Blockchain is a distributed, transactional database that is shared across all the nodes participating in the network. This is the main technical innovation of Bitcoin and it acts as a public ledger for the transactions. However, this technology lacks standardisation and uniform understanding. This is due to a few studies, that would provide a comprehensive model of the blockchain and the distributed ledger technology. In this paper we compare four blockchain technology platforms and focus on their business level properties including actors and roles, services, and processes and data model. Our comparison results in a reference model, which could potentially guide the business analysts, system analysts and software developers when developing new blockchain platforms or their supported implementations. Accuracy of the proposed reference model is validated by considering it against selected blockchain technologies.
The rst implementation of the blockchain technology, i.e. Bitcoin, was
introduced in 2009 [
Di erent businesses have developed various implementations using the
blockchains. However, only limited analysis [
In Section 2 we give an overview of the state of the art of the blockchain technology. Based on it, in Section 3 we present the reference model for the blockchain technology. Section 4 describes how the accuracy of our proposal is validated. Finally, Section 5 concludes the study and presents some future work. 2
In this section, rst we will de ne scope and discuss study background. Next we will survey the blockchain technologies following the scoped properties. 2.1
Blockchain acts as a distributed public ledger. It is a digital record of transactions
and ownership, that is replicated among all of the participants of a peer-to-peer
network. A consensus algorithm ensures that each node owns the same copy
of the ledger as the other nodes. Technically, it is a back-linked ordered list of
blocks, where each block contains transactions [
There are two main types of blockchains: public and private. Bitcoin has a
public ledger i.e., a public blockchain, where anyone is allowed to contribute [
Blockchain technology platforms can be separated into four groups [
They support transferring value from one account to another. Blockchain technology relies on a decentralised network of individual nodes, but nodes have di erent purposes and di erent roles. Table 2 shows an overview of actors who are present in the analysed blockchain platforms.
For each platform, there exists a notion of a Client and a Miner or someone
who builds and agrees upon which transactions are included in a block [
Miners deal with validating transactions and building new blocks. Since Chain Core is a private blockchain, it has Blockchain operators, who are either block generators or signers. Fundamentally, a Blockchain operator is a miner, because miner's tasks in a blockchain environment are to create new blocks, sign them, validate them and submit them to the blockchain.
In conclusion, blockchain technology has two primary actors at the business level: { Human actor who interacts with the blockchain by creating transactions.
This actor can be called a \User". { Human or system actor responsible for veri cation and validation of transactions, building new blocks, signing new blocks and publishing new blocks to the blockchain. This actor supports trust between the parties involved. In case of public blockchains \proof of work" is provided by the mining software (system), but, for example in Chain Core there exist Blockchain operators (human) who decide on the consensus. This actor can be labelled as \Block generator". 2.3
In this section we will compare services used by actors de ned in the previous section. The services are summarized in Table 3.
Firstly, every platform provides a service to create and broadcast transactions to the network. This is an essential service because transactions dictate the state of the blockchain and add new data to the blockchain.
MultiChain and Chain Core have the notion of assets, which is a type of
value, that is issued on the blockchain [
Bitcoin and MultiChain focus on transactions and exchanging value. Ethereum
and Chain Core also rely on state and smart contracts. Ethereum provides a
service to create a new contract, that can be submitted to the network; and Chain
Core supports the use of smart contract while issuing assets, by de ning business
3 Bitcoin (with upper B) stands for protocol, the software and community, bitcoin
(with lower b) stands for a unit of currency
rules for issuing (issuance program [
Since MultiChain and Chain Core are both designed to support private
blockchains, they support services to manage permissions. MultiChain provides
services for granting and revoking permissions to and from speci c users [
All platforms except Chain Core are proof-of-work-based mining solutions.
Mining in Bitcoin and Ethereum is publically available for anyone, while in
MultiChain user needs to have permission to perform mining (or if everyone on
the blockchain are known users, mining can be turned o alltogether [
In conclusion, common services among the technologies are creating transactions, validating blocks and mining / creating blocks. Additionally, permissioned blockchains provide services to manage permissions. Overall, it depends on the features o ered by a blockchain, e.g., with Bitcoin being the most generic blockchain, the number of provided services is di erent compared to Chain or MultiChain. Features like assets, smart contracts and permissions add additional services to the commonly o ered ones. 2.4
Every platform has a network discovery process, which consists of 4 main steps - Peer discovery ( nding peers to connect to, either user already knows the IPs or acquires them), Handshake (version check, establishing connection, providing ownership of private key), Network discovery ( nding neighbouring peers and letting the network know that a new node has connected) and Synchronization (downloading the latest block data from the network).
In Bitcoin and Ethereum, new node connects to a known peer, they verify
that both are running the same version of the software and have the same, latest
and longest chain of blocks. In case of di erences, new node will download the
previous blocks up to the latest one. MultiChain expands the Handshake process
[
Creating the transaction in Bitcoin requires user to enter value and the
receivers address. The transaction is then signed by the user and broadcasted to
the network (to neighbouring nodes). The neighbouring nodes check the
transaction for validity. If valid, they will propagate it forward to other peers [
Another common process is the Mining process (Block generation
process). Mining is based on the proof-of-work. A miner will build a new block,
add collected unveri ed transactions and metadata, and calculate the
computationally exhaustive proof-of-work for the block. If he is the rst one to solve the
task and mine the block, he can submit the block to the network and receive
a reward for that work. In Ethereum, the mining process also requires a state
transition process, since it keeps a state of the blockchain. In the state transition
process, transactions are validated and in case of contracts, code execution is also
performed. When all the state transition functions are valid, miner in Ethereum
will provide the proof-of-work for the block. In MultiChain, proof-of-work [
Chain Core introduces a Chain consensus process [
Bitcoin and Ethereum introduce the block validation process which is
performed by every node once the miner broadcasts a new block. Since in the
public blockchains the miners are anonymous, there has to be a guarantee that
the miner has indeed produced a valid block. In Bitcoin, this is called a
`consensus' if all the nodes validate the new blocks against the same rules. In Ethereum,
the validation is similar, but additionally it includes the state transition process,
which each node has to perform before accepting the block [
The four platforms provide similar general processes: Network discovery, Transaction creation, Block generation and submission (Mining and Chain consensus process) and Block validation. Conceptually, the processes may be similar, but the inner work ow di ers from one technology to another. 2.5
All four blockchain technologies introduce a Block, a Block header and
Transactions. Bitcoin, MultiChain and Chain Core also include Transaction Inputs and
Transaction Outputs (UTXO). Ethereum relies on the state replication, where
each new block's state is the outcome of the transactions that were included in
the block. Ethereum has chosen not to use the Input-Output transaction method,
because it does not support multi-stage contracts or scripts that could keep an
internal state [
Ethereum has the notion of Accounts, which are either user accounts or virtual contract accounts, that hold the balance, contract code and internal storage. In Ethereum case, all this data is stored on the blockchain.
With the addition of Assets, Chain Core keeps an Asset entity containing only the asset ID. The assets are tied with certain programs (i.e. Issuance program, for issuing new assets, or Control program, for spending assets). The Consensus program is used by Block generator to verify that a block is ready to be submitted to the network.
To conclude, the main set of entities are the Block, Block Header and
Transactions. The Input-Output transactions are de facto Bitcoin solution to prevent
double-spending, but there are several arguments about the use of unspent
transaction outputs and their scalability [
4 http://www.opengroup.org/subjectareas/enterprise/archimate-overview
{ Transaction creation - Transactions allow Users to add information to the
blockchain. Transactions can be used to create assets, spend assets, create
smart contracts, call functions on smart contracts, manage permissions etc.
{ Transaction submission - Transaction submission support signing and
broadcasting the transaction to the network.
{ Block validation - Block validation is a general service used by the nodes
to validate the newest blocks that have been added to the blockchain.
{ Block generation - Block generation is broken down into smaller services
that are used di erently depending on the type of consensus. A miner in
a public blockchain would use the block generation services, but in Chain
Core, one party creates the block and other parties sign the block [
The reference model (Figure 1) includes four processes: Network discovery process, Transaction process, Consensus process, and Block generation process. We expand these processes using the BPMN modelling language.5
Network discovery process (Figure 2) consists of four subprocesses - Peer Discovery, Handshake, Discovering additional peers and Synchronization. Process begins by acquiring an IP of the known peer or blockchain (IP is known from the last session or is acquired from an outside source). Once connected, private blockchains will perform a check if the user is allowed to connect via its public IP or a network token. If permitted, a handshake process is carried out to verify versions and check that both nodes have the same blockchain with the latest blocks. Next, the connected node's IP will be propagated to the network. Once the network is discovered the connecting node will synchronise its blockchain if any di erences in terms of the latest block is observed.
Transaction process (Figure 3) starts creation of a new transaction. Relevant metadata are added depending on the transaction type (e.g., standard transfer of funds, creating assets, deploying smart contract, etc.). In the case of private blockchains, there will be a permission check: (i ) whether the speci c user is permitted to create the transaction, (ii ) whether the receiver is permitted to receive funds, (iii ) whether the network permits transfer of funds or creation of speci c type of transactions, etc. If the transaction is permitted (or in case of public blockchains, the creator signs the transaction), it will be broadcasted to the network, i.e., to the neighbouring nodes.
Consensus process (Figure 4) is performed by each node when a new block has been broadcasted. Each blockchain de nes its own \Consensus rules", according to the type of the validated blocks and transactions. If validated, the nodes will append the new block to the blockchain; otherwise it will be rejected.
During the Block generation process (Figure 5), rst, a new block is
created; next, the previous block's metadata is added. In case of the
permissioned blockchains, permission changes have to be applied in order to avoid
5 http://www.bpmn.org/
unwanted actions performed by users, who do not have permission. Next, all
transactions will be validated against the consensus rules on the blockchain. In
public blockchains, the block generators (i.e., miners) will provide a proof of the
work (i.e., proof-of-work or proof-of-stake etc.) and will be rewarded for their
work. In private blockchains, the consensus process might be part of the block
generation process. Once the block is constructed, it is submitted to the network
and propagated to all the nodes.
The data model is presented in Figure 6. The model shows that keeping the
state of the blockchain is preferred here in comparison to the Input-Output
type transaction logic. Since UTXOs are stateless [
State contains accounts, which have balance and address. In case of contracts (marked with grey box in Figure 6), they also have to keep the executable code and storage speci c to the given account. Accounts are linked to transactions and ko r w t e N Create new
block Add previous block metadata
Unverified transactions
Col ect unverified transactions
Apply permission changes each transaction changes the state of the blockchain (balance changes, contracts function calls etc.). Block and Block Header are standard and essential to every blockchain platform. Block contains transactions; the block's metadata is kept in the Block Header. 4
To validate the reference model we investigate its accuracy by comparing it to the existing blockchain solutions. Firstly, we compare the reference model to the four implementations that were used to build it. We de ne a Delta (i.e.
) metric, which represents the di erenc between the reference model and the models of the considered blockchain technologies. Secondly, we select four new blockchain platforms and calculate the Delta value for them. Our goal is to show that using the reference model we are able to capture business perspective of all selected blockchain technologies.
Delta de nition. Finding the delta for the four initial Blockchain technologies (Figure 7), we will compare actors, services, processes and data models separately. The overall will be the sum of all deltas ( = actors + services + processes + datamodels). = 0 means they have the same entities. > 0 means the uni ed model has more entities. < 0 means that the reference model is missing some entities, that the comparable model has. Finding the delta for the four new Blockchain technologies (Figure 7), we will also compare actors, services, processes and data models separately, but we will only take into evaluation if the given entity exists or is present in the newly presented technology.
= 0 means they have the same entities. > 0 means the reference model has more entities.
Results for the values are presented in Table 5. The goal of accuracy validation was to get a value as close to 0 as possible for initial four technologies as well as for four new technologies that were not used as a basis of building the model. From the results we can see that the initial four technologies are covered almost accurately by the reference model, with subtle di erences in Data models (e.g., four di erences - Bitcoin and MultiChain do not support Accounts and State). We consider this result acceptable, because of the di erences that smart contracts introduce to the data model (see Figure 6).
As for the four technologies that were not part of the initial building of the model, we found di erences in the data models (e.g., ve entity di erences CryptoNote and NXT do not keep the blockchain state, Hyperledger Fabric does not have Block Header, Tendermint misses account representation), Networking process (one entity di erence - Block synchronization) and Consensus process (e.g., one entity di erences - Transactions in Hyperledger Fabric are veri ed and endorsed before reaching the miner). This result is also considered acceptable, due to the fact that the new blockchain technologies are di erent from the initial four technologies, but still perform well under the de ned blockchain properties.
Threats to validity. The accuracy validation was done by the rst author of this paper with the technologies in hand. Thus, another validation approach might conceive di erent results, either by de ning the Delta di erently or by selecting di erent implementations of the technology. We did not constructed conceptual models for platforms which were not used to create the reference model. These were assessed following their documentations. Potentially we could miss some concepts from the comparison. However this is less likely as the majority of the entities were in fact captured as shown in Table 5. 5
This paper gives an overview of the disruptive blockchain platforms: Bitcoin, MultiChain, Ethereum and Chain Core. Each of them presented a di erent approach to networking, transactions, mining, validation, security and permissions. The comparison resulted in the reference model, that aims to represent the domain of the blockchain technology. The model contributes to the explicit understanding of the technology and its work processes. The paper also presents a brief overview of validation performed on model with the corresponding results. The results re ect that the reference model performed well to cover the known blockchain technologies, such as Cryptonote, NXT, Hyperledger and Tendermint. The results potentially indicate how to develop this fast growing technology further and in a more secure way by expanding its disruptive nature to other application domains.
When it comes to standardisation of the blockchain technology and its
presentation it in an uni ed way, there exist some study that thrive towards this
goal. In [
This type of research on the blockchain technology is a rst and hopefully a
basis to future research. This being said, we plan to further validate the model
via security assessment [