Access 2019 2961372

This article has been accepted for publication in a future issue of this journal, but has not been
fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2019.2961372, IEEE Access
Date of publication xxxx 00, 0000, date of current version xxxx 00, 0000.
Digital Object Identifier 10.1109/ACCESS.2017.DOI
Machine Learning adoption in

Blockchain-based Smart Applications:
The challenges, and a way forward
SUDEEP TANWAR1 , QASIM BHATIA1 , PRUTHVI PATEL1 , APARNA KUMARI1 , PRADEEP
KUMAR SINGH2 , WEI-CHIANG HONG3 , SENIOR MEMBER, IEEE
1
Department of Computer Science and Engineering, Institute of Technology, Nirma University, Ahmedabad, Gujarat, India (e-mails:
[email protected], [email protected], [email protected], [email protected])
2
Department of Computer Science & Engineering, Jaypee University of Information Technology, Waknaghat, Solan (H.P.) India (e-mail:
[email protected])
3
Department of Information Management, Oriental Institute of Technology, Panchiao, New Taipei, Taiwan. (email: [email protected])
Corresponding author: Wei-Chiang Hong (e-mail: [email protected]).
ABSTRACT In recent years, the emergence of blockchain technology (BT) has become a unique, most
disruptive, and trending technology. The decentralized database in BT emphasizes data security and privacy.
Also, the consensus mechanism in it makes sure that data is secured and legitimate. Still, it raises new
security issues such as majority attack and double-spending. To handle the aforementioned issues, data
analytics is required on blockchain based secure data. Analytics on these data raises the importance of arisen
technology Machine Learning (ML). ML involves the rational amount of data to make precise decisions.
Data reliability and its sharing are very crucial in ML to improve the accuracy of results. The combination
of these two technologies (ML and BT) can provide highly precise results. In this paper, we present a
detailed study on ML adoption for making BT-based smart applications more resilient against attacks.
There are various traditional ML techniques, for instance, Support Vector Machines (SVM), clustering,
bagging, and Deep Learning (DL) algorithms such as Convolutional Neural Network (CNN) and Long
short-term memory (LSTM) can be used to analyses the attacks on a blockchain-based network. Further, we
include how both the technologies can be combined in several smart applications such as Unmanned Aerial
Vehicle (UAV), Smart Grid (SG), healthcare, and smart cities. Then, future research issues and challenges
are explored. At last, a case study is presented with a detailed conclusion.
INDEX TERMS Blockchain, Machine Learning, Smart Grid, Data Security and Privacy, Data Analytics,
Smart Applications
I. INTRODUCTION its communication also raises security concerns. To handle

ROM the past few decades, data has become an essential these issues, BT can be used, which has a distributed database
F source of intelligence and carries new opportunities
to the real-life problems such as wireless communications,
network. It was coined by Satoshi Nakamoto in the year 2008
and contained a time-stamped series of tamper-proof records,
bioinformatics [1], agriculture [2], and finance [3] through which are managed by a cluster of distributed computers.
smart applications. These applications are data-driven and It comprises of a chain of blocks that are connected using
incorporate actionable insights into user experience, which cryptographic primitives. The three mainstays of BT are
enables individuals to complete the desired task more effi- immutability, decentralization, and transparency. These three
ciently. It operationalizes insights, personalizes the customer characteristics opened its door for a wide variety of applica-
experience, optimizes customer interactions, improves options, for example, digital currency existence (currency with
erational efficiency, and enable new business model. There no physical existence) and analysis on the suitability of it
are various smart applications such as SG, UAV, Smart in smart applications [4]. Although BT ensures security and
Cities, which makes the life of an individual easier. These privacy issues, some vulnerabilities also started appearing
applications generate a huge amount of data, and storage after its implementation. For instance, the nature of attacks
of this ever-evolving data in databases is a problem, and began to be increasingly sophisticated such as majority
VOLUME 4, 2016 1
This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
Tanwar et al.: Machine Learning adoption in Blockchain-based Smart Applications: The challenges, and a way forward
attacks (51% attack) that control voting, Sybil attacks for fake blockchain. The proposed survey covers all the fundamental
identity generation to control the consensus [5]. To handle the aspects of ML to be applied BT based applications, for
aforementioned issue, a robust Intrusion Detection System instance, intrusion detection. A review conducted by Meng
(IDS) is required in place because the traditional methods use et al. [12] describes how blockchain help to meet intrusion
a signature-based approach to detect specific patterns. But, detection. The authors mitigate trust issues by establishing a
to detect intrusions and attack patterns, one of the emerging collaborative IDS using smart contracts [17]. Further, Conti
technology known as ML can be used to analyze the data et al. [13] surveyed on security and privacy issues of bitcoin.
traffic. Thus, designing efficient and effective algorithms They discuss various categories of attacks, such as double-
to analyze this massive amount of data is in dire need of spending attacks, client-side security threats, and mining
handling the blockchain-based smart applications. Hence, pool attacks. To handle the aforementioned issue, Rahouti et
ML is highly prevalent today and uses a dozen times a al. [5] discuss specific ML-based solutions and discuss the
day without even realizing it. ML encompasses computers handling of specific social issues such as human trafficking
to study, think, and act without intervention of humans. and drug sales through cryptocurrency. Then, Ucci et al [14]
It is considered to be one of the applications of Artificial explored malware analysis using ML techniques. Features of
Intelligence (AI). ML provides computers the competency to malware were thoroughly discussed, and a detailed taxonomy
learn without being programmed it explicitly. Its basic idea has been proposed. Salah et al. [15] and Casino et al. [16]
is to build an efficient algorithm that can accept input data conducted a review on blockchain-based applications. To
and, with the help of statistical analysis, make a prediction, clarify the main difference between other surveys paper and
and update the outputs. A substantial amount of data can be this survey paper, a comprehensive comparison has been
analyzed by ML to create data-driven decisions. shown in Table 1. It includes objectives, merits, and demerits
In a communication network of blockchain-based smart of peer surveys concerning numerous parameters such as
applications, there is layer-wise handling of security issues. architecture, applications, open issues, taxonomy, and merits,
Some security issues are handled at the network layer, such as demerits of the existing approaches. A master taxonomy of
malicious packets and some at the application layer such as ML for BT is summarized in Figure 3.
malware [6]. At the network layer, malicious packets can be
used to impose the network to establish fraudulent consensus. B. RESEARCH CONTRIBUTIONS
A naive solution to this problem can be to use a firewall
Though several research works exist to address the various
to ensure that packets meet pre-defined security criteria [7].
ML usages for the blockchain-based system but It has not
Though, the attacks are becoming more sophisticated with
been exploited to its full potential. In this paper, we investi-
unseen patterns to bypass a firewall. To prevent this issue,
gated the ML usages for BT. The following are the research
packets header data can be analyzed using ML models [8] in
contribution of this paper.
real-time using historical data. This analysis helps to detect
new and changing patterns. Similarly, ML techniques can • A brief discussion on how ML and blockchain can be
be used to classify malware to end-point such as servers, used together with a proposed architecture diagram.
mobile, or workstations. Further, several blockchain-based • To develop taxonomy covering ML techniques required
smart applications such as UAV [9], Data Trading [10], SG for BT based environment. In each part of the taxonomy,
build trust between data exchangers [11]. It is very crucial existing work has been discussed in detailed to handle
in any smart application at the same time; data should be several issues, such as preventing and predicting attacks
secure. BT ensures data security but to build confidence, on the blockchain network.
and ML techniques are used to predict untrustworthy nodes • A case study is presented to demonstrate the ML tech-
based on past patterns. Similarly, UAVs have significantly niques for blockchain-based smart applications, for ex-
different network topology compared to the conventional ample, SG.
blockchain network topology [9]. It includes communication
using satellites and various ground stations. For UAV, BT is C. ORGANIZATION
used to securely store coordinates and other relevant data to
maintaining graph integrity for the vehicles. In subsequent The rest of the paper organized as follows: Section 2 provides
sections, we explore the recent research work on ML adop- a background of Blockchain, ML, and list all benefits while
tion in the blockchain-based smart application. applying ML for BT based application. Then, we proposed an
architecture. Survey procedures and taxonomy presented in
Section 3. Then, in Section 4-7, we present existing advance-
A. SCOPE OF THIS SURVEY ments in ML approaches for BT-based smart applications.
Several survey work has been published till date by the A discussion on future research issues and challenges is
different researchers on the various aspects of the adoption presented in Section 8. Further, section 9 covers a case study
of ML in blockchain-based smart applications [5]. As per on the SG system, and finally, we concluded the paper.
our knowledge, most of the surveys have focused on specific
areas, fields, or applications where it requires both ML and II. BACKGROUND
2 VOLUME 4, 2016
TABLE 1: Comparison of the existing surveys with the proposed survey.

Authors Year Objectives of Survey Merits Demerits 1 2 3 4 5
Meng et al. [12] 2018 To present use of blockchain in in- Scope of application of blockchain Discusses only data sharing and X X X X X
trusion detection was discussed trust management issues of collab-
orative intrusion detection
Conti et al. [13] 2018 To discuss various security and pri-
A Comprehensive review of possi- Blockchain issues are not high- X X X X X
vacy issues in bitcoin ble attacks on bitcoin and provided lighted
countermeasures
Rahouti et al. [5] 2018 Survey on ML security solutions for In-depth and wide classification of Other applications of blockchain X X X X X
bitcoin major threats and extensive expla- are missing
nation of the role of ML
Ucci et al. [14] 2018 To study ML techniques for mal- Time and space complexity for var- Lacks discussion on uses of these X X X X X
ware analysis ious methodologies has been de- techniques in a blockchain environ-
scribed in detail ment
Salah et al. [15] 2019 Discuss applications, platforms, The decentralization feature of Discussion on privacy is not cov- X X X X X
and protocols in blockchain blockchain is explained with a spe- ered in detail
specifically for AI cific view of AI
Casino et al. [16] 2019 Review blockchain-based applica- Prerequisites for blockchain appli- Focused on applications, not the X X X X X
tions and identify open issues cations are thoroughly discussed open issues
Proposed Survey - To survey how ML can be used Discusses architecture and technol- - X X X X X
in blockchain-based smart applica- ogy at a fundamental level and
tions bridges the gap between two tech-
nology
Parameters- 1:Architecture, 2:Application, 3:Open Issues and Challenges, 4:Taxonomy, 5:Security
Notations- X: considered, and X: not considered.
A. MACHINE LEARNING variables, following which it is used to make best guess

ML is the field of study that focuses on building applications predictions for the unlabeled data. It feeds that predicted data
that learn through experience. It is the ability to teach a back into the supervised ML algorithm as training data and
computer without programming it explicitly [18]. ML en- use the model to make predictions on unseen data.
compasses its work from a diverse set of disciplines, includ-
ing philosophy, information theory, probability and statistics, B. BLOCKCHAIN
control theory, psychology and neurobiology, computational
complexity, and artificial intelligence [19]. ML algorithms Blockchains are an immutable set of records that are cryp-
are used in many application and benefited it as listed below: tographically linked together for audit [23]. It is similar to
an accounting ledger. Here, previous records in accounting
• In Data mining, large databases contain different pat- ledger cannot be changed, and new records need to be veri-
terns that can be discovered automatically by using ML fied by a trusted party. The only difference between these two
techniques to analyze outcomes, for instance, medical is that new blocks (set of records) checked by a decentralized
treatments of a patient from health record databases structure of nodes that have a copy of the ledger. There
or to identify the creditworthiness of a person from is no centralized party to verify the records. Blockchain is
financial databases. formed by linking valid blocks together; the current block
• ML applies in areas where a deterministic algorithm contains the hash of the previous block, and so on, as shown
is not promising, such as human face recognition from in FIGURE 1. This makes blockchain traceable and resistant
images. to change [24]. Older blocks cannot be modified, in case
• Application domains where the adaptable programming they are changed in any way; their hash would change. This
is required, for instance, controlling manufacturing pro- emphasis to link hash in all subsequent blocks to make the
cesses as per the demand of the customer and adapting blockchain network valid again. A copy of the blockchain is
to the varying reading interests of readers. available with every individual within the network; hence-
ML algorithm eare application specific and depends on the forth, any changes can be cross verified by the other users.
output required by the system. There are several ML algo- These copies of the blockchain are updated with the addition
rithms, such as Supervised ML, Semi-Supervised ML, and of a new block. Then, everyone can see the block, depending
Unsupervised ML. (i) Supervised ML uses statistical models on the permissions assigned by the administrator. BT uses
to predict output in numerical data and classify the correct la- a cryptographic secure hash algorithm (SHA) such as SHA-
bel [20]. Here, the most commonly known algorithms include 256 and SHA-512 to maintain the data integrity within the
the regression approach and decision trees. (ii) Unsupervised block. Each block has a unique hash value. For instance,
ML does not have label data. Here, data samples are grouped Ethereum uses Keccak-256 and Keccak-512, while Bitcoin
into clusters depending on their similarity or dissimilarity uses double SHA-256. This SHA is a collision-resistant al-
[21] using a different approach. For example, K means clus- gorithm, where no two different input data could produce the
tering and association rules algorithms. (iii) Semi-Supervised same output (hash value). Henceforward SHA can be used to
ML is also a category of interest [22]. It involves a mixture of check if the data is the same or not. There are various SHA
supervised and unsupervised ML techniques. Unsupervised algorithms developed by the NSA and NIST and belongs
learning may be applied to discover the structure of input to the SHA-2 family [25]. SHA was initially envisioned
VOLUME 4, 2016 3
Block Hash Block Hash ------------- Block Hash
Block Header Block Header Block Header
Timestamp Timestamp Timestamp

Difficulty Difficulty Difficulty
Previous Block Hash Previous Block Hash Previous Block Hash
Nonce Nonce Nonce
Merkle Root Merkle Root -------------- Merkle Root
Transaction Hashes Transaction Hashes Transaction Hashes
Block-1 Block-2 Block-N
FIGURE 1: Blockchain Structure
as a fragment of the Digital Signature Standard (DSS) to

produce the signature. Further, consensus algorithms are
used to determine block validity. The selection of algorithms
depends on the type of blockchain, such as public, private,
and consortium blockchain. The selected algorithm should
ensure the consensus among nodes [26]. It must be able to
use resources efficiently and tolerate a degree of safety during
FIGURE 2: Proposed architecture for Machine Learning
the event of attacks.
adoption in Blockchain-based Smart applications
Further, a smart contract is a program (set of codes) that
runs on the blockchain and adds blocks whenever certain
conditions are met. It is defined by translating actual legal
models may be eliminated. ML-models can be based on
contracts into programs to enforce the legal contract onto
specific segments of the chain rather than the entire data-
the blockchain [27]. It is similar to stored-procedures in
set. This could give custom models for different applications,
relational databases. It is stored as scripts in the blockchain
such as fraud detection and identity theft detection. Few of
and executed according to the data fed to them to produce
the benefit are listed beneath when ML is applied:
outputs that are expected from the original contract [28]. It
governs transactions either executed fully or partially based • User authentication as a legitimate user for requesting or
on the current input. The primary purpose of a smart contract performing any transaction in the blockchain network.
is to provide superior security with a reduction in cost and • BT provides a high level of security and trust.
delays associated with traditional contracts. • Blockchain integrates public ML models into smart
contracts to ensure that the conditions and terms which
C. INTEGRATION OF MACHINE LEARNING IN were previously agreed are sustained.
BLOCKCHAIN-BASED APPLICATIONS • BT helps in the reliable implementation of an incentive-
The learning capabilities of ML can be applied to blockchains based system; thus, it encourages users/customers to
based applications to make them smarter. By using ML contribute data. This huge data will help to improve ML
security of the distributed ledger may be improved. ML may model performance.
also be used to enhance the time taken to reach consensus • ML models can be updated on-chain environment of BT
by building better data sharing routes. Further, it creates an with a small fee and off-chain, locally on an individu-
opportunity to build better models by taking advantage of the alâĂŹs device without any costs
decentralized architecture of BT. We proposed architecture • Good data contributions can happen from users/customers,
for ML adoption in BT-based smart application, as shown in these data consistently computed, and rewards can be
Figure 2. Here, the smart application collects data from dif- given to the users.
ferent data sources such as sensors, smart devices, and Intenet • Tamper-proof smart contracts can be evaluated by dif-
of Things (IoT) devices. Data collected from these devices ferent machines (having different hardware configura-
get processed as part of smart applications. The blockchain tion), ML models will not diverge from their potential
work as an integral part of these smart applications. Then, and produce results exactly as it is supposed to do.
ML can be applied to these application’s data for analysis • Payments processed in real-time with trust on a
(Data analytics and real-time analytics) and prediction. The blockchain environment.
data sets used by ML models could be stored on a blockchain • Blockchains tools, for instance, Ethereum, deal with
network. This reduces errors in the data such as duplica- thousands of decentralized machines all over the world.
tion, missing data value, errors, and noise. Blockchains are This guaranteed users that it is never completely un-
focused on the data, and hence data-related issues in ML reachable or offline.
4 VOLUME 4, 2016
III. SURVEY PROCEDURES AND TAXONOMY CIDS can be addressed with the use of ML classifiers on
In this section, we refer to the procedures used to perform this BT. These classifiers run locally by the data owner, and
study; for example, our search approach and inclusion criteria the results shared with other users within the network. To
for the final set of papers. Likewise, we present a detailed handle the trust computation issue in CIDS, Alexopoulos
taxonomy based on our literature review. et al. [29] proposed an approach to handle the alerts and
data distribution among the participants. Then, it verifies the
A. SYSTEMATIC LITERATURE REVIEW data and executes a suitable consensus algorithm to add a
We used standard databases (for example IEEEXplore, ACM new data block within the blockchain network. These data
Digital Library, ScienceDirect, and Springerlink) and Google alerts could be encrypted with keys distributed to selected
Scholar to search existing research work, using keywords parties. Another approach was to keep them on a separate
such as âĂIJMachine Learning for Blockchain-based Smart blockchain while still being part of the network. Further, an
Application,âĂİ (âĂIJBlockchain for Smart ApplicationsâĂİ anomaly detection system (ADS) was implemented based on
AND âĂIJMachine Learning for Smart applicationsâĂİ ). the assumption that a similar kind of attack may occur, but on
In the initial phase, publications emphasis essentially on different nodes in the blockchain network [32]. This system
blockchain approaches for smart applications. Then, in the does not discard information about orphans and forks that
next phase of the search, we concentrated on an ML-based are usually done by other ADS. The attacked nodes shared
solution for blockchain-based smart applications. Based on this information with other neighborâĂŹs node within the
the results of these searches, we removed duplicate articles network. In the experimental analysis, the system success-
and obtained our first set of more than 350 publications. fully prevented the same kind of attack (on other nodes) with
Further, we steered the search procedure with several maga- negligible overhead.
zines, journals, and conferences dedicated to the parent field,
ML, smart applications, and Blockchain. Based on that, we B. MONITORING
found 130 articles. Then, we studied the different sections A blockchain monitoring system has been prototyped using
of the articles like abstract, conclusion, and introduction. Self Organising Maps (SOMs) [33]. It has modeled the
Then we categorized these articles as âĂIJrelevantâĂİ or dataset without external control. Here, large vectors map
non-relevantâĂİ to ML for blockchain-based smart applica- to smaller and lower dimension vectors using SOMs. The
tions. Lastly, we only selected 60 publications to present the system analyzed blockchain data using Kohonen and SOM-
taxonomy, as shown in Figure 3. Here, each layer is color- brero libraries (in the R programming environment). The
coded to tailor each level of the taxonomy. For example, the result shows the effective key attributes monitoring of the
root represents the level-0 of the classified taxonomy and blockchain nodes and finds the emerging patterns. More,
presented in blue color. In a similar way, the four major another approach included the distributed pattern recognition
dimensions are representing as level-1 in light-green color system with the concept of Graph Neurons (GNs) [31] to
and so onwards. monitor the blockchain system. GNs are scalable and could
recognize patterns from similar or incomplete patterns. The
B. TAXONOMY GN communicated with adjacent nodes to detect events
In this paper, we studied the papers that are predominantly within the network by using input data. Preliminary results
based on either ML, blockchain, or both. The presented showed object detection by the GN was accurate; still, further
survey has been divided into four dimensions: Goal Oriented, work is needed in the area.
Layer Oriented, Countermeasures, and Smart Applications.
Figure 3 shows the taxonomy of ML adoption for blockchain- C. PREDICTION
based smart applications. The ML models are mostly used for prediction. A good
prediction model helps to make the right decision making
IV. GOAL ORIENTED and analytics. Added to this, an ML model to predict the
In this dimension, research work has been included from the bitcoin prices has been proposed by Valenkar et al. [30].
following aspects: (i) preventing, (ii) predicting, (iii) mon- This model uses bayesian regression and random forest with
itoring, (iv) detection, and (v) response of blockchain. For several features such as block size, total bitcoins, day high,
instance, [29] discussed the detection of intrusion in a collab- number of transactions, and trade volume. The trained dataset
orative manner, [30] predicts bitcoin prices using Bayesian was normalized using log, z-score, and box-cox normaliza-
regression method and [31] monitors the blockchain network tion techniques. Further, a price prediction study has been
in a distributed manner. done for several cryptocurrencies such as Ripple, Litecoin,
Dash, Bitcoin, and Ethereum cryptocurrencies [34]. Here,
A. DETECTION correlation matrices for feature selections used and reported
This subsection includes the detection of the attacks to the general trends within the network. The proposed model
handle data security issues. Meng et al. [12] has proposed used multiple regression techniques on bitcoin. The system
a Collaborative IDS (CIDS) with a data-sharing agreement showed the 0.9944 accuracies for price prediction.
on a blockchain-based environment. Data privacy issues of
VOLUME 4, 2016 5
Machine Learning Adoption for Blockchain-

based Smart Applications
Layers Counter- Smart

Goal Oriented
Oriented measures Applications
Application Customer
Detection Analytical
Layer Services
Device
Monitoring Endpoint Real-Time
Personalization
Product
Prediction Process Manufacturing
Response User Data Trading
UAV
Prevention
Smart Cities
Energy and
Utilities
FIGURE 3: Proposed Taxonomy for Machine Learning adoption in Blockchain-based Smart Applications
D. RESPONSE using the key. Then, each work history entry compared
Tsolakis et al. [35] presented an approach for the secure with the database of the previous organization. This ensures
exchange of energy data between a cluster of end-users and consistency and integrity of data as well as the authenticity
a Virtual Node. This approach uses a blockchain-based ar- of the sender.
chitecture to give a demand response solution. This solution A similar idea to tracking studentâĂŹs learning history
included Fog-Enabled Intelligent Devices (FEID) [36] at the has been proposed [40]. The main aim was to maintain the
user side (act as a blockchain node) and managed smart learning histories of the students and ensure access control,
contracts with the energy producer [36] in cloud computing privacy, and security. To ensure privacy and access con-
platform [37] [38]. Then, a group of energy users combined trol, three contracts were suggested: Learner âĂŞ Learning
into a centralized Virtual Node. All predictions and forecast- Provider Contract (LLPC), Registrar âĂŞ Learning Provider
ing about the energy consumption have done on the Virtual Contract (RLPC), and Index Contract (IC). The role of RLPC
Nodes. The end-user learned itself from each experience that is to specify all conditions that describe request access by
helps to improve the accuracy of data passed as the next the provider within the network. LLPC shows that a learner’s
iteration to the virtual node. data exists on the providerâĂŹs database. The IC map the
learners and providers and the corresponding learning his-
tory. Here, specialized Learning Blockchain Application Pro-
E. PREVENTION
gram Interfaces (LB APIs) has been proposed to encourage
One of the based uses of ML in BT based application is the the adoption of this technology. Initially, the system is started
prevention of an issue; for instance, employee verification by with a boot node in the network. A new provider is added
a future employer. A mechanism proposed to reduce the time to the network as per the conditions described in RLPC. A
taken to verify work history details provided by a prospective new learner account is created and noted LLPC. In case, the
employee [39]. The previously worked organization has to new provider wants access to learners’ data, it has to request
compile details of an employee (date of joining, leaving, the learner using the LLPC, and upon granting, data will be
post, etc.), and a public key is generated to encrypt the accessible.
data. Subsequently, a new smart contract is created for the
employee. Then, the address of the smart contract is entered
into the organizationâĂŹs database, and later it verified by
the future employer. In a future organization, work history is
fetched from the smart contract of employees and decrypted
6 VOLUME 4, 2016
TABLE 2: Comparative Analysis of Goal Oriented Approaches

Authors Year Objective Merits Demerits 1 2 3 4 5
Meng et al. [12] 2018 To present use of blockchain in In- Scope of application of blockchain Discusses only data sharing and X X X X X
trusion Detection was discussed trust management issues of col-
labrative intrusion detection
Alexopoulos et al. 2018 Improvement in trust and account- Design considerations thoroughly Use of ML not covered X X X X X
[29] ability in Collaborative Intrusion explained
System using blockchain
Signorini et al. [32] 2018 to provide a blockchain based extremely less bandwidth overhead Only basic testing carried out con- X X X X X
anomaly detection system of about 0.248% sisting of only 3 node
Chawathe et al. [33] 2018 self-organizing maps to monitor Running time for the number of Security issues are not highlighted X X X X X
blockchain data in real-time iterations, attributes and instances in storing bitcoin data in a database
shows a linear relationship
Hudaya et al. [31] 2018 distributed pattern recognition use of graph neuron approach en- integration of IoT-blockchain not X X X X X
for event monitoring within IoT- ables distributed pattern recognition covered in depth
blockchain network
Saad et al. [34] 2018 predicting bitcoin prices with high user activity dynamics and its ef- not enough focus on other factors X X X X X
accuracy using ML fects discussed extensively other than supply demand
Tsolakis et al. [35] 2018 demand response system using security concerns addressed appro- ML aspects not covered X X X X X
blockchain priately
Sarda et al. [39] 2018 To prevent work-history related Extensive use of blockchain to No intersection of blockchain and X X X X X
frauds using blockchain share and verify work history ML was found, discussion on secu-
rity issues was not carried out
Liang et al. [41] 2019 Present micro-blockchain based dy- Scope of application of blockchain Discusses attacks and trust manage- X X X X X
namic intrusion detection was discussed ment issues only
Sgantzos et al. [42] 2019 Present genetic algorithm imple- Discussed framework and future Natural Language interaction with X X X X X
mentation on blockchain applications blockchian based system need to be
explored
Parameters- 1:ML, 2:Blockchain, 3:Detection, 4:Algorithm, 5:Trust Management
Notations- X: Considered, and X: Not Considered.
F. COMPARISON OF EXISTING APPROACHES FOR A. APPLICATION LAYER

GOAL ORIENTED DIMENSION The application layer helps to accomplish the security, liveli-
A detailed comparison of the approaches discussed above ness within the network. An approach to detect malware (in
is shown in Table 2. This comparison of works included the form of a portable executable file) has been proposed
several parameters such as trust management, ML, algorithm, using DL methods [43]. The portable executable file was
blockchain, detection, merits, and demerits. converted to a grayscale image to feed to the Deep belief
network. In this solution, the receiving node broadcast the
file to other nodes within the network. The file executed
V. LAYERS ORIENTED locally in the local detection model of each node. The prob-
At this moment, BT is very young, and understanding of ability value of being malware is appended with the block.
layer division is not enough. Considering the BT as a solitary A weighted average for the trust of the node is applied to
layer is similar to enduring everything among the physical the appended probabilities. The weighted average calculated
layers and transport layers into a single layer. While separat- on the condition that the file is malware or not. The model
ing the BT into multiple layers, we can understand various used Restricted Boltzmann Machines (RBM) techniques with
properties of BT which are needed to be implemented. Some 3000 hidden units. The results showed good accuracy for
of the properties are: (i) Security: Nodes that do not control malware detection.
rare resources (commonly computing power) majorly cannot
convince others for a different version of the ledger. (ii) Live- B. ENDPOINT
ness: Here, new blocks can be added to the blockchain with Endpoints included the small computing devices which are
suitable latency. (iii) Stability: Nodes within the blockchain participating in the BT based application network. To handle
network should not amend their belief of the consensus the issue of fast computing and reduction in computation
ledger except rare cases. (iv) Accuracy: Blocks added to the power, emergent technology Edge Computing comes into the
ledger must signify valid transactions such as they imitate to picture. The Edge Computing Service Provider (ECSP) can
a description of how new blocks relate to previous blocks. meet the requirements, and an approach has been proposed
This dimension discusses the adoption of ML techniques in to maximize the revenue of ECSP [44]. The ECSP allocates
various layers of a blockchain network, such as endpoints, the resource unit to the highest bidder who participated in the
application layer, process, network (intrusion), and finally, bidding process.
user level. The proposed categorization into layers recognizes
each property. Security is accomplished at each layer, Live-
Similarly, Kim et al. [45] proposed an architecture to
ness is realized at the user layer, Stability is achieved at the
offload resources computations in Deep Neural Networks to
network and application layer, and accuracy is achieved at the
edge computing. Within this architecture, the embedded de-
endpoints, wherever blocks have significance.
vice (the computation device at the edge server) and the edge
server had to make initial deposits. After that, the embedded
VOLUME 4, 2016 7
TABLE 3: Comparative analysis for Layer-Oriented approaches

Author Year Objective 1 2 3 4 5 Pros Cons
Raje et al. [43] 2017 To design a decentralized firewall X X X X X Discussed Deep Belief Networks Number of devices in experimen-
using blockchain and deep learning and training procedure in detail and tal setup was small and working of
explored various architectures blockchain not discussed
Portnoff et al. [46] 2017To identify human traffickers using X X X X X A detailed discussion on data anal- Bitcoin transaction linking was not
classified ads and bitcoin transac- ysis perfect, false positives were high,
tions blockchain aspect not covered
Wasim et al. [47] 2017 Proposed a Law as a service X X X X X Results and system model defined Blockchain not included in the
architecture to monitor Contract clearly, proposed unsupervised ML model description
Breaches and issue injunctions algorithm
Luong et al. [44] 2018 Edge computing usage for mining X X X X X ML algorithms thoroughly dis- Focus on optimizing revenue but no
applications and ML for resource cussed, well-documented results optimal allocation of resources
allocation
Kim et al. [45] 2018 Proposed deep neural networks ar- X X X X X Executes complex programs on Experimental results not discussed
chitecture for blokchain based edge ethereum blockchain using virtual in detail, Deep Neural Network ap-
computing application machines plications not covered
Parameters- 1:Blockchain, 2:Edge Computing, 3:Neural Network, 4:Supervised Learning, 5:Unsupervised Learning
Notations-X: considered, and X: not considered.
device sends the computations to edge servers. Then, the E. COMPARISON OF EXISTING APPROACHES FOR
edge server returns the results, which are verified by other LAYER ORIENTED USE CASES
nodes within the network. The server is rewarded for its A detailed comparison of the approaches is discussed in
work after validation. The typical Ethereum architecture had Table 3. This comparison is made based on several param-
been modified in this implementation by replacing Ethereum eters such as edge computing, blockchain, neural network,
virtual machine by V8 javascript virtual machine. The blocks supervised learning, pros, and cons of the proposed approach.
generation part had been separated from the virtual machine
to ensure that smart contract execution cannot affect the VI. COUNTERMEASURES
rate of generation of blocks. The system performance shows The ML approached in response to a threat or monitoring
better results compare to the state-of-art approaches. a system can be classified as real-time analytics or analysis
based on historical data [33]
C. PROCESS
A. ANALYTICAL
This subsection comprises ML techniques to the process in-
In todayâĂŹs world, every enterprise generates huge
volves in the blockchain-based applications. An ML classifier
amounts of data from different sources such as social me-
has been used to detect the human traffickers using adult
dia, smartphones, IoT, and other computing devices. These
classified ads [46]. It included ads dataset from Backpage (a
data are tremendously valuable to organizations. The overall
website for online classifieds). It used a supervised learning
technique to find a meaningful pattern from these data is
model and used logistic regression for classification. The
called data analytics. It is a process to convert data from
labeled dataset with identifiers like email and phone number
foresight to insight. It describes what happened in the past,
has been used for testing. This approach takes two ads at a
draw awareness about the present, and make predictions
time as an input to the model. If the ads are from the same
(with some ML techniques) about the future. ML techniques
author, the output will be same. The trained model showed an
can be categories as supervised, unsupervised, and reinforce-
89.54% true positive rate. Further, this approach used to build
ment learning to analyses the data. The supervised learn-
a graph and found links between ads and their related bitcoin
ing included classification and regression originated on the
transactions. This approach showed a high false-positive rate.
idea of example-based learning. Next, unsupervised learning
techniques perform clustering, dimensionality reduction, and
D. USER recommendation of a system based on the dataset. This used
Users are the individual or system that uses the system to recognize hidden patterns or focus on the well-educated
functionality seating at one end of the application. Wasim behavior of the machine. The reinforcement learning ap-
et. al [47] proposed law as a service ML-based architecture proach helps to reward maximization. In the next decades,
to monitor contract breaches. Previously, contract breaches our society will be driven by new technological developments
were dealt with or without the use of courts. The pro- in ML and BT. Further, a traceability algorithm has been pro-
posed system used an unsupervised ML algorithm called posed for bitcoin mining by using the ANN approach [51].
Probability-based Factor Model to issue injunctions. This It aimed to remove irrelevant data in mining and introduce
model simulated using three service providers Redis [48], traceability to the system. This was helpful in a distributed
MongoDB [49], and Memcached Servers [50]. Services architecture and decrease traceability time.
monitor the contracts for breaches. The results showed that DL is a subset of ML that is originated on definitive
services that perform complex operations are more likely to algorithms influenced by the overall structure and work of
breach contracts. the neural network in the human brain. DL rationalizes
tasks to perform speech recognition, image recognition, make
8 VOLUME 4, 2016
insightful decisions on natural language processing. DL net-

work consisted of input, output, and hidden multi-layer. It
accepted a new block and a history of the previous block as
an input in BT based applications. It used a state-transition
algorithm on the features such as hash value, nonce, address,
and transaction data. An ML-based classification approach is
used to identify the cyber-crime in Bitcoin [52]. The results
provided a glimpse of the size of cybercrime in the Bitcoin
ecosystem.
B. REAL-TIME
ML enables real-time analytics for all types of data, such as
social âĂŞ accessible, transactional, and operational. It uses
in-memory computing though leaving unremittingly updated
data securely. It improves analytics accuracy and accelerates
the predictive behavior of ML models. It has been noticed
that BT makes real-time cross edge transactions in monetary FIGURE 4: A list of Smart Applications
and payment frameworks. Several fintech innovators and
banks are currently investigating blockchain due to fast and
real-time settlement of massive amounts independent of geo- A. CUSTOMER SERVICE
graphic hindrances. Similarly, associations that require real- With the upsurge of the customer, the Customer service has
time analytics of information on a huge scale can approach to be more efficient and automated to meet rising customer
an ML and blockchain-empowered framework to accomplish needs. One of the best ways is to automate the process to
the goal. With ML and BT, financial institutions and other increase a companyâĂŹs capabilities. Wang et al. [58] pro-
associated organizations across the globe can trace the data posed an AutoML framework for the blockchain-based appli-
changes to make quick and fast business decisions regard- cation. It consists six layers: (i) Organization Layer (Includes
less of irregular activities or suspicious transactions. More- entities such as shops, malls, online stores), (ii) Customer
over, the performance of the blockchain-based Software- Layer, (iii) Application Layer (applications provide services
Defined Vehicular Network (SDVN) has been improved us- to the consumers), (iv) Data conversion layer(Data masking
ing deep Q-learning methodology [53] [54]. This approach and unmasking), (v) AutoML layer (Consists various ML
used a permissioned blockchain with Byzantine fault tol- models such as Linear regression, SVM and logistic regres-
erance as a consensus protocol. Results showed that this sion for predictions), (vi) Blockchain Layer (Secure storage
scheme managed network and computing resources better of data and ensures the safe data exchange) This framework
and gave the best throughput in the SDVN [55]. Further, helps organizations to keep their data safe, automate their
Liu et al. [22] recommended a data collection framework for processes and share data with other organizations in a mu-
Industrial-IoT (IIoT) applications. This framework combines tually beneficial and safe way.
the use of deep reinforcement learning (DRL) and Ethereum
blockchain. To store and share data, Ethereum nodes were B. DEVICE PERSONALIZATION
categories into two categories: mining nodes and nonmining
Device Personalization (DP) is a component that uses pre-
nodes. The proposed DRL algorithm used three plain compo-
dictions across smart devices to improves the quality of
nents, actions (moving path and remoteness of mobile termi-
service (QoS) such as actions in the launcher, smart text
nals), states (environment description), and rewards (amount
selection during writing on text pad. DP Services uses system
achieved). The proposed algorithm shows the 34.5% increase
permissions to provide smart predictions. In a smart home
in geographical fairness compared to a random solution.
environment, and ML model-based single DP framework has
been implemented [59]. Here, smart devices connected to a
VII. SMART APPLICATIONS smart hub. Every time a user uses a device, a log is generated
The last dimension covers the adoption of ML techniques with user data, device data, and other parameters for that
to BT-based smart applications such as data trading, UAV, device. For example, uses of air conditioner at home. This
product manufacturing, Medical, and Healthcare [56], Smart log data help to adjust the operation of the device.
Cities, automated Customer Service, and Device Personaliza-
tion as shown in Figure 4. ML and BT are revolutionizing up- C. PRODUCT MANUFACTURING
to-date technologies by transforming customer experiences, As a feature of the manufacturing process, organizations
behaviors, and business models [57]. Both are making strides have started trusting blockchain-based procedures to em-
in the major smart application: power production, security, transparency, and compliance
checks. Instead of following fixed schedules of machine
VOLUME 4, 2016 9
maintenance, ML algorithms are being used to make flexible estate, and utilities [61]. (i) Smart Homes: Smart homes
plans at specific periods. Product testing and quality control can be monitored, and DP can help to improve the quality
have also automated increasingly, with versatile algorithms. of livelihood. (ii) Smart Parking System: Here, arrival and
It successfully detects faulty and good products, particularly departure of vehicles can be tracked for different parking
in profoundly delicate situations. For example, Porsche (a car lots available in a decentralized manner within the smart city
manufacturing company) is one of those early adopters of [61]. Consequently, smart parking lots should be designed by
ML and BT technologies; to improve automobile safety and seeing the number of cars in each region. Furthermore, new
increase capabilities. The organization utilizes blockchain parking lots must be recognized spontaneously to benefits in
innovation to transfer data more safely and rapidly, offering dealers and vehicle ownerâĂŹs daily life. (iii) Smart Weather
its clients peace of mind, regardless of parking, charging, and and Water Systems: Here, the system can use some sensors
third-party access to their car. to generate appropriate data such as temperature, wind speed,
rain, and pressure. The analysis of these data through ML
D. DATA TRADING techniques can contribute to improving the density of smart
DataTrading is an innovative platform that makes advanced cities. (iv) Smart Vehicular Traffic: Vehicular traffic data with
trading possible for retail traders from all over the world. An a suitable analysis will benefit the government and citizens to
Ethereum based data trading framework has presented that a great extent [62]. Everyone can decide the arrival time to
succeeded in preventing single-point failure and preserving a destination by using these data. (v) Surveillance Systems:
privacy at the same time [10]. The framework consists of Physical security is an utmost important concern for citizens
three entities, data provider, a data consumer, and a market anywhere they live. To address this issue, smart technology
manager. Once the network is set up, the data consumer and such as ML and BT can be configured for it. Consequently,
data provider registered themselves with the market manager. collecting and analyzing data and identifying crimes is one of
The data provider needed to deposit an amount with the the challenging tasks. (vi) Smart Healthcare: It includes the
manager greater than the amount to be paid by the consumer. accessibility of the caregivers and doctors, identification of
A list indicating topics of data available with the provider is nearest medical stores, and clinic. Hence, ML and BT play
published. The client referred it and requests some encrypted a vital role while data is getting generated [63] [64]. (vii)
data blocks for content validation using a distance metric Smart Governance and Smart Education: Smart governance
learning technique. After successful validation, the client can maintain a city smartly. A smart city includes a differ-
responded, and the provider sent a signature, and the client ent way of education. It captures the data of the students
published a smart contract to the network. The provider then and employees in educational and government institutions.
sent another signature, and these two signatures, the client, Predictions and analytics required to keep the stuff up to
decrypted the data. There are different protocols such as setup standards.
protocol, register protocol, payment and acknowledgment
protocol, and query protocol that facilitate the communica- G. ENERGY AND UTILITIES
tion between provider, consumer, and the manager. In the Energy industry, BT is assisting to simplify energy
exchanges. For example, IOTA enables smart transformation
E. UNMANNED AERIAL VEHICLE across the entire energy industry by implementing BT [65].
A UAV or drone is an aircraft that runs without a pilot It uses the concept of peer-to-peer (P2P) energy production
(human) aboard. Kuzmin et al. [9] has proposed blockchain- and consumption. Smart energy microgrids are progressively
based UAVNet model includes different devices such as developing fame as a technique of making ecological energy
a network of satellites, cellular base stations, and ground resources. LO3 Energy (a Newyork-based organization) is
control stations. Here, BT served multiple purposes, such also using a blockchain-based development to develop en-
as preserving the integrity of the data and for distributed ergy generation, transmission, conservation, and exchanging
graphs computation. The communication between satellites within neighborhood networks. The technology uses micro-
and base stations is prone to electromagnetic jamming, hence grid and smart meters, together with smart contracts, to
a blockchain-based system enables UAVs to store relevant manage and track energy transactions. GE Digital, together
coordinateâĂŹs data and operate autonomously within the with Evolution Energie (a startup), has created an application
jamming zone. This solution used a proof-of-graph consen- that encourages the tracking of renewable energy in SG and
sus algorithm with a simplified memory bounded algorithm uses blockchain to give energy sources certificate. The idea
of any existing shortest path to validating a new transaction. is allowing organizations and individuals to trade renewable
energy sources without the involvement of third-party [36].
F. SMART CITIES Globally, most of the industries are dogged by third-party,
Smart cities improve the living experiences of individuals. who increase the business cost. BT has already interrupted
ML and blockchain emergent technologies play a crucial role that model by facilitating the P2P model for customers.
in the innovation of smart cities to provide critical services The revolution works even better when combined with the
and infrastructure components such as healthcare [60], city ML and BT together. We have been witnessed the financial
administration, smart homes, education, transportation, real revolution by using the cryptocurrency world; similarly, we
10 VOLUME 4, 2016
are going to observe more innovation disruptions as a result

of combining ML and BT.
Infrastruct
VIII. FUTURE RESEARCH ISSUES AND CHALLENGES ure
Researchers are looking forward to these technologies across
the globe, but still, various obstacles resist the integration of Suitability Memory
BT and ML [66]. Their integration is still in its infancy. Many
open issues and challenges are yet to be addressed. Here,
we discuss the futuristic open issues and challenges of ML
adoption in BT for secure communication, as shown in Figure Research
5. We highlighted the challenges as Suitability, Infrastructure, Challenges
Privacy, security, Memory, Implementation, and Quantum Implement
Privacy
ation
resilience.
A. SUITABILITY
Blockchain is a viable solution if the source of data cannot be
Quantum
trusted, and several entities are high in the distributed envi- Security
resilience
ronment. If performance is required, then a simple database
is a better option. Therefore, the architecture of blockchain
must be understood before its use in any application [67].
FIGURE 5: Future Research Issues and Challenges
B. INFRASTRUCTURE
Blockchain specific hardware and network infrastructure im-
addition of blocks and transactions needs to be decreased to
prove the performance of many blockchain-based applica-
meet the inevitable demand.
tions. These could include network administration, mining
hardware, decentralized storage, and communication proto-
F. SECURITY
cols [30]. However, products tailored to use in blockchain
are still under investigation (involving major tech companies Blockchains are decentralized, and they are prone to security
and financial institutions). issues [68]. The most common concern is that due to attacks,
the consensus protocol may be compromised, such that the
mining power of a few farms will control which blocks are
C. PRIVACY
added to the network. This particular concern is present in
Data generated by devices to be stored on the blockchain is public blockchains. Private versions are unaffected by this
available to the entire blockchain nodes [33]. This leads to a attack as they have each node identified, and an appropriate
potential privacy concern for data that needs to be kept either consensus protocol is in place.
private or confidential. Such issues could be resolved by the
use of private blockchains, controlled access, and encryption. G. QUANTUM RESILIENCE
However, the ML models adoption on these limited data
The hashing algorithms used by blockchains could soon be
imposed barriers for predictions and analytics.
broken with quantum computers. A blockchain is predom-
inantly at risk from this because it uses one-way functions
D. MEMORY for encryption (only protection in digital signature). This
The size of blockchain keeps growing as new blocks are would cause all the features that make blockchains a viable
added to it. Consequently, the entire chain must be stored storage structure (obsolete). Luckily, quantum computing
by all nodes, which creates a significant memory constraint offers opportunities to boost the performance and security of
on these devices. The performance is also affected by an blockchains. Quantum cryptography can reinforce the secu-
increase in chain size. Besides that, the storage of irrelevant rity of the blockchain network as quantum communication is
data also wastes computational resources. Blockchains are authenticated inherently (users cannot mimic another user). It
immutable, and hence storage management is a major issue can encrypt entire P2P communications and replace classical
in most implementations. digital signatures in the blockchain network. Research is
underway to design blockchain with quantum computing
E. IMPLEMENTATION [69].
A large- scale blockchain network will require an equiva-
lently large number of transactions. Implementation of this IX. CASE STUDY
size of blockchain invites potential issues; for instance, high In order to demonstrate the proposed architecture for ML
demand for internet bandwidth, which not easy to reduce, adoption in blockchain-based smart applications, we present
transactions will be a burden on the network. Hence, the a case study, as shown in Figure 6. In this case study, we study
VOLUME 4, 2016 11
Energy Storage
Device
Smart Blockchain
Grid
Energy Trading Platform

Accounts
Buildings
Industries
Individual Houses
Entities participating in energy trading
FIGURE 6: An Energy Trading system
a blockchain-based SG system for energy transactions using utility provider, or SG. Blockchain entity is a distributed
cryptocurrency [70] [71]. This framework uses a blockchain digital ledger, which is encompassing all energy transactions
and DL approach to complete an energy transaction. It is in the SG system.
a reliable P2P energy system. It is based on the Byzantine Once an energy seller produces energy from renewable
fault tolerance algorithm to produce high system throughput. energy sources and uses that energy for his own use. After
This approach consists of five phases (i)setup phase, (ii) that, if the seller left with energy, then he publishes his per-
agreement phase, (iii) consensus-making phase, (iv) block unit prices per kilowatt energy on the blockchain network.
creation phase, and (v) a change view phase. Here, blocks In this P2P energy trading system, the needed energy buyer
are generated using hash functions and a short signature. It will look upon the published price and unit of energy. First of
is an IDS that works on recurrent neural networks to detect all, the energy buyer checks his account that he has sufficient
fraudulent transactions and network attacks in blockchain- cryptocurrency balance for energy trading. Then, if the en-
based energy applications. The performance of this IDS has ergy demand of energy buyers and per-unit price matched his
been studied on different energy datasets. It consists of four requirement of energy at that moment, he sends a purchase
entities for communications: (a) Energy Buyer, (b) Energy request to the energy seller through the blockchain network.
Seller, (c) Blockchain, and (d) IDS [41]. Energy buyer trade After validation of the buyer on the blockchain network seller
energy with the energy seller. An energy buyer shows that sells the energy to the buyer and receives cryptocurrency in
he has sufficient energy money (cryptocurrency) that fulfills his account. This P2P energy trading system required further
the minimum asset requirement of energy sellers. An energy data analysis to identify the frequent buyer and seller for
buyer will be a person or commercial building or industry. the recognition of malicious transactions within the network.
Energy buyers can be located in a Home Area Network A malicious block or node can impact the P2P system and
(HAN), Building Area Network (BAN), or Neighborhood compete for energy trading. IDS can help to figure out the
area Network (NAN). Energy seller entities demonstrate that deceitful transactions and network attacks on energy trading
it has sufficient energy to sell. Energy sellers can be located applications [72]. The result shows the good performance of
in HAN, BAN, NAN, or energy companies. Energy seller the system compares to the state-of-art approaches.
(entities) produces energy from renewable sources such as
wind energy, solar energy, and biomass. The seller can be
a neighbor, local society with renewable energy resources,
12 VOLUME 4, 2016
X. CONCLUSION [13] M. Conti, E. Sandeep Kumar, C. Lal, and S. Ruj, “A survey on security
and privacy issues of bitcoin,” IEEE Communications Surveys Tutorials,
The recent advancements in Blockchain and ML have made vol. 20, pp. 3416–3452, Fourthquarter 2018.
them path-breaking technologies. The distributed ledger has [14] D. Ucci, L. Aniello, and R. Baldoni, “Survey of machine learning tech-
the possibility to work as the backbone of various smart niques for malware analysis,” Computers & Security, vol. 81, pp. 123 –
147, 2019.
applications such as smart cities, UAV, SG, data trading. In
[15] K. Salah, M. H. U. Rehman, N. Nizamuddin, and A. Al-Fuqaha,
this paper, we have presented detailed information on BT “Blockchain for ai: Review and open research challenges,” IEEE Access,
and ML, along with their usages in smart applications and vol. 7, pp. 10127–10149, 2019.
proposed an ML-BT based architecture. This architecture [16] F. Casino, T. K. Dasaklis, and C. Patsakis, “A systematic literature review
of blockchain-based applications: Current status, classification and open
can be used to design and deploy an ML-BT based data issues,” Telematics and Informatics, vol. 36, pp. 55 – 81, 2019.
analysis system. A discussion and comparison of various [17] W. Xiong and L. Xiong, “Smart contract based data trading mode using
existing surveys are presented. Then, we presented ML-BT blockchain and machine learning,” IEEE Access, vol. 7, pp. 102331–
102344, 2019.
solution taxonomy, focusing on goal oriented, layer oriented, [18] T. M. Mitchell, Machine Learning. New York, NY, USA: McGraw-Hill,
countermeasures, and smart application dimensions. A com- Inc., 1 ed., 1997.
parative analysis of available methodologies and approaches [19] P. Louridas and C. Ebert, “Machine learning,” IEEE Software, vol. 33,
pp. 110–115, Sep. 2016.
is presented in each dimension. Then, we have listed several
[20] R. Saravanan and P. Sujatha, “A state of art techniques on machine learn-
research challenges being faced during ML adoption in BT- ing algorithms: A perspective of supervised learning approaches in data
based systems, which require solutions. We also emphasized classification,” in 2018 Second International Conference on Intelligent
Computing and Control Systems (ICICCS), pp. 945–949, June 2018.
a number of research prospects such as infrastructure avail-
[21] V. Vats, L. Zhang, S. Chatterjee, S. Ahmed, E. Enziama, and K. Tepe, “A
ability, quantum resilience, and privacy issues that can serve comparative analysis of unsupervised machine techniques for liver disease
as future research directions in this field. Then, we presented prediction,” in 2018 IEEE International Symposium on Signal Processing
a case study on the energy trading system to verify the and Information Technology (ISSPIT), pp. 486–489, Dec 2018.
[22] C. Liu, X. Xu, and D. Hu, “Multiobjective reinforcement learning: A
effectiveness of the proposed architecture and concluded the comprehensive overview,” IEEE Transactions on Systems, Man, and Cy-
paper at last. bernetics: Systems, vol. 45, pp. 385–398, March 2015.
[23] Z. Zheng, S. Xie, H. Dai, X. Chen, and H. Wang, “An overview of
blockchain technology: Architecture, consensus, and future trends,” in
REFERENCES 2017 IEEE International Congress on Big Data (BigData Congress),
pp. 557–564, June 2017.
[1] S. Kaneriya, J. Vora, S. Tanwar, and S. Tyagi, “Standardising the use of
[24] G. B. Mermer, E. Zeydan, and S. S. Arslan, “An overview of blockchain
duplex channels in 5g-wifi networking for ambient assisted living,” in
technologies: Principles, opportunities and challenges,” in 2018 26th
2019 IEEE International Conference on Communications Workshops (ICC
Signal Processing and Communications Applications Conference (SIU),
Workshops), pp. 1–6, May 2019.
pp. 1–4, May 2018.
[2] S. Tyagi, M. S. Obaidat, S. Tanwar, N. Kumar, and M. Lal, “Sensor
[25] W. Penard and T. van Werkhoven, “On the secure hash algorithm family,”
cloud based measurement to management system for precise irrigation,”
Cryptography in Context, pp. 1–18, 2008.
in GLOBECOM 2017 - 2017 IEEE Global Communications Conference,
[26] M. Vukolić, “The quest for scalable blockchain fabric: Proof-of-work vs.
pp. 1–6, Dec 2017.
bft replication,” in Open Problems in Network Security (J. Camenisch
[3] U. Bodkhe, P. Bhattacharya, S. Tanwar, S. Tyagi, N. Kumar, and M. S.
and D. Kesdoğan, eds.), (Cham), pp. 112–125, Springer International
Obaidat, “Blohost: Blockchain enabled smart tourism and hospitality
Publishing, 2016.
management,” in 2019 International Conference on Computer, Information
[27] S. Thompson, P. L. Seijas, and D. Adams, “Scripting smart contracts for
and Telecommunication Systems (CITS), pp. 1–5, Aug 2019.
distributed ledger technology.” December 2016.
[4] S. Nakamoto, “Bitcoin: A peer-to-peer electronic cash system,” 2009.
[28] K. Christidis and M. Devetsikiotis, “Blockchains and smart contracts for
[5] M. Rahouti, K. Xiong, and N. Ghani, “Bitcoin concepts, threats, and the internet of things,” IEEE Access, vol. 4, pp. 2292–2303, 2016.
machine-learning security solutions,” IEEE Access, vol. 6, pp. 67189– [29] N. Alexopoulos, E. Vasilomanolakis, N. R. Ivánkó, and M. Mühlhäuser,
67205, 2018. “Towards blockchain-based collaborative intrusion detection systems,” in
[6] A. P. Namanya, A. Cullen, I. U. Awan, and J. P. Disso, “The world of Critical Information Infrastructures Security (G. D’Agostino and A. Scala,
malware: An overview,” in 2018 IEEE 6th International Conference on eds.), (Cham), pp. 107–118, Springer International Publishing, 2018.
Future Internet of Things and Cloud (FiCloud), pp. 420–427, Aug 2018. [30] S. Velankar, S. Valecha, and S. Maji, “Bitcoin price prediction using
[7] S. Krit and E. Haimoud, “Overview of firewalls: Types and policies: machine learning,” in 2018 20th International Conference on Advanced
Managing windows embedded firewall programmatically,” in 2017 Inter- Communication Technology (ICACT), pp. 144–147, Feb 2018.
national Conference on Engineering MIS (ICEMIS), pp. 1–7, May 2017. [31] A. Hudaya, M. Amin, N. M. Ahmad, and S. Kannan, “Integrating dis-
[8] G. Betarte, E. Gimenez, R. Martinez, and A. Pardo, “Improving web tributed pattern recognition technique for event monitoring within the iot-
application firewalls through anomaly detection,” in 2018 17th IEEE In- blockchain network,” in 2018 International Conference on Intelligent and
ternational Conference on Machine Learning and Applications (ICMLA), Advanced System (ICIAS), pp. 1–6, Aug 2018.
pp. 779–784, Dec 2018. [32] M. Signorini, M. Pontecorvi, W. Kanoun, and R. D. Pietro, “BAD:
[9] A. Kuzmin and E. Znak, “Blockchain-base structures for a secure and blockchain anomaly detection,” CoRR, vol. abs/1807.03833, 2018.
operate network of semi-autonomous unmanned aerial vehicles,” in 2018 [33] S. Chawathe, “Monitoring blockchains with self-organizing maps,” in
IEEE International Conference on Service Operations and Logistics, and 2018 17th IEEE International Conference On Trust, Security And Privacy
Informatics (SOLI), pp. 32–37, July 2018. In Computing And Communications/ 12th IEEE International Conference
[10] Y. Zhao, Y. Yu, Y. Li, G. Han, and X. Du, “Machine learning based privacy- On Big Data Science And Engineering (TrustCom/BigDataSE), pp. 1870–
preserving fair data trading in big data market,” Information Sciences, 1875, Aug 2018.
vol. 478, pp. 449 – 460, 2019. [34] M. Saad and A. Mohaisen, “Towards characterizing blockchain-based
[11] C. Pop, T. Cioara, M. Antal, I. Anghel, I. Salomie, and M. Bertoncini, cryptocurrencies for highly-accurate predictions,” in IEEE INFOCOM
“Blockchain based decentralized management of demand response pro- 2018 - IEEE Conference on Computer Communications Workshops (IN-
grams in smart energy grids,” Sensors, vol. 18, no. 1, p. 162, 2018. FOCOM WKSHPS), pp. 704–709, April 2018.
[12] W. Meng, E. W. Tischhauser, Q. Wang, Y. Wang, and J. Han, “When [35] A. C. Tsolakis, I. Moschos, K. Votis, D. Ioannidis, T. Dimitrios, P. Pandey,
intrusion detection meets blockchain technology: A review,” IEEE Access, S. Katsikas, E. Kotsakis, and R. GarcÃ a-Castro, “A secured and trusted
vol. 6, pp. 10179–10188, 2018. demand response system based on blockchain technologies,” in 2018
VOLUME 4, 2016 13
Innovations in Intelligent Systems and Applications (INISTA), pp. 1–6, [56] J. Vora, A. Nayyar, S. Tanwar, S. Tyagi, N. Kumar, M. S. Obaidat,
July 2018. and J. J. P. C. Rodrigues, “Bheem: A blockchain-based framework for
[36] A. Kumari, S. Tanwar, S. Tyagi, N. Kumar, M. S. Obaidat, and J. J. securing electronic health records,” in 2018 IEEE Globecom Workshops
Rodrigues, “Fog computing for smart grid systems in the 5g environment: (GC Wkshps), pp. 1–6, Dec 2018.
Challenges and solutions,” IEEE Wireless Communications, vol. 26, no. 3, [57] S. K. Singh, M. M. Salim, M. Cho, J. Cha, Y. Pan, and J. H. Park, “Smart
pp. 47–53, 2019. contract-based pool hopping attack prevention for blockchain networks,”
[37] J. Vora, P. DevMurari, S. Tanwar, S. Tyagi, N. Kumar, and M. S. Obaidat, Symmetry, vol. 11, no. 7, p. 941, 2019.
“Blind signatures based secured e-healthcare system,” in 2018 Inter- [58] W. M. Wang, H. Guo, Z. Li, Y. Shen, and A. V. Barenji, “Towards open
national Conference on Computer, Information and Telecommunication and automated customer service: A blockchain-based automl framework,”
Systems (CITS), pp. 1–5, July 2018. in Proceedings of the 2Nd International Conference on Computer Science
[38] S. Tanwar, J. Vora, S. Kaneriya, and S. Tyagi, “Fog-based enhanced safety and Application Engineering, CSAE ’18, (New York, NY, USA), pp. 27:1–
management system for miners,” in 2017 3rd International Conference on 27:6, ACM, 2018.
Advances in Computing,Communication Automation (ICACCA) (Fall), [59] K. Singla, J. Bose, and S. Katariya, “Machine learning for secure device
pp. 1–6, Sep. 2017. personalization using blockchain,” in 2018 International Conference on
[39] P. Sarda, M. J. M. Chowdhury, A. Colman, M. A. Kabir, and J. Han, Advances in Computing, Communications and Informatics (ICACCI),
“Blockchain for fraud prevention: A work-history fraud prevention sys- pp. 67–73, Sep. 2018.
tem,” in 2018 17th IEEE International Conference On Trust, Secu- [60] R. Gupta, S. Tanwar, S. Tyagi, N. Kumar, M. S. Obaidat, and B. Sadoun,
rity And Privacy In Computing And Communications/ 12th IEEE In- “Habits: Blockchain-based telesurgery framework for healthcare 4.0,” in
ternational Conference On Big Data Science And Engineering (Trust- 2019 International Conference on Computer, Information and Telecom-
Com/BigDataSE), pp. 1858–1863, Aug 2018. munication Systems (CITS), pp. 1–5, Aug 2019.
[40] P. Ocheja, B. Flanagan, and H. Ogata, “Connecting decentralized learning [61] A. Kumari, S. Tanwar, S. Tyagi, N. Kumar, M. Maasberg, and K.-K. R.
records: A blockchain based learning analytics platform,” in Proceedings Choo, “Multimedia big data computing and internet of things applications:
of the 8th International Conference on Learning Analytics and Knowledge, A taxonomy and process model,” Journal of Network and Computer
LAK ’18, (New York, NY, USA), pp. 265–269, ACM, 2018. Applications, vol. 124, pp. 169–195, 2018.
[41] H. Liang, J. Wu, S. Mumtaz, J. Li, X. Lin, and M. Wen, “Mbid: Micro- [62] A. Mewada, S. Tanwar, and Z. Narmawala, “Comparison and evaluation
blockchain-based geographical dynamic intrusion detection for v2x,” of real time reservation technologies in the intelligent public transport
IEEE Communications Magazine, vol. 57, no. 10, pp. 77–83, 2019. system,” in 2018 Fifth International Conference on Parallel, Distributed
and Grid Computing (PDGC), pp. 800–805, Dec 2018.
[42] K. Sgantzos and I. Grigg, “Artificial intelligence implementations on the
[63] A. Kumari, S. Tanwar, S. Tyagi, and N. Kumar, “Fog computing for
blockchain. use cases and future applications,” Future Internet, vol. 11,
healthcare 4.0 environment: Opportunities and challenges,” Computers &
no. 8, p. 170, 2019.
Electrical Engineering, vol. 72, pp. 1–13, 2018.
[43] S. Raje, S. Vaderia, N. Wilson, and R. Panigrahi, “Decentralised firewall
[64] S. Tanwar, K. Parekh, and R. Evans, “Blockchain-based electronic health-
for malware detection,” in 2017 International Conference on Advances in
care record system for healthcare 4.0 applications,” Journal of Information
Computing, Communication and Control (ICAC3), pp. 1–5, Dec 2017.
Security and Applications, vol. 50, p. 102407, 2020.
[44] N. C. Luong, Z. Xiong, P. Wang, and D. Niyato, “Optimal auction for [65] “Iota.” https://www.iota.org/. Online; Accessed: 4-November-2019.
edge computing resource management in mobile blockchain networks: [66] S. K. Singh, S. Rathore, and J. H. Park, “Blockiotintelligence: A
A deep learning approach,” in 2018 IEEE International Conference on blockchain-enabled intelligent iot architecture with artificial intelligence,”
Communications (ICC), pp. 1–6, May 2018. Future Generation Computer Systems, 2019.
[45] J.-Y. Kim and S.-M. Moon, “Blockchain-based edge computing for deep [67] K. Salah, M. H. U. Rehman, N. Nizamuddin, and A. Al-Fuqaha,
neural network applications,” in Proceedings of the Workshop on INTel- “Blockchain for ai: review and open research challenges,” IEEE Access,
ligent Embedded Systems Architectures and Applications, INTESA ’18, vol. 7, pp. 10127–10149, 2019.
(New York, NY, USA), pp. 53–55, ACM, 2018. [68] A. Kumari, S. Tanwar, S. Tyagi, and N. Kumar, “Verification and validation
[46] R. S. Portnoff, D. Y. Huang, P. Doerfler, S. Afroz, and D. McCoy, “Back- techniques for streaming big data analytics in internet of things environ-
page and bitcoin: Uncovering human traffickers,” in Proceedings of the ment,” IET Networks, vol. 8, no. 2, pp. 92–100, 2018.
23rd ACM SIGKDD International Conference on Knowledge Discovery [69] A. K. Fedorov, E. O. Kiktenko, and A. I. Lvovsky, “Quantum computers
and Data Mining, KDD ’17, (New York, NY, USA), pp. 1595–1604, ACM, put blockchain security at risk,” 2018.
2017. [70] M. A. Ferrag and L. Maglaras, “DeepCoin: A Novel Deep Learning and
[47] M. U. Wasim, A. A. Z. A. Ibrahim, P. Bouvry, and T. Limba, “Law as Blockchain-Based Energy Exchange Framework for Smart Grids,” IEEE
a service (laas): Enabling legal protection over a blockchain network,” in Transactions on Engineering Management, vol. PP, pp. 1–13, 2019.
2017 14th International Conference on Smart Cities: Improving Quality of [71] N. Wang, X. Zhou, X. Lu, Z. Guan, L. Wu, X. Du, and M. Guizani, “When
Life Using ICT IoT (HONET-ICT), pp. 110–114, Oct 2017. energy trading meets blockchain in electrical power system: The state of
[48] “redis.” https://redis.io/. Online; Accessed: 4-December-2019. the art,” Applied Sciences, vol. 9, no. 8, p. 1561, 2019.
[49] “mongodb.” https://www.mongodb.com/. Online; Accessed: 4-November- [72] N. Kabra, P. Bhattacharya, S. Tanwar, and S. Tyagi, “Mudrachain:
2019. Blockchain-based framework for automated cheque clearance in financial
[50] “Memcached.” https://memcached.org/. Online; Accessed: 11-November- institutions,” Future Generation Computer Systems, vol. 102, pp. 574 –
2019. 587, 2020.
[51] R.-Y. Chen, “A traceability chain algorithm for artificial neural networks
using tâĂŞs fuzzy cognitive maps in blockchain,” Future Generation
Computer Systems, vol. 80, pp. 198 – 210, 2018.
[52] H. Sun Yin and R. Vatrapu, “A first estimation of the proportion of
cybercriminal entities in the bitcoin ecosystem using supervised machine
learning,” in 2017 IEEE International Conference on Big Data (Big Data),
pp. 3690–3699, Dec 2017.
[53] C. Qiu, F. R. Yu, F. Xu, H. Yao, and C. Zhao, “Blockchain-based
distributed software-defined vehicular networks via deep q-learning,” in
Proceedings of the 8th ACM Symposium on Design and Analysis of In-
telligent Vehicular Networks and Applications, DIVANet’18, (New York,
NY, USA), pp. 8–14, ACM, 2018.
[54] J. Bhatia, R. Dave, H. Bhayani, S. Tanwar, and A. Nayyar, “Sdn-based
real-time urban traffic analysis in vanet environment,” Computer Commu-
nications, vol. 149, pp. 162–175, 10 2019.
[55] J. Bhatia, Y. Modi, S. Tanwar, and M. Bhavsar, “Software defined vehicu-
lar networks: A comprehensive review,” International Journal of Commu-
nication Systems, vol. 32, no. 12, p. e4005, 2019.
14 VOLUME 4, 2016
SUDEEP TANWAR is an Associate Professor APARNA KUMARI pursuing Ph.D. from Depart-
in the Computer Science and Engineering De- ment of Computer Science and Engineering, In-
partment at Institute of Technology, Nirma Uni- stitute of Technology, Nirma University, Ahmed-
versity, Ahmedabad, Gujarat, India. He is visit- abad, Gujarat, India. She received M.Tech from
ing Professor in Jan Wyzykowski University in Jawaharlal Nehru University, Delhi, India in 2012.
Polkowice, Poland and University of Pitesti in Her research interest includes big data analytics,
Pitesti, Romania. He received B.Tech in 2002 from smart grid, blockchain technology and cloud com-
Kurukshetra University, India, M.Tech (Honor’s) puting.
in 2009 from Guru Gobind Singh Indraprastha
University, Delhi, India and Ph.D. in 2016 with
specialization in Wireless Sensor Network. He has authored or coauthored
more than 100 technical research papers published in leading journals
and conferences from the IEEE, Elsevier, Springer, Wiley, etc. Some of
his research findings are published in top cited journals such as IEEE
Transactions on TVT, IEEE Transactions on Industrial Informatics, Applied
Soft Computing, Journal of Network and Computer Application, Pervasive PRADEEP KUMAR SINGH is currently work-
and Mobile Computing, International Journal of Communication System, ing as Assistant Professor (Senior Grade) in De-
Telecommunication System, Computer and Electrical Engineering and IEEE partment of CSE at Jaypee University of Informa-
Systems Journal. He has also published three edited/authored books with tion Technology (JUIT), Waknaghat, H.P. He has
International/National Publishers. He has guided many students leading to completed his Ph.D. in Computer Science & En-
M.E./M.Tech and guiding students leading to Ph.D. He is Associate Editor gineering from Gautam Buddha University (State
of IJCS, Wiley and Security and Privacy Journal, Wiley. His current interest Government University), Greater Noida, UP, In-
includes Wireless Sensor Networks, Fog Computing, Smart Grid, IoT, and dia. He received his M.Tech. (CSE) with Distinc-
Blockchain Technology. He was invited as Guest Editors/Editorial Board tion from GGSIPU , New Delhi, India. Dr. Singh
Members of many International Journals, invited for keynote Speaker in is having life membership of Computer Society of
many International Conferences held in Asia and invited as Program Chair, India (CSI), Life Member of IEI and promoted to Senior Member Grade
Publications Chair, Publicity Chair, and Session Chair in many International from CSI and ACM. He is Associate Editor of International Journal of
Conferences held in North America, Europe, Asia and Africa. He has been Information Security and Cybercrime (IJISC) a scientific peer reviewed
awarded best research paper awards from IEEE GLOBECOM 2018, IEEE journal from Romania. He has published nearly 85 research papers in various
ICC 2019, and Springer ICRIC-2019. International Journals and Conferences of repute. He has received three
sponsored research projects grant from Govt. of India and Govt. of HP worth
Rs 25 Lakhs. He has edited total 8 books from Springer and Elsevier and
also edited several special issues for SCI and SCIE Journals from Elsevier
and IGI Global. He has Google scholar citations 401, H-index 12 and i-10
Index 15 in his account.
QASIM BHATIA is a graduate student of Nirma

University, Ahmedabad, India. His research inter-
est includes machine learning, network security, WEI-CHIANG HONG (M’04-SM’10) is a Pro-
fog computing, and cloud computing. fessor in the Department of Information Manage-
ment, Oriental Institute of Technology, Taiwan.
His research interests mainly include computa-
tional intelligence (neural networks and evolution-
ary computation), and application of forecasting
technology (ARIMA, support vector regression,
and chaos theory) and machine learning algo-
rithms. Professor Hong serves as the program
committee of various international conferences
including premium ones such as IEEE CEC, IEEE CIS, IEEE ICNSC,
IEEE SMC, IEEE CASE, and IEEE SMCia, etc. In May 2012, his paper
had been evaluated as Top Cited Article 2007-2011 by Elsevier Publisher
(Netherlands). In Sep. 2012, once again, his paper had been indexed in
ISI Essential Science Indicator database as Highly Cited Papers, in the
meanwhile, he also had been awarded as the Model Teacher Award by
PRUTHVI PATEL is a graduate student of Nirma
Taiwan Private Education Association. Professor Hong is a senior member of
University, Ahmedabad, India. His research inter-
IIE and IEEE. He is indexed in the list of Who’s Who in the World (25th-30th
est includes big data analytics, fog computing, and
Editions), Who’s Who in Asia (2nd Edition), and Who’s Who in Science and
cloud computing.
Engineering (10th and 11th Editions). Professor Hong is currently appointed
as the Editor-in-Chief of the International Journal of Applied Evolutionary
Computation, in addition, he serves as a guest editor of the Energies, and
is appointed as an Associate Editor of the Neurocomputing, Forecasting,
and International Journal of System Dynamics Applications. He has Google
scholar citations 5424, H-index 38 and i-10 Index 64 in his account..
VOLUME 4, 2016 15

Access 2019 2961372

Uploaded by

Copyright:

Available Formats

Access 2019 2961372

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Access 2019 2961372

Uploaded by

Copyright:

Available Formats

This article has been accepted for publication in a future issue of this journal, but has not been

Machine Learning adoption in

I. INTRODUCTION its communication also raises security concerns. To handle

TABLE 1: Comparison of the existing surveys with the proposed survey.

A. MACHINE LEARNING variables, following which it is used to make best guess

Block Hash Block Hash ------------- Block Hash

Block Header Block Header Block Header

Timestamp Timestamp Timestamp

Transaction Hashes Transaction Hashes Transaction Hashes

Block-1 Block-2 Block-N

FIGURE 1: Blockchain Structure

as a fragment of the Digital Signature Standard (DSS) to

Machine Learning Adoption for Blockchain-

Layers Counter- Smart

Response User Data Trading

TABLE 2: Comparative Analysis of Goal Oriented Approaches

F. COMPARISON OF EXISTING APPROACHES FOR A. APPLICATION LAYER

TABLE 3: Comparative analysis for Layer-Oriented approaches

insightful decisions on natural language processing. DL net-

are going to observe more innovation disruptions as a result

Energy Trading Platform

Entities participating in energy trading

FIGURE 6: An Energy Trading system

QASIM BHATIA is a graduate student of Nirma

You might also like