered aws which could de-anonymize the environment. Thus,

the anonymity scheme with the zero knowledge approach as used in Zerocoin will

be implemented.

This method can be implemented on the existing protocol, the security parameters

can be created by the terminal and the central entity can pursue the mining trans-

action. Furthermore, the zero knowledge technology has a proven research record

while generating a high level of anonymity with unlinkable transactions. Addition-

ally, even if pre-dened nodes want to send anonymous transactions to each other

it is possible with the untraceable properties.

The process of an anonymous transaction works as follows: at rst, a user will

approach a pre-dened node and will allow all transaction details to be known

including the anonymity option. Next, the node gives the transaction list including

the desired anonymous transactions to the terminal overview. On the third point,

43
 Figure 23: Process of optional Anonymity

the central entity prepares all transactions and performs a mint transaction for the

anonymous option to transform it to a Zerocoin token. After the mining transaction,

the terminal overview transmits the anonymous transaction back to the current node

leader. However, the mining process might take longer than the normal process of

checking transactions by the central entity and the anonymous transaction might

be part of the transaction list after several rounds. In the next step, the node

leader veries the transaction and relays it to the other nodes including all the other

transactions for the following transaction round. After the verication procedure,

the anonymous transaction will be appended on the blockchain and stays there.

Finally, in the spending process of the anonymous transaction, one ChoiceCoin token

can be redeemed on the blockchain with a serial number made public, performing a

zero knowledge proof, and a random Zerocoin.

A user can make an anonymous transaction with Zerocoins any time, for instance

a few moments before an actual transaction to obfuscate the transaction link even

more. For the spending transaction all that is needed is the serial number and all

members of the network. The recipient of the base coin only knows it can be from

one of the many participants who commissioned a Zerocoin mint process without

unveiling the true identity. Furthermore, there will be a xed denomination rate of

0.25 Zerocoins in a one-to-one exchange with the existing coins in the ecosystem.

For instance, a transaction of ve Choicecoins can be made anonymous, then the

user will give a normal transaction to one of the pre-dened nodes. After that,

the terminal splits up the transactions in this case to 20 individual transactions

with 0.25 Zerocoins each. The user can activate the spend process as soon as the

transactions are on the blockchain.

To be more precise in a mint process, a hash value H will be computed by the

central entity. The hash value H(S, r) consists of a serial number S and a random

secret r. The serial number will be made public in the spending process later on.

44
However, the random secret r will never be made public to ensure unlinkability.

After the mint transaction and the verication process of the pre-selected nodes,

the hash value will stay on the blockchain. On the blockchain, only the hash value

H will be visible. The input values with the serial number S and the random secret

r including the recipient address are invisible.

The user who wants to execute an anonymous transaction can wait as long as desired

because its Zerocoin is on the blockchain with the respective hash value H and can

contact the pre-dened nodes or the terminal to spend it. To redeem a transaction

into ChoiceCoins two components are required: First, the serial number S has to

be included which has been generated and not been spent before. The second

requirement is to create the zero knowledge proof which includes that the users

know a number r such thatH(S, r) is one of the Zerocoins on the blockchain. The
randomness r combined with S picks an arbitrary Zerocoin on the blockchain and

uses it as the input in a new ChoiceCoin transaction.

The mint -and spend transactions break the links of the base coin completely and

the process can be seen as a huge laundry network. Selecting an arbitrary Zerocoin

in the spending procedure determines the anonymity in the system as soon as more

than one user relies on the anonymous option regarding transactions. Since r is

a secret no one knows which Zerocoin corresponds to the serial numbers. Even

after serial numbers have been revealed and it was inside the hash value H, nobody

knows which hash value it really was. Moreover, the presented approach should be in

accordance with anonymity and privacy mechanisms that have to present a solution

to both issues between the protection of privacy and the maintenance of public

veriability. The reason is the public veriability is reached through a permissioned

blockchain and the privacy is not mandatory but it is a possible option for each

participant in the system. The users can weigh the timing and anonymity argument

individually and it allows them to make a choice while providing a exible approach

for the presented ecosystem of ChoiceCoin. It is thinkable to change to a complete

anonymous transaction scheme as soon as the transaction scalability can reach the

same level as without the anonymous option through a voting decision by the pre-

dened nodes.

4 Conclusion

One of the main goals of this thesis was to build a cryptocurrency from the ground

up with specic properties. Three key components are required for the creation:

a blockchain, a currency and a protocol. Within those boundaries the design of a

45
cryptocurrency can have various spectrums. The core properties as presented in

ChoiceCoin are built on a private blockchain with 24 pre-dened nodes to generate

fast transactions, a consensus mechanism based in principle on the Ripple algorithm

and an optional anonymous transaction possibility created within the protocol. Fur-

thermore, the terminal overview providing a exible adaptation to change the set

of rules via a voting system by the nodes, and the node leader component congure

ChoiceCoin with unique details that other cryptocurrencies do not oer directly.

A centralized approach as seen in ChoiceCoin has a key vulnerability: if a central

authority, in this case the terminal overview, is not available, the transaction pro-

cess cannot be executed. That is a risk factor that all participating users have to

reconsider. Furthermore, one aspect remains whether it makes sense to introduce a

blockchain for ChoiceCoin or if a distributed databank is more usable. For Bitcoin,

a blockchain is a reasonable design for a group of un-trusted nodes, no central au-

thority and almost no possibility to shut it down by one or several attacks on its

network. The cost of maintenance and support, the limited performance and the

fault tolerance factor aecting the robustness of the ecosystem are clear properties

that inuence the decision making of choosing a blockchain. Blockchains are supe-

rior in the provision of a robust and fault tolerant system, while its shortcomings can

be found in a limited performance compared to a databank. For ChoiceCoin, it is le-

gitimate to think of introducing a SQL databank. However, as the terminal overview

already presents a centralized property and to provide a maximum degree of decen-

tralized approach for the underlying network, a blockchain is more reasonable than

relying on a databank. One reason is that the 24 pre-dened nodes should verify

the transactions and update its blockhain themselves without any dependency on a

central authority. Furthermore, if one node cannot be reached, there are sucient

nodes available and up-to-date to execute the verication process. The robustness

aspect is favored over the performance component as the blockchain's development

and research is still at an early stage and it is highly probable that improvements

regarding performance are possible in the future.

The analysis of anonymous schemes emphasizes that from an economic perspective

the network is responsible for providing anonymity and privacy to the participants.

The users have to rely on the system and its operating entities. But participants

can enhance the anonymity aspect indirectly by encouraging numerous transactions

resulting in noise for the system. In the cryptocurrency sphere, each anonymity and

privacy structure has its individual set of benets and drawbacks including respective

trade-os. ChoiceCoin includes an anonymous option for users' transactions based

on the zero knowledge proof method pursued by Zerocoin. The reason to provide

an optional approach only is the transaction scalability which takes longer if a user

46
wants to make an anonymous transaction. Furthermore, anonymousness depends

also on the user's preferences and oering an option between choosing anonymity

and fast transactions will result in a more exible usage.

A faster anonymous transaction is also possible, for instance with the zk snark pro-

cedure. But the technology is not well established, yet and as anonymity can only

be provided by the implemented network, the reliability aspect cannot be guaran-

teed as a core property for the underlying ecosystem. Thus, pure zero knowledge

proofs have a detailed research history and a proven record as a resilient and stable

technology among others for providing anonymity and privacy.

47
A Appendices

A.1 Glossary

51% attack: A mining pool is able to control 51% of the mining power (hash rate)

meaning it can include its own blocks into the blockhain. Additionally, it can pursue

a fork to make an independent branch that can merge with the main blockchain part.

ASIC: It is called Application Specic Integrated Circuit (ASIC) and it is a chip

that is custom-designed for just one application for instance Bitcoin or Ethereum

mining.

Asynchromous distributed system: In those systems messages can be delayed

for an unlimited amount of time. The period of message transmission is unknown

between nodes. Furthermore, there are no strong assumptions on time and ordered

events.

Byzantine generals' problem: The main idea of this problem is that there are

several generals and every general has the command of its Byzantine army to attack

an enemy city. However, to execute a successful attack, all generals need to reach

agreement on a common battle plan. The generals can only communicate with

messengers. It is possible that those messengers might get captured by the enemy

and the original message does not arrive the others. The main diculty of that

problem is that some generals could be traitors and are interested in sabotaging the

battle plan. Within this environment some generals might send inaccurate messages

but the loyal generals should be unaected tolerating a small fraction of traitors and

not adopting the manipulated battle plan.

Condential Transaction (CT): Originally, CT is a cryptographic tool that

should improve the privacy and security of Bitcoin. The transferred amounts will

be visible only to the users in the transaction and to those that verify it. The

cryptographic technology used is called additive homomorphic commitments. A by

product is the possibility to transfer private invoice numbers or refund addresses

data without an increase of transaction size. The idea according to Back (2013) is

to use a Pedersen Commitment.

GHOST: Greedy Heaviest Observed Sub Tree. It is a revised version of Bitcoin's

PoW and used in Ethereum. The GHOST-protocol was introduced in 2013 to com-

bat the way that quick block time in blockchains suer from a big amount of or-

phaned blocks. This protocol includes orphaned blocks, called uncle blocks, and its

rewards are 87.5% for normal valid blocks, while the nephew, the child of the unlce

i
blocks, gets 12.5% of the rewarded block.

Hash rate: The total amount of existing gear in the network to solve puzzles/ tasks

in a PoW mechanism. In Bitcoin, the hash level changes over time as miners can

join or leave the network. A higher hash rate results in a better network protecting

against attacks.

Invisible Internet Project(I2P): It is used in network systems. I2P provides

increased anonymity with an overlay network. The i2P network is a network within a

network. The main goal is to protect the communication within the system from drag

net surveillance by additional parties such as ISPs. The users' trac is encrypted

and runs through several thousand computers worldwide. The nearly unlimited

possibility of trac paths makes it harder to follow for monitoring and surveillance

processes by third parties.

Latency: It is a way of delay publication in the computing -and network system.

Latency announces the delay time it takes from an input to a desired outcome within

a system. This term aects the usage especially of communication of network system.

Log replication: A log is a strictly ordered append-only sequence of numerous

operations. In computing, a replication handles storing information to enable con-

sistency between dierent resources to improve reliability, fault tolerance or acces-

sibility.

Memory hardness: It is the ability of a computer to move data around in mem-

ory and not with calculations. Furthermore, it is a property that general purpose

computer hardware is already designed to perform well but it cannot really lead to

ecient results on ASICs. With a resistant algorithm to ASIC, it can prevent large

powerful rms from out taking control of mining power in the Ethereum system.

Multilayered Linkable Spontaneous Anonymous Group Signature (ML-

SAG): The MLSAG following Noether (2017) is a generalized form of Back's Link-
able Spontaneous Anonymous Group Signature (LSAG). This LSAG is not covered

by keys but by key vectors.

Denition: A key vector is just a collection ȳ = (y1 , ..., yr ) of public keys with

respective private keys x̄ = (x1 , ..., xr ).

In a MLSAG approach, it is supposed that each signer of a generalized ring of n

j i=1,...n
members has exactly m keys:"{Pi }j=1,...m .

The rst intention behind the MLSAG is to provide a proof that one of the n signers
knows the secret keys to the complete key vector. In addition to that another

component sketches the MLSAG signature. If a signer uses one of the m signing

ii
keys in a dierent MLSAG signature this means both rings are linked and the second

MLSAG signature will be discarded. The MLSAG is introduced by an algorithm

and satises three key properties: Unforgeability, Linkability, and Signer Ambiguity.

Network File System (NFS): It consists of a distributed le system and gives

users access to a client computer to demand les over a computer system. The NFS

has an open standard form and anyone can implement the protocol.

Nonce: It is a crypto graphical term and an arbitrary number that is in principle

only used once.

Peer to peer (P2P): Originally, it is a computer that participates in the network

peers and does not have any privileges but equal treatment.

Protocol: It describes how communication should work and provides several ways

of implementation by using dierent programming languages and so on.

RSA-accumulator: An accumulator is used in cryptography and it is a one way

membership application. It will respond to whether a participant is a member of

a set while not publishing the individual members of the set entirely. The RSA

argument is a public-key cryptosystem by Rivest, Shamir and Adleman (RSA) for

data transactions. In this system a user builds and shows a public key based on

two large prime numbers while those prime numbers must be kept secret to mitigate

fraudulent use of the system.

secp256k1: It is a cryptographic component and part to the parameters of the

Elliptic Curve Digital Signature Algorithm (ECDSA) curve. The secp256k1 is used

among others in Bitcoin and its denition can be found in Standards for Ecient

Cryptography (SEC).

Segregated Witness: It is an upgrade to the Bitcoin protocol. Its technology

separates signature data from bitcoin transactions. It is a soft fork and should make

Bitcoins' protocol rules more restrictive.

Segwit 2x: It is a software upgrade and improves the capacity of the Bitcoin

protocol. The deployment of Segwit2x is signaled by the majority of Bitcoin miners.

The expected date of the hard fork was November 16th and it should be done when

Bitcoin block 497,784 was mined. However, the date was postponed and the hard

fork took place on December 28, 2017 on the block number 501,451.

Selsh mining: In this form of attack, the attacker gains a lot of mining power

at the cost of his/her short term revenue with the maintenance of an own private

blockchain to the existing blockchain. He/She publishes a lot of blocks at once and

forces the rest of the network to give away their blocks and revenue. This is an

iii
incentive to honest miners to follow the attacker and to increase its revenue that

nally can get the 51% of the network mining power. It is another way of a 51%

attack.

Serenity: It is a name used in Ethereum ecosystem for the last phase to switch

from the PoW mechanism to the PoS mechanism called Casper. The Serenity pahse

will be implemented via a hard fork from the ocial Ethereum blockchain.

Software Development Kit (SDK): It is a collection of Software and used for

development issues for a device or an operating system. An SDK consists of an

integrated development environment for a central programming hub and interface.

In most cases, SDKs have a sample code that supports developers with example

programs and libraries. The developers can create basic programs via SDKs and it

eventually helps them to build more complex applications. Furthermore, SDKs can

also include technical documentation or sample graphics.

State: All or a part of the data that a program deals with.

Throughput: It is a way of measurement in the computing -and network system.

Throughput tells the amount of units of information a system can process in a xed

period of time. The response time between single users to request and receive the re-

sponse, and the speed of workload time are related measurements to the throughput

component.

The Onion Router(Tor): It is a free software and enables anonymous communi-

cation. Tor is an overlay network system and provides internet trac with the help

of several thousand relays. The relays are based globally and try to help making the

location and usage of the participants' trac and network surveillance invisible.

Virtual machine: It is a device that actually executes a program or code.

A.2 Source Codes

The following Python code has been executed on Python version 3.4

ˆ Python code for the created blockchain:

import hashlib , json , sys

def newHash ( msg=" " ) :

# This function helps to f a c i l i t a t e the encoding of the hashing_

algorithm with an " utf −8" approach

iv
i f type ( msg ) != str :
msg = j s o n . dumps ( msg , s o r t _ k e y s=T r u e ) # the keys w i l l be sorted_
to guarantee repeatability

i f s y s . v e r s i o n _ i n f o . m a j o r == 2 :
return ( h a s h l i b . s h a 2 5 6 ( msg ) . h e x d i g e s t ( ) , ' u t f −8 ' )
else :
return h a s h l i b . s h a 2 5 6 ( str ( msg ) . e n c o d e ( ' u t f −8 ' ) ) . h e x d i g e s t ( )

import random

import time

random . s e e d ( 0 )

def m a k e T r a n s a c t i o n ( maxValue = 5 ) :

# It w i l l create valid random transactions in the range of 1 to with_

a timestamp

sign = int ( random . g e t r a n d b i t s ( 1 ) ) * 2 − 1 # This w i l l randomly choose_

−1 or 1

amount = random . r a n d i n t ( 1 , maxValue )

aPays = sign * amount

bPays = −1 * aPays

# timestamp based on milliseconds

timestamp1 = time . time ( ) * 1000

print ( t i m e s t a m p 1 )
time . s l e e p ( 0 . 1 )

amount = random . r a n d i n t ( 1 , maxValue )

cPays = sign * amount

dPays = −1 * cPays

# timestamp based on milliseconds

timestamp2 = time . time ( ) * 1000

print ( t i m e s t a m p 2 )
# It w i l l always return transactions that respect the conservation_
of tokens .

# However , since now no checks pursued whether these overdraft_

an account

return { u 'A ' : a P a y s , u 'B ' : bPays , u 'C ' : c P a y s , u 'D ' : d P a y s }

txnRawList = [ makeTransaction ( ) for i in range ( 2 0 ) ]

# The txnRawList w i l l create 20 transactions

v
# Send the txnList to the central authority
# The central authority w i l l check the defined set of rules and_
rank the transactions in an order regarding the timestamps

def u p d a t e S t a t e ( txn , state ):

# Inputs : transfer amount ( txn ) and account balance ( state )

# Returns : Updated state but no validation of transaction only_
update the state

# I f the transaction == valid −> update the state

state = # Creates a working copy of the data .
s t a t e . copy ( )

for k e y in t x n :
i f k e y in s t a t e . k e y s ( ) :
s t a t e [ key ] += txn [ key ]

else :
s t a t e [ key ] = txn [ key ]

return state

def i s V a l i d T x n ( txn , state ):

# Assumption : Transaction is a dictionary keyed by account names

# Checking sum of the deposits and withdrawals is 0
i f sum ( t x n . v a l u e s ( ) ) i s not 0:

return F a l s e

# Checking overdraft of transactions

for k e y in t x n . k e y s ( ) :
i f k e y in s t a t e . k e y s ( ) :
acctBalance = s t a t e [ key ]

else :
acctBalance = 0

i f ( acctBalance + txn [ key ] ) < 0:

return F a l s e

return True

# Determine the node leader

vi
random . s e e d ( 0 )

master_node = random . r a n d i n t ( 1 , 2 4 ) ;

print ( " The leader node will be the following node : " + str ( m a s t e r _ n o d e ) )
# The node leader w i l l relay the transaction l i s t to a l l other_
pre defined nodes

# Consensus algorithm
#The consensus mechanism w i l l be explained in the section_
" the consensus algorithm "

# Block creation
# Generating the s t a t e s and the genesis block
state = { u 'A ' : 6 0 , u 'B ' : 6 0 , u 'C ' : 6 0 , u 'D ' : 6 0 }

# Definition of i n i t i a l s t a t e s
genesisBlockTxns = [ state ]

genesisBlockContents = { u ' b l o c k N u m b e r ' : 0 , u ' p r e v i o u s H a s h ' : None , _

u ' txnCount ' : 1 , u ' t x n s ' : g e n e s i s B l o c k T x n s , ' timestamp ' : time . time ( ) * 1000}

genesisHash = newHash ( genesisBlockContents )

genesisBlock = {u ' hash ' : g e n e s i s H a s h , u ' c o n t e n t s ' : g e n e s i s B l o c k C o n t e n t s }

genesisBlockStr = j s o n . dumps ( g e n e s i s B l o c k , s o r t _ k e y s=T r u e )

chain = [ genesisBlock ]

print ( c h a i n )

def makeBlock ( t x n s , chain ) :

time . s l e e p ( 0 . 1 )

timeStamp = time . time ( ) * 1000

time . s l e e p ( 0 . 1 )

parentBlock = chain [ −1]

previousHash = p a r e n t B l o c k [ u ' hash ' ]

blockNumber = p a r e n t B l o c k [ u ' c o n t e n t s ' ] [ u ' blockNumber ' ] + 1

txnCount = len ( t x n s )
blockContents = {u ' timestamp ' : timeStamp , u ' blockNumber ' : blockNumber ,_

u ' previousHash ' : previousHash ,

u ' txnCount ' : len ( t x n s ) , ' txns ' : txns }

blockHash = newHash ( b l o c k C o n t e n t s )

block = {u ' hash ' : blockHash , u ' contents ' : blockContents }

vii
return block

blockSizeLimit = 4 # Randomly chosen number of transactions per block

while len ( t x n R a w L i s t ) > 0 :
b u f f e r S t a r t S i z e = len ( t x n R a w L i s t )

## Bring together a set of valid transactions for gathering

txnList = []

while ( len ( t x n R a w L i s t ) > 0) & ( len ( t x n L i s t ) < blockSizeLimit ) :

newTxn = t x n R a w L i s t . pop ( )

validTxn = i s V a l i d T x n ( newTxn , state ) # I f txn is invalid −>False

if validTxn :

t x n L i s t . a p p e n d ( newTxn )

state = u p d a t e S t a t e ( newTxn , state )

else :
print ( " T r a n s a c t i o n is invalid ")

sys . stdout . f lu s h ()

continue # I f invalid transaction −> ignore i t and continue

## Create f i n a l block
theAddedBlock = makeBlock ( t x n L i s t , chain )

c h a i n . append ( theAddedBlock )

chain [ 0 ]

chain [ 1 ]

print ( c h a i n [ 1 ] )

state

print ( s t a t e )

def checkingBlockHash ( block ) :

# Exception raised i f the hash is no match to the block contents

expectedHash = newHash ( b l o c k [ ' c o n t e n t s ' ] )

i f b l o c k [ ' hash ' ] ! = expectedHash :

raise E x c e p t i o n ( ' Hash d o e s n o t match contents of block %s '%

b l o c k [ ' c o n t e n t s ' ] [ ' blockNumber ' ] )

return

viii
def ch ecki ngBlo ckVa lidi ty ( block , parent , state ):

# Checking following conditions :

# Is each transaction a valid update to the system state ?
# Is Block hash valid for the block contents ?
# Does block number go up by one compared to the parent block number?
# Is the parent block ' s hash referenced properly ?
parentNumber = p a r e n t [ ' c o n t e n t s ' ] [ ' blockNumber ' ]

previousHash = p a r e n t [ ' hash ' ]

blockNumber = b l o c k [ ' c o n t e n t s ' ] [ ' blockNumber ' ]

# Checking transaction v a l i d i t y ; i f an invalid transaction −> error .

for t x n in b l o c k [ ' c o n t e n t s ' ] [ ' txns ' ] :

i f i s V a l i d T x n ( txn , s t a t e ) :
state = u p d a t e S t a t e ( txn , state )

else :
raise Exception ( ' I n v a l i d transaction in block %s : %s ' %_

( blockNumber , txn ) )

checkingBlockHash ( block ) # Checks hashes −> error i f not accurate

i f b l o c k N u m b e r != ( p a r e n t N u m b e r + 1 ) :
raise E x c e p t i o n ( ' Hash d o e s n o t match c o n t e n t s of block %s ' %_

blockNumber )

i f b l o c k [ ' c o n t e n t s ' ] [ ' p r e v i o u s H a s h ' ] != p r e v i o u s H a s h :

raise E x c e p t i o n ( ' P r e v i o u s h a s h i n a c c u r a t e a t b l o c k %s '_
% blockNumber )

return state

def checkingChain ( chain ) :

# Checks the chain from the genesis block

# Checks i f a l l transactions are valid , no overdrafts found , and_
accurate linked hashes of blocks

# This returns the state as a dictionary of accounts and balances_

or False if error

ix
## Data input processing : Verifying that the chain is a list_
of dictionaries

i f type ( c h a i n ) == str :
try :
chain = json . loads ( chain )

assert ( type ( c h a i n ) == list )

except : # A catch a l l exception
return F a l s e
e l i f type ( c h a i n ) != l i s t :
return F a l s e

state = {}

## Preparing a l l by checking the genesis block

# Checking the following conditions :
# Is each of the transactions a valid update to the system state ?
# Is the block hash valid for the block contents ?
for txn in chain [ 0 ] [ ' contents ' ] [ ' txns ' ] :

state = u p d a t e S t a t e ( txn , state )

checkingBlockHash ( chain [ 0 ] )

parent = chain [ 0 ]

## Checking subsequent blocks : Additionally checks are needed for :

# The reference to the previous block ' s hash
# The v a l i d i t y of the block number
for block in chain [ 1 : ] :

state = chec kingB lock Vali dity ( block , parent , state )

parent = block

return state

checkingChain ( chain )

chainAsText = j s o n . dumps ( c h a i n , s o r t _ k e y s=T r u e )

checkingChain ( chainAsText )

#conclusion of a l l together
import copy

nodeBlockchain = copy . copy ( c h a i n )

x
nodeBlockTxns = [ makeTransaction ( ) for i in range ( 4 ) ]
newAddedBlock = makeBlock ( nodeBlockTxns , n o d e B l o c k c h a i n )

print ( "Amount o f b l o c k s on the blockchain on Node 1 is c u r r e n t l y :_

"+ str ( len ( c h a i n ) ) )

print ( "Amount o f b l o c k s on the blockchain on Node 2 is c u r r e n t l y :_

"+ str ( len ( c h a i n ) ) )

try :
print ( "A new block is in the system and in the verification process ! ")

state = c h e c k i n g B l o c k V a l i d i t y ( newAddedBlock , c h a i n [ −1] , s t a t e )

# Updating the current state
c h a i n . a p p e n d ( newAddedBlock )

except :
print ( "No valid block ! Waiting for the next block . . . " )

print ( " The Blockchain has been updated s u c c e s f u l l y !_

On Node 1 it is %s blocks l o n g "% len ( c h a i n ) )

print ( " The Blockchain has been updated s u c c e s f u l l y !_

On Node 2 it is %s blocks l o n g "% len ( c h a i n ) )

ˆ Python code for cryptography property inspired by the bitcoin package:

The output will be a randomely created private key in hex and decimal form and a

subsequent public key in hex form and compressed hex form.

import bitcoin

# make a random private key

private_key = False

while not private_key :

key = b i t c o i n . random_key ( )

decoded_key = b i t c o i n . d e c o d e _ p r i v k e y ( key , ' hex ' )

private_key = 0 < decoded_key < b i t c o i n .N

print ( " P r i v a t e key in hex form is : " , key )

print ( " P r i v a t e key in decimal form is " , decoded_key )

# mulitply the EC generator point G with the private key_

xi
to get a public key point

public_key = b i t c o i n . f a s t _ m u l t i p l y ( b i t c o i n . G, decoded_key )

print ( " C o o r d i n a t e s (X , Y) of public key is :" , public_key )

#encode as a hex and prefix 04

hex_encoded_public_key = b i t c o i n . encode_pubkey ( p u b l i c _ k e y , ' hex ' )

print ( " P u b l i c key in hex form is :" , hex_encoded_public_key )

#compress public key , adjust te prefix depending on whether y_

is even or odd

( public_key_x , public_key_y ) = public_key

if ( public_key_y % 2) ==0:

compressing_prefix = ' 02 '

else :
compressing_prefix = ' 03 '

hex_compressing_public_key = compressing_prefix +_

b i t c o i n . e n c o d e ( public_key_x , 16)

print ( " The compressed public key in hex form i s :_

" , hex_compressing_public_key )

ˆ The following R source code has bee used for the representation of
gures 17 and 18:

# subplots implemented ( optional )

par ( mfrow=c ( 1 , 4 ) )
#representation of i n t i a l network setup
g <− make_ f u l l _g r a p h ( 6 ) %du% make_ f u l l _g r a p h ( 6 ) %du%
make_ f u l l _g r a p h ( 6 )_
%du% make_ f u l l _g r a p h ( 6 )

g <− add_e d g e s ( g , c ( 1 , 7 , 2 ,8 , 3 ,9 , 1 0 , 1 3 , 1 1 , 1 4 , 1 2 , 1 5 ,_

13 ,19 , 14 ,20 , 15 ,21 ))

com <− c l u s t e r_s p i n g l a s s ( g , s p i n s =6)

V( g ) $ c o l o r <− com $ m e m b e r s h i p

g <− set_g r a p h_attr ( g , " l a y o u t " , layout_w i t h_kk ( g ) )

plot ( g , v e r t e x . l a b e l . d i s t = 1 . 5 , v e r t e x . l a b e l . c o l o r =" b l a c k " )

#node leader gets transaction l i s t
xii
g <− make_ f u l l _g r a p h ( 6 ) %du%_
make_ f u l l _g r a p h ( 6 ) %du%
make_ f u l l _g r a p h ( 6 ) % du% make_ f u l l _g r a p h ( 6 )
g <− add_e d g e s ( g , c ( 1 , 7 , 2 ,8 , 3 ,9 , 10 ,13 , 11 ,14 , _
12 ,15 ,

13 ,19 , 14 ,20 , 15 ,21 ))

com <− c l u s t e r_s p i n g l a s s ( g , s p i n s =6)

V( g ) $ c o l o r <− " b l a c k "

V( g ) $ c o l o r [ 2 ] <− " w h i t e "

g <− set_g r a p h_attr ( g , " l a y o u t " , layout_w i t h_kk ( g ) )

plot ( g , v e r t e x . l a b e l . c o l o r =" b l a c k " , x l a b =" 1 " )

# 50% threshold
V( g ) $ c o l o r [ 1 ] <− " w h i t e "
V( g ) $ c o l o r [ 3 ] <− " w h i t e "

V( g ) $ c o l o r [ 4 ] <− " w h i t e "

V( g ) $ c o l o r [ 7 ] <− " w h i t e "

V( g ) $ c o l o r [ 1 0 ] <− " w h i t e "

V( g ) $ c o l o r [ 1 1 ] <− " w h i t e "

V( g ) $ c o l o r [ 1 3 ] <− " w h i t e "

V( g ) $ c o l o r [ 1 4 ] <− " w h i t e "

V( g ) $ c o l o r [ 1 5 ] <− " w h i t e "

V( g ) $ c o l o r [ 1 9 ] <− " w h i t e "

V( g ) $ c o l o r [ 2 0 ] <− " w h i t e "

V( g ) $ c o l o r [ 2 1 ] <− " w h i t e "

plot ( g , v e r t e x . l a b e l . c o l o r =" b l a c k " , xlab =" 2 " )

# 60% threshold
V( g ) $ c o l o r [ 5 ] <− " w h i t e "
V( g ) $ c o l o r [ 8 ] <− " w h i t e "

V( g ) $ c o l o r [ 9 ] <− " w h i t e "

V( g ) $ c o l o r [ 1 6 ] <− " w h i t e "

V( g ) $ c o l o r [ 2 2 ] <− " w h i t e "

plot ( g , v e r t e x . l a b e l . c o l o r =" b l a c k " , xlab =" 3 " )

# 80% threshold
V( g ) $ c o l o r [ 6 ] <− " w h i t e "
V( g ) $ c o l o r [ 1 2 ] <− " w h i t e "

V( g ) $ c o l o r [ 1 7 ] <− " w h i t e "

V( g ) $ c o l o r [ 2 3 ] <− " w h i t e "

plot ( g , v e r t e x . l a b e l . c o l o r =" b l a c k " , xlab =" 4 " )

xiii
References

Acquisti, A., Dingledine, R., Syverson, P. (2003), 'On the Economics of Anonymity',

Volume 2742 of the series Lecture Notes in Computer Science pp. 84-102.

Alwen, J., Fuchshuber, G., Gazi, P., Park, S., Pietrzak, K. (2015), `Spacecoin: A

Cryptocurrency Based on Proofs of Space `.

Androulaki, E., Capkun, S., Karame, G. O., Roeschlin, M., Scherer, T. (2013),`Eval-

uating User Privacy in Bitcoin` Financial Cryptography and Data Security pp. 34-

51.

Antonopoulos, M. A. (2014), `Mastering Bitcoin. Programming The Open Blockchain',

O'Reilly.

Antonopoulos, M. A. (2014), `Bitcoin security model: trust by computation',

http://radar.oreilly.com/2014/02/bitcoin-security-model-trust-by-computation.html,

20.02.2014, online accessed: 11.01.2018.

Armknecht, F., Karame, G. O., Mandal, A., Youssef, F., Zenner, E. (2015), `Ripple:

Overview and Outlook`.

Back, A., Möller, U., Stiglic, A. (2001), `Trac Analysis Attacks and Trade-Os in

Anonymity Providing Systems', Proceedings of Information Hiding Workshop (IH

2001), April 2001, pp. 245-257.

Baliga, B. (2017), `Understanding Blockchain Consensus Models. Whitepaper', Per-

sistent Systems Ltd.

Barber, S., Boyen, X., Shi, E., Uzun, E. (2012), `Bitter to Better- How to Make

Bitcoin a Better Currency', Financial Cryptography and Data Security pp. 399-

414.

Baran, P., (1962), `On Distributed Communication Networks', The RAND Corpo-

ration, Santa Monica, California.

Bech, M., Garatt, R., (2017), `Central bank cryptocurrencies', BIS Quartely Review,

September 2017.

Benaloh, J., de Mare, M. (1994), One-way accumulators: a decentralized alternative

to digital signatures, EUROCRYPT '93, vol. 765 of LNCS, 1994, pp. 274285.

Ben-Sasson, E., Chiesa, A., Garman, C., Green, M., Miers, I., Tromer, E., Virza,

M. (2014),' Zerocash: Decentralized Anonymous Payments from Bitcoin', IEEE

Computer Society, pp. 459-474.

xiv
Ben-Sasson, E., Chiesa, A., Tromer, E., Virza, M . (2015), `Succinct Non-Interactive

Zero Knowledge for a von Neumann Architecture'.

Berentsen, A., Schär, F. (2017), `Bitcoin, Blockchain und Kryptoassets', BoD -

Books on Demand, Norderstedt.

Bernanke, B. (2013), `Bitcoin may hold long term promise', qz.com/148399/ben-

bernanke-bitcoin-may-hold-long-term-promise/, online accessed: 11.01.2018.

BitcoinBlog (2017), `Dash: Der neue Überieger-Altcoin, der Bitcoins Probleme

löst?', https://bitcoinblog.de/2017/03/02/dash-der-neue-ueberieger-altcoin -der -

bitcoins-probleme-loest/,02.03.2017, online accessed: 11.01.2018.

Blockgeeks (2016), `What is Cryptocurrency: Everything You Need To Know [Ul-

timate Guide]',https://blockgeeks.com/guides/what-is-cryptocurrency/, online ac-

cessed: 11.01.2018.

Blog.Ethereum (2014), `The Issuance Model in Ethereum', https://blog.ethereum.org

/2014/04/10/the-issuance-model-in-ethereum/,10.04.2014, online accessed: 11.01.2018.

Bonneau, J., Clark, J., Felten, E. W., Kroll, J.A., Miller, A., Narayanan A. (2014),

`Mixcoin: Anonymity for Bitcoin with Accountable Mixes',

Brennan, C., Lunn, W. (2016), `Blockchain, Connection Series-The trust disrupter'.

Brown, A., Godsi, P., Kewell, B., Maull, R., Mulligan, C. (2017), `Distributed

ledger technology: Applications and implications*', 2017 John Wiley & Sons, Ltd

Strategic Change. 2017;26(5) pp. 481489.

Brown, R. G., Carlyle, J., Grigg, I., Hearn, M., (2016), `Corda: An introduction'.

Böhme, R., Möser, M. (2017), `The price of anonymity: empirical evidence from a

market for Bitcoin anonymization`, Journal of Cybersecurity, 2017, pp. 19.

Cachin, C. (2001), `Distributing trust on the internet`, IBM Research Zurich Re-

search Laboratory.

Cachin, C., Liu, S., Viotti, P., Vukolic, M., (2016), `XFT: Practical Fault Toler-

ance beyond Crashes', 12th USENIX Symposium on Operating Systems Design and

Implementation.

Cachin, C., Schubert, S., Vukolic, M. (2016) ` Non-determinism in Byzantine Fault-

Tolerant Replication`, IBM Research  Zurich.

Camenisch, J., Lysyanskaya, A. (2001), Dynamic accumulators and application to

ecient revocation of anonymous credentials',CRYPTO '02, 2002, pp. 6176.

xv
Castro, M., Liskov, B. (1999),'Practical Byzantine Fault Tolerance', Proceedings

of the Third Symposium on Operating Systems Design and Implementation, New

Orleans, USA, February 1999.

Chaum, D. L. (1981), `Untraceable electronic mail, return addresses, and digital

pseudonyms', Communications of the ACM CACM Homepage archive, Volume 24

Issue 2, Feb. 1981, pp. 84-90.

Chaum, D. L. (1983), `Security Blind Signatures for Untraceable Payments', Ad-

vances in Cryptology pp. 199-203.

Chaum, D. L. (1985), `Security without Identication: Transaction Systems to Make

Big Brother Obsolete', Communications of the ACM, Volume 28 Issue 10, pp. 1030-

1044.

Coinchoose (2017), `Litecoin', https://www.coinchoose.com/coins/litecoin/, online

acceseed: 11.01.2018.

Coindesk (2017), `A (Short) Guide to Blockchain Consensus Protocols', https://

www.coindesk.com/short-guide-blockchain-consensus-protocols/,04.03.2017, online ac-

cessed: 11.01.2018.

Coinmarketcap (2017),'Cryptocurrency Market Capitalizations', https:// coinmar-

ketcap.com/, online accessed: 11.01.2018.

Cryptographyengineering (2013), `Zerocoin: making Bitcoin anonymous', https://

blog.cryptographyengineering.com/2013/04/11/zerocoin-making-bitcoin-anonymous/,

11.04.2013, online accessed: 11.01.2018.

Danezis, G., Meiklejohn, S. (2015), `Centrally Banked Cryptocurrencies', NDSS '16,

21-24 February 2016, San Diego, CA, USA

DashMasternode (2017), `What is Dash?', http://dashmasternode.org/what-is-dash/,

online accessed: 11.01.2018.

Decker, C., Wattenhofer, R. (2013),'Information Propagation in the Bitcoin Net-

work', 13-th IEEE International Conference on Peer-to-Peer Computing.

Diaz, D., Dueld, E. (2014), `Dash: A Privacy-Centric Crypto-Currency'.

Diedrich, H. (2016), `ethereum`, Wildre Publishing.

Dolev, D., Lamport, L., Pease, M., Shostak, R. (1987), `The Byzantine Generals',

Concurrency Control and Reliability in Distributed Systems, Bharat K. Bhargava,

editor, Van Nostrand Reinhold, pp. 349-369.

xvi
Druschel, P., Rowstron, A. (2001), `Pastry: Scalable, Decentralized Ob ject Location,

and Routing for Large-Scale Peer-to-Peer Systems', R. Guerraoui (Ed.): Middleware

2001, LNCS 2218, pp. 329350, 2001.

Ecomusing (2017), `Build Your Own Blockchain: A Python Tutorial', http://

ecomunsing.com/build-your-own-blockchain, 30.08.2017, online accessed:11.01.2018.

Emmadi, N., Narumanchi, H., (2017), `Reinforcing Immutability of Permissioned

Blockchains with Keyless Signatures' Infrastructure', ICDCN '17, January 04-07,

2017, Hyderabad, India.

European Central Bank (2015), `Virtual currency schemes - a further analysis',

www.ecb.europa.eu/pub/pdf/other/virtualcurrencyschemesen.pdf, online accessed:

11.01.2018.

Eyal, T., Gencer, A. E., Sirer, E. G., van Renesse R. (2016), `Bitcoin-NG: A Scalable

Blockchain Protocol', 13th USENIX Symposium on Networked Systems Design and

Implementation

Eyal, T., Sirer, E. G. (2013), `Ma jority is not enough: Bitcoin mining is vulnerable',

Department of Computer Science, Cornell University.

Fischer, M. J., Lynch N. A., Paterson, M. S. (1985), `Impossibility of Distributed

Consensus with One Faulty Process`, Journal of the Association for Computing

Machinery, Vol. 32, No.2, April 1985, pp. 372-384.

Gernandt, A., Gipp, B., Meuschke, N., (2015), `Decentralized Trusted Timestamping

using the Crypto Currency Bitcoin, iConference 2015.

Github (2017), `snakecoin-server-full code.py, https://gist.github.com/aunyks

/47d157f8bc7d1829a729c2a6a919c173, 23.07.2017, online accessed: 11.01.2018.

Goldreich, O., Micali, S., Wigderson A. (1991), `Proofs that Yield Nothing But Their

Validity or All Languages in NP Have Zero-Knowledge Proofs Systems`, Journal of

the Association for Computer Machinery, Vol. 38, No. 1, July 1991, pp.691-729.

Goldwasser, S., Micali, S., Racko, C. (1985), `The Knowledge Complexity of Inter-

active Proof-Systems`, Journal of the Association for Computer Machinery, pp.291-

304.

Gomber, P., Hinz, O., Nofer, N., Schiereck, D. (2017), `Blockchain`, Springer Fachme-

dien Wiesbaden.

Green, M., Garman, C., Miers, I., Rubin, A. D. (2013),` Zerocoin: Anonymous

Distributed E-Cash from Bitcoin`.

xvii
Greenspan, G. (2015), `MultiChain Private Blockchain-White Paper`.

Hyperledger (2015), `Hyperledger Whitepaper', https://docs.google.com/document/

d/1Z4M_qwILLRehPbVRUsJ3OF8Iir-gqS-ZYe7W-LE9gnE/edit

#heading=h.m6iml6hqrnm2, online accessed: 11.01.2018.

Ishai, Y., Kushilevitz, E., Ostrovsky, R., Sahai, A. (2006),`Cryptography from

Anonymity`.

Johnson, B., Grossklags, J., Laszka, A., Moore, T., Vasek, M. (2014), 'Game-

theoretic analysis of DDoS attacks against Bitcoin mining pools', Workshop on

Bitcoin Research

Kosba, A., Miller, A., Papamanthou, C., Shi, E., Wen, Z. (2016),' Hawk: The

Blockchain Model of Cryptography and Privacy-Preserving Smart Contracts', Uni-

versity of Maryland and Cornell University.

Lamport, L., Pease, M., Shostak, R. (1981), `The Byzantine Generals Problem',

ACM Transactions on Programming Languages and Systems, Vol. 4, No.3, July

1982, Pages 382-401.

Lamport, L. (2001), `Paxos Made Simple'.

Lippencott, J., Versluis, R., Viglione, R. (2017), `Zen White Paper'.

Liskov, B., Oki, B. M. (1988), `Viewstamped Replication: A New Primary Copy

Method to Support Highly-Available Distributed Systems', Massachusetts Institute

of Technology.

Mazieres, D. (2015),` The Stellar Consensus Protocol: A Federated Model for Internet-

level Consensus`, Stellar Development Foundation.

Medium (2017), `Let's Build the Tiniest Blockchain-In Less Than 50 Lines of Python',

https://medium.com/crypto-currently/lets-build-the-tiniest-blockchain-e70965a248b,

16.07.2017, online accessed: 11.01.2018.

Nakamoto, S. (2008), `Bitcoin: A peer-to-peer electronic cash system', bitcoin.org,

2008.

NavCoin (2016),'The Unbreakable Code NAVTECH WHITEPAPER 2016 Beta Re-

lease v0.9', https://www.navcoin.org/les/navtech-whitepaper-beta-v0.9.pdf, online

accessed: 11.01.2018.

Noether, S. (2017) `Ring Condential Transactions` Monero Research Labs.

Okamoto, T., Ohta, K. (1991), `Universal Electronic Cash', Advances in Cryptology

xviii
CRYPTO '91, LNCS 576, pp. 324-337.

Ongaro, D., Ousterhout, J. (2014), `In Search of an Understandable Consensus

Algorithm (Extended Version), Stanford University'.

Schnorr, C. P. (1991), Ecient signature generation for smart cards, Journal of

Cryptology, vol. 4, no. 3, pp. 239252, 1991.

Schwartz, D., Britto, A., Youngs, N. (2014), `The Ripple Protocol Consensus Algo-

rithm`,Ripple Labs Inc, 2014.

Schwartz, E., Thomas, S. (2015), `A Protocol for Interledger Payments'.

Steemit (2017), `An overview of blockchain privacy mechanisms and how Zerocoin

in Zcoin $XZC (not Zcash) stacks up', https://steemit.com/zcoin/@zcoinocial/an-

overview-of-blockchain-privacy-mechanisms-and-how-zerocoin-in-zcoin-usdxzc-not-zcash-

stacks-up, online accessed: 11.01.2018.

Swan, M. (2015), `Blockchain-A blueprint for a New Economy`, O'Reilly Media Inc.

Swanson, T. (2015), `Consensus-as-a-service: a brief report on the emergence of

permissioned, distributed ledger systems'.

Van Saberhagen, N. (2012), `CryptoNote v 1.0'.

Verge (2016), `Verge Blackpaper', https://vergecurrency.com/assets/Verge-Anonymity-

Centric-CryptoCurrency.pdf, online accessed: 11.01.2018.

Waves Platform (2017), `Review of blockchain consensus mechanisms', https://blog.

wavesplatform.com/review-of-blockchain-consensus-mechanisms-f575afae38f2, 31.07.2017,

online accessed: 11.01.2018.

Wood, G. (2017), `Ethereum: A Secure Decentralised Generalised Transaction Ledger'

EIP-150 Revision (1e18248 - 2017-04-12).

ZCoin (2017), `Understanding how Zerocoin in Zcoin works and how it compares to

other anonymity solutions Part 1`, http://zcoin.io/understanding-how-zerocoin-in-

zcoin-works-and-how-it-compares-to-other-anonymity-solutions-part-1/, 20.03.2017,

online accessed: 11.01.2018.

Zerocash-pro ject (2017),`How Zerocash works', http://zerocash-project.org/how_zerocash

_works, online accessed: 11.01.2018.

xix