A Formal Approach for the Analysis of the XRP Ledger Consensus
Protocol
Lara Mauri
1 a
, Stelvio Cimato
1 b
and Ernesto Damiani
1,2 c
1
Computer Science Department, Universit
`
a degli Studi di Milano, Milan, Italy
2
EBTIC - Khalifa University of Science and Technology, Abu Dhabi, U.A.E.
Keywords:
XRP Ledger, Ripple, Cryptographic Protocol, Consensus, Distributed Ledger.
Abstract:
Distributed ledger technology is envisioned as one of the cornerstones of promising solutions for building the
next generation of critical applications. However, there is still quite a bit of confusion and hype around the
real security guarantees this technology offers. This is especially due to the fact that for the vast majority
of existing blockchain-based consensus protocols it is really hard to find sufficiently detailed documentation
that fully captures their behavior. A number of recent papers have formalized the behavior of Bitcoin-like
protocols in order to rigorously study the security and privacy properties of their underlying structure, but
surprisingly very little work has been devoted to the formalization of distributed ledger systems using BFT-
like approaches. In this work, we focus on XRP Ledger, better known as Ripple, and take the first steps
towards the complete formalization of its consensus protocol. To this end, we have investigated all the existing
documentation and analyzed its source code. We present a formal description of its consensus protocol for
every step. Furthermore, we provide an accurate view of its security guarantees in terms of safety and liveness
and show how to increase the desired tolerance by changing the value of specific protocol parameters.
1 INTRODUCTION
Although the classical problem of achieving mutual
agreement in the presence of faulty nodes (Pease
et al., 1980) has been extensively studied in the liter-
ature from different perspectives (e.g. resiliency, net-
work connectivity, communication complexity, run-
ning time and channel reliability), interest in consen-
sus protocols has risen quite significantly over the last
few years because of their application to various novel
settings such as blockchain systems. The novelty of
blockchain technology is a powerful combination of
well-known research results taken from distributed
computing, cryptography and game theory that to-
gether create a system providing a trustworthy service
to parties who otherwise have no reason to trust each
other. The term blockchain is self-explanatory and
denotes a distributed ledger (i.e. a public data struc-
ture) where inputs called transactions are aggregated
in the form of blocks which are added one after an-
other. A copy of the ledger is locally maintained by
a
https://orcid.org/0000-0002-4024-1015
b
https://orcid.org/0000-0003-1737-6218
c
https://orcid.org/0000-0002-9557-6496
all the nodes, which operate in a peer-to-peer commu-
nication setting and continuously send and validate
transactions in an attempt to update the state of the
shared ledger. In principle, any participant can pro-
pose the addition of a new block to the blockchain,
but this addition has to be approved by the other peers
to ensure its legitimacy.
The idea behind what is today known as the
Nakamoto consensus (or ledger consensus) was first
conceptualized in 2008 with the birth of the Bitcoin
payment system (Nakamoto, 2008) and refers to the
consensus mechanism allowing the nodes in the Bit-
coin network to decide which next block to include
into the ledger. Specifically, the ultimate goal of
the consensus protocol is to ensure that parties have
a unique common view of the ledger even in the
presence of possible conflicting inputs and arbitrary
Byzantine behaviors of some network nodes. In or-
der to guarantee consistency between potentially dif-
ferent versions of the ledger, Bitcoin relies on the
concept known as proof-of-work (PoW), which re-
quires a party to invest computational power to af-
fect the behavior of the system. However, the ex-
traordinary amounts of energy it demands have raised
concerns over its sustainability and environmental im-
52
Mauri, L., Cimato, S. and Damiani, E.
A Formal Approach for the Analysis of the XRP Ledger Consensus Protocol.
DOI: 10.5220/0008954200520063
In Proceedings of the 6th International Conference on Information Systems Security and Privacy (ICISSP 2020), pages 52-63
ISBN: 978-989-758-399-5; ISSN: 2184-4356
Copyright
c
2022 by SCITEPRESS Science and Technology Publications, Lda. All rights reserved
pact. These limitations motivated an interesting re-
search direction that proposes the adoption of a new
class of consensus protocols based on the use of alter-
native resources (Bano et al., 2017; Wang et al., 2018)
among them, the proof-of-stake (PoS) paradigm is
the most implemented so far. Such developments,
combined with recent efforts to leverage the core fea-
tures of blockchains for novel use cases in both finan-
cial and non-financial domains (Rawat et al., 2019),
have led to a significant increase in proposals for
blockchain-based solutions, with new software appli-
cations and services appearing daily. But this rush to
quickly enter the market poses serious risks to the en-
tire blockchain ecosystem since vulnerabilities of in-
sufficiently tested code can be exploited with signifi-
cant consequences due to the decentralized nature and
irreversibility of the technology (Saad et al., 2019).
According to Halpin and Piekarska (2017), an
emerging field that needs further research is the study
of the security and privacy properties of the structure
underlying blockchain-based systems. In this context,
the first formal security analysis of the fundamental
principles behind the Nakamoto consensus was pro-
posed by Garay et al. (2015). In their work, the au-
thors presented an abstraction of the Bitcoin proto-
col in synchronous networks, which they referred to
as the Bitcoin backbone, and proved that it satisfies
certain security properties. Later, Pass et al. (2017)
extended the analysis to asynchronous networks, and
further refinements of the initial model of computa-
tion were presented in subsequent works (Garay et al.,
2017, 2018). While a number of other papers have
studied the security of several PoW- and PoS-based
blockchains in a rigorous manner (Pass and Shi, 2017;
Badertscher et al., 2017; Daian et al., 2016; Kiayias
et al., 2017), it is worth noting that for the vast ma-
jority of existing blockchain protocols it is really hard
to find sufficiently detailed information on the func-
tioning of the consensus mechanism they use (though
there is a lot of work presented online in the form of
white papers, pre-prints and blog posts). Moreover,
the majority of current research is focusing on study-
ing blockchains consistent with the so-called permis-
sionless setting, whereas much less attention has been
paid to the formalization of distributed ledger sys-
tems operating in permissioned settings, which em-
brace aspects of traditional Byzantine-fault tolerant
(BFT) consensus protocols. In contrast to permission-
less blockchains such as Bitcoin, permissioned ones
are operated by known entities and only a restricted
group of nodes can participate in the process for de-
termining the correct update of the ledger (Cachin and
Vukolic, 2017).
Motivated by this emerging new era of research,
in this work we focus our attention on XRP Ledger
better known as Ripple (Ripple Labs Inc., a) ,
the third-largest digital currency in market capitaliza-
tion after Bitcoin and Ethereum. The XRP Ledger
protocol is a distributed payment system that uses a
unique consensus protocol which sets it apart from
previous approaches. Based on the idea of subjective
trust assumptions, in Ripple every participant is free
to specify which nodes it believes will behave cor-
rectly and which ones it considers faulty by means of
so-called Unique Node Lists (or UNLs). The fact that
the XRP Ledger protocol does not rely on a process
of mining to verify transactions has many advantages,
in terms of transaction speed, compared to other con-
sensus algorithms (Mauri et al., 2018). Indeed, while
for Bitcoin-like protocols transactions are confirmed
after an hour on average, in the XRP Ledger network
the settlement process takes around 5 seconds, with a
throughput of up to 1500 transactions per second.
Although mentioned in several surveys on the
consensus mechanisms that govern the most popu-
lar blockchains (Xiao et al., 2019; Gramoli, 2017),
the Ripple protocol is often described in a vague and
not very clear way. By now the original white pa-
per (Schwartz et al., 2014) is deprecated, and avail-
able documentation about how the consensus process
works is restricted to the developer portal (Ripple
Labs Inc., b) and a recent analysis provided by the cre-
ators of Ripple themselves (Chase and MacBrough,
2018). Even if the latter presents a detailed explana-
tion of the Ripple algorithm and derives conditions
for its safety and liveness, there remains a degree
of uncertainty around many aspects of the protocol
and, in any case, not all steps have actually been de-
scribed. Furthermore, to date the only existing peer-
reviewed analysis of the XRP Ledger consensus pro-
tocol (Armknecht et al., 2015) was conducted on the
original white paper and showed that some specifica-
tions were flawed.
Our Contributions. In light of the above, we be-
lieve that having a clear and accurate description of
the XRP Ledger consensus protocol is fundamental
not only to gain in-depth understanding of its behav-
ior, but also to increase trust in the foundational prin-
ciples of the algorithm and identify its vulnerabilities.
To this end, we have carefully investigated all the ex-
isting documentation relating to XRP Ledger and, by
directly analyzing its source code (Ripple Labs Inc.,
c), we have created an in-depth description of the con-
sensus protocol for every step. In particular, our work
contributes in the following aspects: (1) taking the
first steps towards the complete formalization of the
XRP Ledger Consensus Protocol; (2) providing an ac-
curate view of the current security guarantees of the
A Formal Approach for the Analysis of the XRP Ledger Consensus Protocol
53
protocol in terms of safety and liveness; (3) show-
ing how the inherent correlation between the safety
and liveness tolerances, the validation quorum and the
UNLs overlapping size can be leveraged to increase
either liveness or safety.
Organization. The remainder of the paper is or-
ganized as follows. Section 2 provides the neces-
sary background to understand the consensus mech-
anism powering the XRP Ledger network. Section 3
presents our formalization of the different phases of
the XRP Ledger Consensus Protocol, expressed by
means of a set-theoretic notation and mathematical
formulas. Section 4 reports and clarifies results from
a previous analysis with respect to safety and liveness
properties and extends it by showing that the value of
some protocol parameters can be changed to guaran-
tee more liveness or safety. Section 5 concludes and
proposes directions for future research.
2 BACKGROUND ON THE XRP
LEDGER NETWORK
2.1 Basic Concepts and Terminology
In the XRP Ledger network, the collaborative process
by which the shared replicated transaction system is
updated and kept consistently synchronized is known
as the XRP Ledger Consensus Protocol (or XRP LCP,
for short). The primary objective of each participant
involved in the XRP LCP is to advance the chain of
validated ledgers by reaching agreement on a new set
of transactions to apply to the prior ledger. The over-
all consensus process, which can be viewed as a pro-
gression through three phases collection, deliber-
ation and validation is controlled by the so-called
validators. Unlike simple tracking nodes, which are
uniquely responsible for relaying transactions from
clients throughout the network, participants acting as
validators perform the additional task of proposing
and revising candidate sets of transactions to be pro-
cessed.
Several permissioned blockchain platforms rely
on classical BFT consensus protocols, which typi-
cally require every node to know the entire set of
peers participating in the consensus process. How-
ever, one of the major challenges for those protocols
that prevents their wider adoption in blockchain sys-
tems is their scalability in terms of the number of
nodes (Vukoli
´
c, 2016). Conversely, the XRP Ledger
protocol does not rely on a strict notion of member-
ship and not every node has to trust all others. This
gives the XRP Ledger network a clear advantage over
BFT-based blockchains because it is more scalable in
terms of the number of nodes. In order to achieve con-
sensus, validators share information about their candi-
date transaction sets by exchanging signed messages
(whose format will be described in detail later on), but
nodes only consider messages sent by specific valida-
tors they believe will behave honestly. In particular,
each participant individually chooses which valida-
tors it will listen to when making decision about the
next ledger, and such choice is declared in the form
of a list, referred to as the Unique Node List (UNL),
which is locally maintained by each validator; valida-
tors belonging to the UNL are called trusted.
A fundamental component of the first phase of the
consensus protocol is represented by the open ledger,
which is the ledger to which all pending transactions
are applied. When certain specific conditions are met,
the ledger is closed, and the closure implies that new
transactions are no longer accepted for inclusion in
the new version of the ledger. In essence, the set of
transactions included in the closed ledger constitutes
the first position and reflects the validator’s initial be-
lief of which transactions should be considered for the
next ledger. Nodes are equipped with a public/private
key pair. Digital signatures are used to check that sub-
mitted transactions are correct, i.e. authorized to do
a specific set of actions, but also serve as a means
to authenticate the issuers of the messages received.
The fact that all protocol communications are signed
makes it possible to reliably identify validators in the
peer-to-peer network even if messages are relayed by
untrusted nodes.
As mentioned above, validators communicate by
means of messages in an attempt to reach network
agreement. One of such messages is represented by
the so-called proposals. At the start of the consen-
sus process, each node decides whether to operate as
an active or passive validator. A proposing valida-
tor is a normal participant taking on its own position
and sending proposals to other peers. Conversely, an
observing validator updates its position, but does not
propose it to its peers. The first proposal issued by
a validator corresponds to its initial position, so pro-
posals can be seen as subsequent positions taken by
nodes as the network progresses. As we will see later,
the concept of dispute is strictly connected to that of
proposal. It is used to denote an individual transaction
that is either not included in a validator’s proposal or
not included in a trusted validator’s proposal.
In addition to the closed ledger, we can distinguish
two other types of ledger representing subsequent
evolutions, in terms of content validity and finality, of
the initial ledger version under consideration. Specif-
ically, the ledger resulting from the completion of de-
ICISSP 2020 - 6th International Conference on Information Systems Security and Privacy
54
liberation, known as the last closed ledger, is the most
recent ledger on which, from the point of view of a
single validator, the network reached agreement. This
ledger version is not immediately considered final. In-
deed, a ledger is fully validated only once it passes a
certain validation threshold. It is only at that point that
it is considered authoritative and irrevocable, meaning
it becomes part of the permanent public history and its
content is truly reliable.
Lastly, a validation is a signed message containing
the hash of the last closed ledger, whose purpose is
to let trusted validators know which particular ledger
was built by the other validators and come to a com-
mon conclusion about which last closed ledger should
be considered authoritative (i.e. declared fully vali-
dated).
2.2 Informal Protocol Description
The XRP LCP proceeds in rounds and consists of
three different phases that are continually repeated to
process groups of transactions. The typical execution
of a single ledger round goes from collection to de-
liberation to validation. Below we provide a brief de-
scription of each phase:
Collection: During this phase, validators collect
new incoming transactions, apply the well-formed
ones to their open ledger and subsequently relay
them to the other peers. When certain conditions
are met, each validator determines the open ledger
period should end and closes its ledger.
Deliberation: Validators begin an iterative pro-
cess in which they try to establish consensus by
exchanging proposals with their trusted valida-
tors. Starting from their own position, they con-
tinuously change their view of the transaction set
by adding or removing disputed transactions from
their proposal and, after each update, they check
the percentage of trusted validators agreeing with
them. When a validator sees a supermajority
agreement, it declares that consensus has been
reached. Then, it creates a new last closed ledger
by applying the consensus transaction set to the
last validated ledger and issues a validation.
Validation: Participants determine whether to
consider the outcome of deliberation (i.e. their
last closed ledger), fully validated by comparing
the hashes received from their trusted peers. The
decision is taken on the basis of the percentage
of agreement on the newly-created ledger. Only
when a quorum of validations for the same ledger
is reached, that ledger is deemed final.
3 FORMALIZING THE XRP LCP
3.1 Preliminaries
In the XRP Ledger network, the shared global
ledger is actually a series of ledger versions (or
ledgers). A ledger is any quintuple of the form
L = hsn,h,h
1
,ts,T i where sn is the sequence num-
ber of the ledger, h is the unique identifying hash of
the ledger’s contents, h
1
is the h value of the pre-
vious ledger version this ledger was derived from, ts
is a timestamp, and T is a set of transactions. The
hash serves as a unique identifier for the ledger and
its contents, whereas the sequence number identifies
the order in which ledgers occur in the ledger chain.
The very first ledger, known as the genesis ledger, has
sequence number 1; each successive ledger has a se-
quence number one greater than that of the preced-
ing one. We say a tuple L = hsn, h,h
1
,ts,T i is valid
with respect to a predecessor (fully validated) ledger
L
0
= hsn
0
,h
0
,h
0
1
,ts
0
,T
0
i iff:
1. h
1
= h
0
;
2. sn = sn
0
+ 1;
3. |Σ
L
| q
v
, where |Σ
L
| is the support of the ledger L
and q
v
is the minimum percentage of participation
required to reach network agreement.
Below we introduce the notation used in the formal-
ization of the XRP LCP provided in Section 3.2, as
well as the timing assumptions.
3.1.1 Notation
We denote by N the universe of nodes of the peer-
to-peer XRP Ledger network. Nodes play differ-
ent roles: client applications submit transactions to
server nodes, which are differentiated in tracking
nodes and validators. The consensus process is ex-
clusively driven by the latter. Since only validators
actively participate, check and validate all transac-
tions taking place in the system, we believe that for
the purposes of formalization it is meaningful to con-
sider only the subset N
v
N of participants, where
N
v
denotes exactly the nodes acting as validators. As
already mentioned, during the consensus process, val-
idators only evaluate proposals and validations issued
by their trusted nodes, discarding those received from
validators not belonging to their UNL. For any node
i N
v
, U NL
i
= {u : u N
v
, F(u)} denotes the set of
all nodes u for which F(u) is true, where F(u) means
that validator i trusts u. Also, we use n
i
= |U NL
i
|
to denote the size of node is UNL. Giving as a fact
that the individual choice to include (or omit) a given
transaction in the next ledger is influenced only by
A Formal Approach for the Analysis of the XRP Ledger Consensus Protocol
55
the trusted nodes’ messages, we decide not to directly
specify UNL membership every time (in our formal-
ization, we will specify this dependency only in the
context of the last phase). Deliberation and validation
phases are parameterized by q
c
and q
v
, respectively.
In particular, q
c
specifies the consensus quorum (i.e.
the minimum number of validators having issued the
same proposal needed to declare consensus), whilst
q
v
see the third condition required for ledger valid-
ity represents the validation quorum (i.e. the min-
imum number of validations needed to fully validate
a ledger). Both these parameters are a function of a
constant k and n
i
, where n
i
is the previously defined
size of UNL
i
. We let
˜
L
i
,
˜
L
c
i
and L
0
i
denote the open
ledger, the closed ledger and the last closed ledger of
node i, respectively. Moreover, we use
ˆ
L to denote
the last fully validated ledger. tx is used to represent a
single transaction under consideration by consensus.
Any transactions excluded from a node’s proposal are
added to its local queue we denote by T
Q
. For nota-
tional simplicity, we suppose the existence of another
set T
I
which contains all received new transactions.
Also, we use P
i
to denote the node is proposal of
candidate transactions and the symbol θ to indicate
the threshold for including a given transaction in the
proposal P
i
. Lastly, D
i
denotes the node is set of dis-
puted transactions, whereas σ
i
denotes a validation is-
sued by i. Table 1 provides a summary of the notation
used throughout this paper (some parts however will
be slightly modified to suit the context).
Table 1: Summary of notation.
N
v
The set of validators
i A validator in the network
UNL
i
is Unique Node List
n
i
The size of U NL
i
tx A single transaction
T A set of transactions
T
Q
The set of queued transactions
T
I
The set of new transactions
q
c
The consensus quorum
q
v
The validation quorum
˜
L
i
is open ledger
˜
L
c
i
is closed ledger
L
0
i
is last closed ledger
ˆ
L The last fully validated ledger
P
i
is proposal of transactions
D
i
is set of disputed transactions
σ
i
is validation
3.1.2 Timing
An important aspect that must be considered when de-
signing any consensus protocol is the ability of parties
to reach a certain degree of synchronization during
the protocol execution. In the XRP Ledger network,
the protocol execution is driven by a heartbeat timer
which allows nodes to advance the consensus process.
In practice, at regular intervals each validator checks
whether the necessary conditions to move to the next
phase of the consensus protocol are met. Although
each validator can begin a new round of consensus at
arbitrary times based on when the initial round started
and the time taken to apply the transactions, the tran-
sition to both deliberation and validation phases can
only occur at the end of the heartbeat timer.
Furthermore, each node maintains an internal
clock that it uses when calculating its own close time,
i.e. the time at which the ledger has been closed (see
Sections 2.1 and 3.2.1 for details). In practice, at the
closure of the ledger (which reflects the end of col-
lection phase) each validator uses its current time as
the initial close time estimate and includes this value
in its initial position. Later, this approximate time is
rounded based on a dynamic resolution. As a result,
each validator uses the consensus process not only to
reach agreement with its trusted nodes about the set of
transactions, but also attempts to agree on a common
time for the closure of the ledger. Validators update
their close time positions in response to the effective
close time included in their trusted nodes’ positions,
and in this way, they can derive a common close time
for the ledger without the need to synchronize their
clocks. When there is no clear majority of trusted val-
idators agreeing on a close time, nodes agree to dis-
agree on an actual close time. Even though in this
case the network has no official close time, the effec-
tive close time of the ledger is set to 1 second past the
close time of the previous ledger.
Although reaching agreement on the effective
close time is part of the XRP LCP, in order to make
our formalization more readable, we decided to in-
clude in it only some fundamental protocol timing pa-
rameters.
3.2 In-depth Protocol Description and
Its Formalization
In Ripple, the ledger chain structure starts with the
genesis ledger and ends with the last fully validated
ledger. Generally, given the prior fully validated
ledger
ˆ
L and a set T of new candidate transactions, the
execution of the XRP Ledger Consensus Protocol de-
termines an output ledger
ˆ
L
0
which, together with the
previous validated ledgers, forms the ledger history.
As introduced in the foregoing section, the XRP LCP
is an asynchronous round-based protocol consisting
of three phases. The diagram of Fig. 1 shows the out-
ICISSP 2020 - 6th International Conference on Information Systems Security and Privacy
56
put expected from each phase from the perspective of
an individual node i N
v
: (a) the output of the col-
lection phase is represented by is initial proposal P
0
i
,
which contains the starting position taken by the val-
idator itself before considering any trusted validator
position; (b) at the end of the second phase, i builds
a new last closed ledger L
0
i
; (c) the validation phase
establishes which last closed ledger, amongst those
proposed by all participants, must be considered the
authoritative one (
ˆ
L
0
).
We now proceed with the formalization of the
three main consensus phases. Note that in our formal-
ization we follow the flow depicted in Fig. 1. That
is, we consider the perspective of a single validator
i N
v
, which makes decisions only by listening to
its unique node list UNL
i
. In order to formalize the
behaviour of the XRP LCP we referred to the cur-
rent implementation of the core XRP Ledger server,
called rippled, whose source code is available in the
GitHub repository (Ripple Labs Inc., c).
3.2.1 Collection
The first stage of the process is a quiescent period al-
lowing the validator i to create an individual percep-
tion of the state of the network. This perception is
referred to as the open ledger. The node’s open ledger
is filled with the candidate transactions which failed
to be included in the last round (i.e. those queued in
T
Q
), as well as newly submitted ones. is open ledger
˜
L
i
contains the set of transactions T :
T = T
Q
T
I
, T
˜
L
i
(1)
Under normal circumstances, the closure of the ledger
occurs after a predetermined minimum time t
minclose
has elapsed and only if the open ledger contains at
least one transaction. Thus, i closes its open ledger if
the following holds:
T 6=
/
0 (t
open
t
minclose
), T
˜
L
i
(2)
where t
open
is used to denote how long the ledger has
been open. We let
˜
L
c
i
indicate that is open ledger
has been closed (superscript letter c). In other cases,
the protocol either (1) closes the ledger if more than
half of the proposers have closed the ledger or vali-
dated the last closed ledger, or (2) postpones the clo-
sure to the end of a certain idle interval so that it is
more likely to include some transactions in the next
ledger. Regardless of the condition which led to the
closure of the ledger, once the pre-close phase is com-
pleted, i establishes its initial position on the basis of
the transactions included in the newly closed ledger.
Then, i proposes it to its trusted validators in the form
of a proposal, i.e. a signed message containing the
transactions it believes should be included in the next
ledger (recall from Section 2.1 that a node not operat-
ing in a proposing mode maintains its position inter-
nally without sharing it with peers):
P
0
i
= {tx : tx T, T
˜
L
c
i
} (3)
Here the notation used for the proposal slightly dif-
fers from that given in Section 3.1.1; we introduced
the superscript 0 to denote that P
i
is the initial pro-
posal shared by i. Like ledgers, also proposals have a
sequence number that is incremented each time a val-
idator updates the transaction set contained therein.
As easy to guess, different nodes may receive dif-
ferent unconfirmed transactions due to network de-
lays. As this variation in arrival time allows for each
potential new ledger version to be different for each
node, immediately after sharing its initial position, i
considers all the proposals received from its trusted
validators and creates a dispute for each transaction
discovered not to be in common with the peer’s posi-
tion under consideration. We define the validator is
dispute set D
i
as:
D
i
= {tx : (tx P
r
i
tx / P
r
j6=i
)
(tx / P
r
i
tx P
r
j6=i
)}
(4)
where r refers to one of the successive proposals is-
sued by i during deliberation (at this time, r = 0).
Also, the validator keeps track of each peer’s vote on
each disputed transaction (voting in favour of a dis-
puted transaction simply means believing that such
transaction should be included in the next ledger). For
each tx D
i
, we use v(tx) to denote the support given
to tx by is trusted validators:
v(tx) = | j 6= i : tx P
r
j
| (5)
where, as before, r is currently 0 (in a more advanced
stage of the process, i.e. during deliberation, P
r
j
will
refer to the most recent proposal issued by j).
Unless they are part of a peer’s proposal, transac-
tions received after the closure of the ledger are not
considered until the next round.
3.2.2 Deliberation
In order to achieve consensus on the specific set of
transactions to be processed next, i begins an itera-
tive process during which it issues updated proposals
(i.e. updates of its initial position), which are modi-
fied based on the support obtained by each individual
transaction. A heartbeat timer is the key ingredient
that drives the consensus process forward: it is only
on timer calls, which occur at a regular frequency, that
i adjusts and issues new proposals (see Section 3.1.2).
Going into detail, the mechanism allowing differ-
ences between peers’ proposals to converge is based
A Formal Approach for the Analysis of the XRP Ledger Consensus Protocol
57
Figure 1: Output of each phase from the perspective of a single validator i.
on a majority voting scheme. The evaluation of the
achieved convergence degree is performed taking the
value of a certain threshold θ as a reference. In turn,
the value θ depends on a parameter d which expresses
the percentage of time that consensus is expected to
take to converge. Specifically, d is a function of the
previous round duration, which corresponds to the du-
ration of the last deliberation phase (measured from
closing the open ledger to accepting the consensus re-
sult) and the duration of the deliberation phase for the
current consensus round:
d = f (t
(deliberation1)
, t
deliberation
) =
=
t
deliberation
· 100
max(t
(deliberation1)
, t
minconsensus
)
(6)
where t
minconsensus
(currently, 5 seconds) expresses the
minimum amount of time to consider that consensus
was reached in the previous round.
As deliberation proceeds, i changes its candidate
transaction set in response to what its trusted valida-
tors propose. As a result, i may either add a dis-
puted transaction to its current set if the percentage of
trusted validators that have proposed the same trans-
action in their most recent proposal exceeds the afore-
mentioned threshold θ or, otherwise, remove it. New
proposals are formalized as follows:
P
0
i
= {tx : (tx P
r
i
tx / D
i
)
(tx D
i
(v(tx) θ))}.
(7)
The current implementation uses an initial threshold
for inclusion of 0.5. This means that if a particular
disputed transaction is supported by half of the val-
idators or more, i agrees to include it in its set. On the
contrary, a transaction that, at this early stage, does
not have the support of at least 50% of the trusted
nodes, is removed from is position. Omitted transac-
tions are added to the transaction queue T
Q
and con-
sidered again for inclusion in the next ledger version.
Therefore, the queue is updated whenever a transac-
tion is removed from a proposal:
T
0
Q
= T
Q
{tx : tx D
i
(v(tx) < θ)}. (8)
The specific values of θ are subject to change, in ac-
cordance to the XRP Ledger implementation. De-
pending on the value of the function d (defined in Eq.
6), θ can currently assume one of the following val-
ues:
0.5 d < 50
0.65 50 d < 85
0.7 85 d < 200
0.95 otherwise
(9)
After changing its transaction view, i broadcasts its
new proposal (increasing the related sequence num-
ber) and determines whether consensus has been
reached in case i has not changed its set of trans-
actions, no new proposal needs to be issued. Consen-
sus can be declared only if there is a supermajority
support for each of the transactions the candidate set
contains. Formally, the support of a proposal P
r
i
is
expressed as follows:
v(P
r
i
) = |{ j 6= i : P
r
j
= P
r
i
}|. (10)
Actually, consensus is reached when the following
three conditions are met: (a) a certain minimum
amount of time has elapsed since deliberation be-
gan, (b) at least 3/4 of the previous round proposers
are participating or the current deliberation phase is
longer (of a given minimum amount of time) than
the deliberation phase of the previous round, (c) i,
together with its trusted peers, has reached the per-
centage threshold q
c
above which consensus can be
declared. However, for the purpose of formalization,
the really meaningful condition is the third. Hence,
we assume i declares consensus reached when the fol-
lowing holds:
v(P
r
i
) q
c
, q
c
= k · n
i
(11)
where n
i
is the number of trusted nodes belonging to
UNL
i
(see Section 3.1.1) and k = 0.8 in the current
implementation. Once consensus is reached, i builds
a new last closed ledger L
0
i
by adding the agreed-upon
set of transactions to the prior fully validated ledger
ˆ
L:
L
0
i
=
ˆ
L {tx P
r
i
: v(P
r
i
) q
c
} (12)
and then, it broadcasts its resulting ledger in the form
of a signed message σ
i
containing the identifying hash
of this ledger. At this point, the round is completed
and i now builds a new open ledger on top of the last
closed ledger. As previously said, the open ledger
ICISSP 2020 - 6th International Conference on Information Systems Security and Privacy
58
represents the basis of any validator’s initial proposal
and hence, the first transactions to be added to the
new open ledger are those held over from the previ-
ous consensus round. Next, all valid transactions that
in the previous round were received after the ledger
was closed are applied. In practice, the protocol starts
a new round of consensus before the third phase, i.e.
validation, ends. Each participant begins a new col-
lection phase concurrently, preparing its proposal for
the next ledger version while the consensus process
related to the prior ledger version is ongoing. Accord-
ingly, at any given time, the process is characterized
by a series of in-progress open ledgers, individual par-
ticipants’ last closed ledgers that have not been fully
validated and historical ledgers that have been fully
validated.
3.2.3 Validation
At the end of deliberation, as seen above, validators
independently build what they believe the next state
of the ledger should be. Due to latency in propagating
transactions throughout the network, different valida-
tors may compute different last closed ledgers. These
ledger versions have the same sequence number, but
different hash values that uniquely identify their con-
tents (see Section 3.1). It is important to note that the
sequence number and the hash correlate 1:1 only for
fully validated ledgers. Thus, for a given sequence
number, only one ledger version can ever be fully val-
idated. In order to converge on a universal status of
the ledger and, consequently, resolve the above dif-
ferences, i checks how many trusted validators have
built a last closed ledger equal to its own by compar-
ing its validation, i.e. the hash of the ledger it com-
puted, with the hashes received from its peers; among
all the most recent validations, i considers only those
with the greatest sequence number. We denote by Σ
L
0
i
the set of trusted validators that published the same
validation hash as the validator i:
Σ
L
0
i
= { j UNL
i
: σ
j
= σ
i
}. (13)
Based on these validations, i recognizes whether the
previous consensus round succeeded or not. In partic-
ular, the node declares its last closed ledger L
0
i
fully
validated in the event that a supermajority agreement
on the calculated hash is reached. Therefore, if i de-
tects that it is in the majority, having built a ledger that
received enough matching validations to meet the val-
idation quorum q
v
, it considers the ledger fully vali-
dated. Formally, L
0
i
is declared fully validated iff:
|Σ
L
0
i
| q
v
, q
v
= k · n
i
(14)
where, as for the consensus quorum q
c
(cf. Eq. 11),
the constant k is 0.8. Conversely, if i is in the minority,
it cannot consider its ledger instance fully validated.
Instead, it has to find the supermajority ledger, i.e.
the one with the sufficient number of validations (and
highest sequence number), and accept it as the new
fully validated ledger:
ˆ
L
0
= L
0
x
: (x = i x U NL
i
) (|Σ
L
0
x
| q
v
). (15)
Lastly, in case for a given sequence number no last
closed ledger meets q
v
, no ledger is declared fully
validated and i switches to a strategy allowing it to
determine the preferred last closed ledger for the next
consensus round (see (Chase and MacBrough, 2018)
for an overview on this strategy).
To summarize, the three outputs depicted in Fig. 1
correspond to Eq. 3, 12 and 15 respectively.
4 XRP LEDGER PROPERTIES
4.1 Safety and Liveness
One of the fundamental problems in distributed com-
puting is ensuring that in a system where processors
exchange information, all correct processes make a
decision and eventually reach a common understand-
ing or agreement even if one or more processors have
failed. There are several flavors of the consensus
problem (Pease et al., 1980), but all variations are sub-
ject to conditions similar to the following:
Validity: If every process begins with the same
initial value v, then the final decision of a non-
faulty process must be v;
Agreement: Final decisions by non-faulty proces-
sors must be identical;
Termination: Every non-faulty process must even-
tually decide.
These conditions are intended to capture two crucial
properties, namely safety and liveness. The former,
which derives from the conjunction of validity and
agreement, traditionally guarantees that something
bad will never happen. On the other hand, liveness
guarantees that something good will happen eventu-
ally, and derives from the termination condition. In
the context of the XRP Ledger network, the above
definitions can be refined as follows:
Safety: If an honest node fully validates a ledger
L, then all honest nodes cannot fully validate a
contradictory ledger L
0
6= L;
Liveness: If an honest node broadcasts a valid pro-
posal P to all honest nodes, then P will eventually
be accepted by all nodes and included in a fully
validated ledger.
A Formal Approach for the Analysis of the XRP Ledger Consensus Protocol
59
The phenomenon called fork is the biggest threat to
the ability of a distributed ledger system to operate
properly and produce results that can be relied upon.
In the XRP Ledger network, the occurrence of a fork
corresponds to the situation in which two honest val-
idators fully validate conflicting ledgers, i.e. ledgers
having the same sequence number, but different iden-
tifying hash value.
As already mentioned, the XRP LCP features a
layered notion of trust. The network is divided into
subsets of nodes that are collectively trusted to not
collude in an attempt to defraud the other peers, and
each node has complete discretion in selecting its
own UNL. Since validators have influence only over
nodes configured to trust them, the presence of a cer-
tain honest validator in more UNLs directly implies a
higher influence by that validator on the process of de-
termining the next ledger state. As a result, protocol
safety relies on the fact that the majority of the val-
idators act honestly and it is strongly dependent on the
degree of intersection between the UNLs of every pair
of nodes. The original Ripple whitepaper (Schwartz
et al., 2014) provided an initial analysis of the over-
lap condition required to ensure consistency, coming
to the conclusion that the needed minimum overlap
was roughly 1/5 of the UNL size. Later, an indepen-
dent analysis (Armknecht et al., 2015) proved that the
above condition was insufficient and suggested that
the correct bound was > 2/5 of the UNL size. To be
precise, they showed that in order to ensure the ab-
sence of any fork, the intersection set size between
the UNL of any two validators needs to be > 40%
of the maximum UNL size. A recent work by Chase
and MacBrough (2018) provided a further analysis of
the XRP Ledger Consensus Protocol and also derived
new conditions for its safety and liveness, changing
the expectation for the overlap requirement. In the
remainder of this section, we show the major find-
ings of Chase and MacBrough which will be useful in
the subsequent section. Prior to this, we present the
model used to analyze the safety and liveness of the
protocol.
The network is modeled as if an adversary is in
full control of at most t
i
nodes in U NL
i
for any val-
idator i. The nodes controlled by the adversary are
called Byzantine and can deviate from the protocol
by performing at least one of the following actions:
(i) not responding to messages; (ii) sending incor-
rect messages; (iii) sending differentiated messages
to different nodes. On the contrary, we consider as
honest any node that follows exactly the prescribed
XRP LCP. The honest proportion is equal to n
i
t
i
.
Due to the FLP Impossibility Result (Fischer et al.,
1985), safety and liveness cannot be simultaneously
guaranteed by any consensus algorithm in the pres-
ence of arbitrary asynchrony and Byzantine nodes. In
(Chase and MacBrough, 2018), the authors assumed
a weak form of asynchrony in order to prove that the
system is able to not fall in a state where some hon-
est nodes can never fully validate a new ledger. This
last assumption, however, does not seem to be suf-
ficient to guarantee that the system cannot get stuck
even in networks where two UNLs disagree only by
few nodes. In this regard, the paper showed an exam-
ple where even with 99% UNL overlap and no Byzan-
tine faults, the system fails to successfully apply the
preferred branch strategy, consequently maintaining
a ledger chain with two distinct branches (in other
words, the nodes are unable to determine a preferred
chain of ledgers to converge on). It follows that, actu-
ally, only in the restricted case in which the network
consists of a single agreed-upon UNL with an arbi-
trary number of extra nodes, the XRP LCP cannot get
stuck.
The most relevant contribution by Chase and
MacBrough was the re-analysis of the overlap con-
dition required to ensure safety. According to them,
the XRP LCP guarantees fork safety if for every pair
of nodes i, j the following holds:
O
i, j
>
n
j
2
+ n
i
q
v
+t
i, j
(16)
where O
i, j
= |UNL
i
UNL
j
|, q
v
= k · n
i
(cf. Eq. 14)
and t
i, j
= min{t
i
,t
j
,O
i, j
} is the maximum number of
allowed Byzantine faults in U NL
i
UNL
j
here we
have slightly changed the original notation in order to
adapt it to the one used in our formalization. Assum-
ing 80% validation quorum (q
v
) (as prescribed in the
actual implementation) and 20% faults (t
i, j
), the con-
dition in essence requires roughly > 90% UNL agree-
ment.
4.2 Analysis
In order to provide a more accurate view of the secu-
rity provisions of the XRP LCP in terms of safety and
liveness, here we make some observations about the
three key parameters of the protocol, i.e. UNL over-
lap, validation quorum and tolerated Byzantine nodes,
considering also the discussion recently appeared on
the Ripple GitHub repository (Wilson, 2018). We in-
vestigate the values O
i, j
, q
v
and t
i, j
displayed in Eq.
16 and in the following we show how safety and live-
ness tolerances are related to each other and vary ac-
cording to the assumptions made on the above param-
eters.
To this end, we find it convenient to introduce two
additional parameters, namely µ
i
and µ
j
, which de-
note the size of the set of surplus nodes for UNL
i
ICISSP 2020 - 6th International Conference on Information Systems Security and Privacy
60
Figure 2: µ
i
and µ
j
are the set sizes of surplus nodes in
UNL
i
and UNL
j
.
and UNL
j
, respectively, that is the nodes in U NL
i
that are not in common with UNL
j
and viceversa:
µ
i
= n
i
O
i, j
, µ
j
= n
j
O
i, j
(Fig. 2). Moreover, we
denote by t
s
the safety fault tolerance and by t
l
the
liveness tolerance of the system, i.e. the maximum
number of Byzantine nodes the XRP LCP tolerates in
order to guarantee safety and liveness, respectively.
Given a pair of nodes i, j with their respective
UNLs, and assuming that n
i
< n
j
and both UNLs
have the same percentage of faulty nodes, t
i, j
=
min{t
i
,t
j
,O
i, j
} = t
i
. By substituting O
i, j
with n
i
µ
i
in Eq. 16, we get the following inequality:
t
s
<
n
i
2
µ
i
2
µ
j
2
+ q
v
. (17)
On the other hand, the liveness tolerance t
l
is given
by:
t
l
< n
i
q
v
. (18)
In general, a consensus protocol providing results that
can be relied upon is preferable, rather than one that
is able to progress in the presence of faulty nodes but,
at the same time, reports impaired results that could
undermine consistency. As a result, in order to obtain
a validation quorum q
v
whose safety fault tolerance t
s
meets or exceeds the liveness tolerance t
l
, we can use
t
s
n
i
q
v
and get the following:
n
i
q
v
<
n
i
2
µ
i
2
µ
j
2
+ q
v
q
v
>
3n
i
4
+
µ
i
4
+
µ
j
4
. (19)
4.2.1 Unique UNL
Let us now consider the simplest case where the
XRP Ledger network consists of a single agreed-upon
UNL. In this case, n
i
= n
j
= n, and, consequently,
µ
i
= µ
j
= 0. By Eq. 19, we obtain q
v
> (3/4)n,
meaning that when there is 100% agreement on par-
ticipants, reaching a 75% validation quorum is suffi-
cient to fully validate a ledger. Therefore, in Eq. 16
we have n > n/2 + n (3/4)n + t
i
, from which we
obtain t
s
(= t
i
) < (1/4)n, whereas from Eq. 18 we
obtain t
l
< n(3/4)n = (1/4)n. As a result, in case
UNL
i
= UNL
j
and q
v
= 0.75n, both the tolerances are
0.25n. This means that when less than 25% of trusted
nodes are Byzantine, the protocol functions properly.
Keeping valid the assumption U NL
i
=UNL
j
, now
we show how t
s
and t
l
vary when q
v
is equal to 0.8n,
as required by the current XRP LCP specification. In
this case, the value of the two tolerances are no longer
equal: t
s
< (3/10)n, and t
l
< (1/5)n. Compared to the
previous case in which q
v
= 0.75n, here the liveness
tolerance is lower than the safety fault tolerance, and
this implies that the system ability to not get stuck is
slightly weakened.
4.2.2 Overlapping UNLs
So far the analysis has focused on the circumstance
where the network is composed of a unique UNL.
However, the validation quorum increase takes on
greater significance in the context in which the UNLs
of any two nodes i, j do not completely overlap, i.e.
when at least one of the two parameters µ
i
and µ
j
is
> 0. In this regard, we now turn our attention to the
total non-overlapping component resulting from the
sum of the two individual surpluses µ
i
and µ
j
of Eq.
17:
µ
i
+ µ
j
< n + 2q
v
2t
s
. (20)
Let us now consider two cases in which we set the
validation quorum first to 80% and then to 90% of the
nodes. In the former case, q
v
= 0.8n and assuming
t
s
= 0.2n, we can safely allow up to a non-overlap
of 0.2n. In the latter case, q
v
= 0.9n and assuming
t
s
= 0.1n, the maximum non-overlap we can safely
allow is 0.6n. Thus, as the above bounds show, as q
v
increases, the degree of freedom of any node in the
choice of validators to trust, in turn, increases.
To conclude, if we want the system to have a little
more liveness, we can increase the sizes of the sur-
plus node sets µ
i
and µ
j
. Recalling that t
l
depends on
the quantity n
i
q
v
, as the values µ
i
and µ
j
increase,
we obtain a higher validation quorum (cf. Eq. 19),
and hence the maximum number of allowed Byzan-
tine nodes to guarantee liveness increases. Let us con-
sider an overlap of 90%, i.e. a non-overlap of 10%. If
the safety fault tolerance is 0.2n
i
, from Eq. 16 we ob-
tain q
v
> 0.75n
i
, and t
l
< 0.25n
i
. In contrast, if the
safety fault tolerance is 0.25n
i
, we obtain q
v
> 0.8n
i
,
and hence, t
l
< 0.2n
i
. From this analysis, it emerges
that safety and liveness tolerances, validation quorum,
and UNLs overlapping set size are strictly correlated,
and it is possible to tune these parameters according
to the desired properties the system needs to satisfy.
Currently, if no configuration changes are made,
each node adopts the default and recommended UNL
provided by Ripple. This essentially implies that no
disagreement on the participants in the network is al-
lowed, since all the nodes listen to a single list of val-
idators. Accordingly, the XRP LCP is really able to
A Formal Approach for the Analysis of the XRP Ledger Consensus Protocol
61
guarantee that the network cannot get stuck as long as
the number of Byzantine nodes within the system is
limited.
5 CONCLUSIONS
Understanding the fundamental mechanisms and
security properties of the structure underlying
blockchain-based protocols is becoming increasingly
important. Thus, detailed analyses and formalizations
of existing distributed ledger systems are crucial to
prove the correctness of the algorithms and gain con-
fidence that they achieve their goals.
In this paper, we have presented a formalization
of the XRP Ledger Consensus Protocol, whose func-
tioning has been studied in a rather superficial way
so far. To the best of our knowledge, our work is the
first one that describes the phases of the Ripple pro-
tocol in such great detail, trying to avoid ambiguities
in defining its behavior. We have included the analy-
sis of two key security properties, namely safety and
liveness, based upon which the efficacy of the proto-
col can be determined. Furthermore, we have shown
that the correlation between some protocol parame-
ters can be leveraged to meet a desired liveness/fault
tolerance. Our work represents the first step towards
the complete analysis of the XRP Ledger Consensus
Protocol and opens several research directions. For
instance, our set-theoretic approach could serve as the
basis for the construction of an extended formalism
that is capable of capturing the behaviour of the pro-
tocol by means of a strictly formal language. Also
the use of different model-driven approaches and au-
tomated verification tools can greatly benefit from our
specification efforts and facilitate the comprehensive
description of the protocol. For the immediate future,
we plan to extend our work considering the formal-
ization of additional security properties and using dif-
ferent models of computation.
REFERENCES
Armknecht, F., Karame, G. O., Mandal, A., Youssef, F.,
and Zenner, E. (2015). Ripple: Overview and Out-
look. In Conti, M., Schunter, M., and Askoxylakis,
I., editors, Trust and Trustworthy Computing, pages
163–180, Cham. Springer International Publishing.
Badertscher, C., Maurer, U., Tschudi, D., and Zikas, V.
(2017). Bitcoin as a Transaction Ledger: A Compos-
able Treatment. In Katz, J. and Shacham, H., editors,
Advances in Cryptology – CRYPTO 2017, pages 324–
356, Cham. Springer International Publishing.
Bano, S., Sonnino, A., Al-Bassam, M., Azouvi, S., Mc-
Corry, P., Meiklejohn, S., and Danezis, G. (2017).
Consensus in the Age of Blockchains. CoRR,
abs/1711.03936.
Cachin, C. and Vukolic, M. (2017). Blockchain Consensus
Protocols in the Wild. CoRR, abs/1707.01873.
Chase, B. and MacBrough, E. (2018). Analysis of the XRP
Ledger Consensus Protocol. CoRR, abs/1802.07242.
Daian, P., Pass, R., and Shi, E. (2016). Snow White:
Provably Secure Proofs of Stake. Cryptology ePrint
Archive, Report 2016/919.
Fischer, M. J., Lynch, N. A., and Paterson, M. S. (1985).
Impossibility of Distributed Consensus with One
Faulty Process. J. ACM, 32(2):374–382.
Garay, J., Kiayias, A., and Leonardos, N. (2015). The Bit-
coin Backbone Protocol: Analysis and Applications.
In Oswald, E. and Fischlin, M., editors, Advances
in Cryptology - EUROCRYPT 2015, pages 281–310,
Berlin, Heidelberg. Springer Berlin Heidelberg.
Garay, J., Kiayias, A., and Leonardos, N. (2017). The Bit-
coin Backbone Protocol with Chains of Variable Dif-
ficulty. In Katz, J. and Shacham, H., editors, Ad-
vances in Cryptology CRYPTO 2017, pages 291–
323, Cham. Springer International Publishing.
Garay, J. A., Kiayias, A., Leonardos, N., and Panagiotakos,
G. (2018). Bootstrapping the Blockchain, with Appli-
cations to Consensus and Fast PKI Setup. In Abdalla,
M. and Dahab, R., editors, Public-Key Cryptography
PKC 2018, pages 465–495, Cham. Springer Interna-
tional Publishing.
Gramoli, V. (2017). From Blockchain Consensus Back to
Byzantine Consensus. Future Generation Computer
Systems.
Halpin, H. and Piekarska, M. (2017). Introduction to Secu-
rity and Privacy on the Blockchain. EuroS&P 2017 -
2nd IEEE European Symposium on Security and Pri-
vacy, Workshops.
Kiayias, A., Russell, A., David, B., and Oliynykov, R.
(2017). Ouroboros: A Provably Secure Proof-of-
Stake Blockchain Protocol. In Katz, J. and Shacham,
H., editors, Advances in Cryptology – CRYPTO 2017,
pages 357–388, Cham. Springer International Pub-
lishing.
Mauri, L., Cimato, S., and Damiani, E. (2018). A Compar-
ative Analysis of Current Cryptocurrencies. In Pro-
ceedings of the 4th International Conference on In-
formation Systems Security and Privacy - Volume 1:
ICISSP, pages 127–138. INSTICC, SciTePress.
Nakamoto, S. (2008). Bitcoin: A Peer-to-Peer Electronic
Cash System.
Pass, R., Seeman, L., and Shelat, A. (2017). Analysis of
the Blockchain Protocol in Asynchronous Networks.
In Coron, J.-S. and Nielsen, J. B., editors, Advances
in Cryptology EUROCRYPT 2017, pages 643–673,
Cham. Springer International Publishing.
Pass, R. and Shi, E. (2017). The Sleepy Model of Consen-
sus. In Takagi, T. and Peyrin, T., editors, Advances
in Cryptology ASIACRYPT 2017, pages 380–409,
Cham. Springer International Publishing.
Pease, M., Shostak, R., and Lamport, L. (1980). Reach-
ICISSP 2020 - 6th International Conference on Information Systems Security and Privacy
62
ing Agreement in the Presence of Faults. J. ACM,
27(2):228–234.
Rawat, D. B., Chaudhary, V., and Doku, R. (2019).
Blockchain: Emerging Applications and Use Cases.
CoRR, abs/1904.12247.
Ripple Labs Inc. (a). Ripple. https://ripple.com/. Last
checked on Oct, 2019.
Ripple Labs Inc. (b). XRP Ledger Dev Portal.
https://xrpl.org/index.html. Last checked on Oct,
2019.
Ripple Labs Inc. (c). Ripple Source, GitHub repository.
https://github.com/ripple/rippled/tree/develop/src/
ripple. Last checked on Oct, 2019.
Saad, M., Spaulding, J., Njilla, L., Kamhoua, C. A., Shetty,
S., Nyang, D., and Mohaisen, A. (2019). Explor-
ing the Attack Surface of Blockchain: A Systematic
Overview. CoRR, abs/1904.03487.
Schwartz, D., Youngs, N., and Britto, A. (2014). The Rip-
ple Protocol Consensus Algorithm. Ripple Labs Inc.
White Paper.
Vukoli
´
c, M. (2016). The Quest for Scalable Blockchain
Fabric: Proof-of-work vs. BFT Replication. In Ca-
menisch, J. and Kesdo
˘
gan, D., editors, Open Problems
in Network Security, pages 112–125, Cham. Springer
International Publishing.
Wang, W., Hoang, D. T., Xiong, Z., Niyato, D., Wang, P.,
Hu, P., and Wen, Y. (2018). A Survey on Consensus
Mechanisms and Mining Management in Blockchain
networks. CoRR, abs/1805.02707.
Wilson, B. (2018). Raise quorum / increase fault tol-
erance. https://github.com/ripple/rippled/issues/2604.
Last checked on Oct, 2019.
Xiao, Y., Zhang, N., Lou, W., and Hou, Y. T. (2019).
A Survey of Distributed Consensus Protocols for
Blockchain Networks. CoRR, abs/1904.04098.
A Formal Approach for the Analysis of the XRP Ledger Consensus Protocol
63