Cloud based Secure Service Providing for IoTs
Mubariz Rehman1, Nadeem Javaid1,∗, Muhammad Awais1, Muhammad Imran2, Nidal Naseer3
1Department of Computer Science, COMSATS University Islamabad, Islamabad 44000, Pakistan
2College of Computer and Information Sciences, King Saud University, Riyadh 11451, Saudi Arabia
3College of Engineering, Alfaisal University, Saudi Arabia
∗Correspondence: email@example.com; www.njavaid.com
Abstract—Internet of Things (IoTs) is widely growing domain
of the modern era. With the advancement in technologies, the use
of IoTs devices also increases. However, security risks regarding
service provisioning and data sharing also increases. There are
many existing security approaches, although these approaches
are not suitable for IoT devices due to their limited storage and
limited computation resources. These secure approaches also
require a speciﬁc hardware. With the invention of blockchain
technologies, many security risks are eliminated. With the help
of blockchain, data sharing mechanism is also possible. In this
paper, we proposed a novel secure service providing mechanism
for IoTs by using blockchain. We introduced cloud nodes for
maintaining the validity states of edge service providers. The
rating and cryptocurrency is given to edge servers. Given rating
and incentive is stored in cloud node and updated with respect
to time. The smart contract is proposed to check the validity
state of the edge server as well as compare and verify the service
provided by edge servers. In our proposed system we perform
service authentication at cloud layer as well as edge server
layer. Moreover, by using Proof of Authority (PoA) consensus
mechanism overall performance of our proposed system also
enhanced.By experimental analysis it is shown, our proposed
model is suitable for resource constrained devices.
Index Terms—Internet of Things, Cloud computing, Fog
computing, Proof of Work, Proof of Authority, Light Weight
The Internet of Things (IoTs) is widely growing domain
and some existing studies , predicts the total number of IoT
devices will be 26 billion in 2020. With the use of emerging
network technologies such as cloud computing, edge transpar-
ent computing and fog computing, the functionalities of IoT
constrained devices increases. In order to handle huge number
of devices, there is a need of some standardize protocol and
proper layer for an architecture which provide secure services
for IoT devices. The existing architectures involve centralized
architecture in which IoT devices are connected to cloud
servers with the use of internet. However due to rapidly
increase of IoT devices, network issues like bottle neck
problem, network congestion, bandwidth constraint, security,
single point of failure and service delay may occurs. To avoid
these issues there is a need of some decentralized architecture.
There are some existing decentralized architectures for large
scale peer to peer network , ,  however, security
and privacy issues are not considered in these systems.
From existing literature, we conclude that there is need of
some mechanisms for protecting the IoT devices from illegal
services and service provider in network computing.
Blockchain is growing list of records called as blocks.
These blocks are connected using cryptography. Each block
have cryptographic hash of previous block, a timestamp
at which block is generated and transaction data which is
represented in the form of merkle root tree. Invention of
blockchain technologies overcome the limitation of central-
ized architecture. Blockchain provides excellent features like
transparency, decentralized architecture, tamper proof system.
Blockchain is also used for secure service provisioning and
for data sharing systems. The excellent features of blockchain
technologies can ensure the secure service provisioning for
clients. However, use of blockchain technologies for IoT
constrained devices causes low latency, low throughput and
network delay like issues. Mostly existing blockchain systems
are designed for high computing and high storage devices.
Constrained devices are lacks from these resources.
Motivating by these existing systems, we proposed a se-
cure service providing architecture for IoT devices using
blockchain. Due to blockchain technologies, secure service
provisioning and involvement of secure edge computing de-
vices are possible. In our proposed architecture we maintain
the validity state of the edge servers depending on the service
provided to end users and rating given by end uses. Due to
involvement of blockchain technology the risk of malicious
activities in a network is eliminated. The use of smart contract
enhances the system performance as well as throughput of
the system. All the transaction occurs in network are stored
in cloud network layer.
A. Our Contributions
The main contributions of our proposed system are:-
•We proposed a novel blockchain system for lightweight
devices by considering their resource limitations. Valid-
ity states of the edge servers are stored in cloud servers,
to help the IoT devices for secure service provisioning.
•Smart contract is introduced for proposed system to
check the validity state of the edge servers and their
rating given by the end users. Instead of using Proof
of Work (PoW), We consider Proof of Authority (PoA)
consensus mechanism for ensuring low latency and high
throughput of proposed system.
•We introduced the cryptocurrency based incentive mech-
anism for edge servers. If the service provided by
the edge servers is valid then lightweight clients give
incentive to edge servers. If service provided by edge
servers is invalid then deﬁned amount of cryptocurrency
is deducted from the edge servers account.
•We deﬁne a speciﬁc threshold for validity states of
edge servers. Cryptocurrency and Rating given by edge
servers can determine the validity state of edge server.
IoTs industry grows remarkably in last few years. The
number of IoT devices increases rapidly and become 30
million in near future. The use of devices with cloud com-
puting, edge transparent computing extended the data sharing
and service provisioning of resource constrained devices.
However, existing systems can’t handle the security problems
of the network. The invention of blockchain technology 
solve security related issues by providing the feature of
openness, decentralization and tamper proof system. There are
few limitations of blockchain technology such as high com-
puting resources, data storage resources and enough power
supply. By keeping these issues in mind and motivated by
the existing systems mentioned above, we proposed a novel
service providing architecture to provide secure services to
C. Problem Statement
Yang et al. in  proposed a novel blockchain based ser-
vice provisioning mechanism to protect Light Weight Clients
(LWCs). Consortium blockchain is used in this paper with
POA consensus algorithm to achieve high throughput with
low latency of the system. A prototype is implemented based
on Ethereum to evaluate the effectiveness and security of
the proposed system. In proposed system adding more IoT
devices does not impose a great overhead. Hence, system
is scalable. However incentive mechanism and end user
feedback is not deﬁned in the system which leads to low
participation rate of edge servers. Proper mechanism for
veriﬁcation and validation of the server involved in network
is not considered.
The paper is organized as follows. In Section II the literature
review is described with limitations. Section III contains the
complete description of proposed system model. In Section
IV simulation results and discussions are considered and the
ﬁnal section V contains the conclusion.
II. LITERATURE REVIEW
From the existing literature we categorize the literature into
two main categories according to their domains.Below each
of the category is presented.
A. Blockchain In IoT
In  authors designed a hybrid network system for the
smart cities by combining the two most emerging technolo-
gies Software Deﬁned Network (SDN) and blockchain. To
obtain maximum efﬁciency from the system, authors divides
the network into two sub networks: core network and edge
network. Authors proposed consensus mechanism based on
Argon2 for hybrid network architecture to obtain security
and privacy. With the experimental analysis the effectiveness
of the system is evaluated. However, efﬁcient edge node
deployment and caching techniques at edge nodes are not
considered in this paper. Pardip et al. in  presents an
analysis of challenges that a large-scale IoTs network faced
due to new communication schemes. Distblocknet is pro-
posed architecture in which SDN and blockchain are used
for distributed secure IoT network. With the help of this
architecture model, overall system performance is increased.
The main aim of the model is to protect the system against
different threats. In the future authors want to implement this
model for cloud-fog based environment. Oscar in  presents
a Proof of Concept architecture that follows the blockchain
mechanism for access management in IoT. Other state-of-the-
art management systems are also evaluated in this article.
When there is a single management hub the system performs
less efﬁcient than the other existing systems. However, the
model in this research work is signiﬁcant scalable when
the load is distributed. The system is designed for multiple
management hubs where wireless sensor nodes are connected
to multiple hubs.
In  the main aim of the authors is to optimize the
computation complexity and storage complexity. The authors
introduced a new green consensus mechanism Proof of Col-
laboration (POC). With the help of this mechanism edge
device competes for new block generation by sharing their
collaboration credit instead of solving mathematical puzzles.
The authors propose a furtile transaction theory. Express
transaction and Hollow blocks to improve the efﬁciency of
the network. The proposed POC mechanism can reduce the
wastage of computational resources. However, the security
of POW is based on the wastage resource so the security
limitation is one of the factors involved in Polk.
B. Blockchain in WSN
In  a new concept of Rolling blockchain for Wireless
Sensor Networks (WSNs) is presented. In this paper authors
describe the limitation of WSNs and the model is presented
for how to implement blockchain in a WSN. POW mechanism
is impossible to implement in the WSNs due to lack of
storage. To overcome this problem a novel mechanism is
introduced. The simulation shows that if the network is denser
than node failure will be less affected, but if the network is
less dense then a small percentage of node failure turns to
network breakdown. Security analysis and protection against
hacking of the proposed system is not considered in this paper.
In  the two emerging technologies artiﬁcial intelligence
and blockchain is implemented together to design data sharing
framework. The data owners can only give permission of data
requirement. The authors leverage the features of data access
control and supervision measures to obtain a secure data shar-
ing architecture. In this article both data chain and behavior
chain are combined together to get a secure architecture. This
blockchain model for data sharing is designed for system level
if the data level increases a problem would occur.
Lin et al. in  present a framework of Device to
Device (D2D) based on cellular network architecture for
the authentication of Channel State Information (CSI). In
a data-intensive system D2D communication is not possible
for huge number of mobile end users. By the use of access
control system in the network is very beneﬁcial. The proposed
algorithm outperforms among “Q-learning algorithm”, which
is commonly used. The proposed algorithm enhanced spectral
efﬁciency without using a resource consuming consensus
mechanism of blockchain. This paper only considered non-
cooperative mobile users. Jiao et al. in  considers the
node failure problem in a network which leads to network
breakdown or sometimes useless. In this article authors con-
sider node failure problem and proposed a data transmission
scheme in which ﬁrstly, investigate the node failure and then
using greedy algorithm create concurrent communication tree
to organize the node data. Due to this technique overall
transmitting complexity of nodes increases. As a result, author
concludes that when 15 % node failure occurs, the network
remain stable but when 30 % node failure occurs, link stress
increased greatly. The simulations were performed on limited
network size. In future author wants to implement on real
In  ﬁrst incentive mechanism is built for WSNs by
using blockchain. The WSN node who stores the data is
awarded from digital currency. With increment in data size the
rewarded amount also increases. Two blockchains are used
in this paper. One is for data storage and other is for data
access. Provable Data Possession (PDP) is used in blockchain
instead of POW. Due to PDP the computation complexity
also decreases. The storage space is also decreased by using
C. Comparision With Existing Work
In , authors proposed a blockchain-based framework
for IoT. The main aim of the research is to provide services
for IoT. The network latency is considered as a performance
parameter. However, our proposed work focused on secure
service providing for IoTs. In , there is no secure ser-
vice sharing mechanism. In , author proposed software-
deﬁned networking enabled controller fog nodes to manage
the raw data generated by IoT devices. The fog node is
introduced to reduce the end to end delay. The proposed
architecture leads to the real-time response, high scalability
and low latency. However, energy-efﬁcient communication is
not considered. In , author proposed a Blockchain-based
Mobile Edge Computing sharing system. In proposed system
artiﬁcial intelligence infrastructure can offer sharing services
of IoT economically. Incentive mechanism is also introduced
to potentially support smart city network. However, there is no
service authentication mechanism. In , author proposed a
prototype-based EdgeChain framework. The proposed system
consists of a blockchain and a smart contract. Edge chain
links the edge nodes with IoT devices. With the help of the
edge chain, IoT devices are managed and enforce policies.
However, IoT proxy and heterogeneous IoT devices are not
considered. In our proposed system model, we implement
blockchain-based secure service providing an architecture
for IoT. However, in existing systems, incentive mechanism
and edge node reputation system are not presented. Due to
the lack of incentive mechanism, participation rate of edge
server become low. A proper mechanism for veriﬁcation and
validation of the server involved in the network is also not
III. PROP OS ED SY ST EM MO DE L
By motivated from system model of  we proposed
sytem model consists of four layered architecture in which
cloud servers,edge servers, LWC and blockchain mechanism
are placed. LWC are resource constrained devices with lim-
ited computation, storage and power resorces. The proposed
model consists of both legacy entities of edge transparent
computing and blockchain entities. Cloud servers are placed
in cloud layer which provides trusted service codes to edge
servers in an off chain manner where blockchain mechanism
is not involved. In the proposed model different cloud servers
are placed in cloud layers.These cloud servers have peer
to peer relation among them. Blockchain is implemented at
cloud layer among cloud servers to provide secure service
The main use of edge servers is to provide trusted service
codes to LWC with minimum delay. Edge servers are weak
service providers placed near to LWC.These servers are able
to get service codes from cloud servers and deliver to LWC
upon request. Frequently used sevice codes are stored in their
cache memory.Due to this, service codes are provided to LWC
with minimum delay. Lightweight clients are abstract form of
IoTs and consider as end users.
In proposed system model we introduce incentive mech-
anism for edge servers. After service validation process, if
the service provided by edge server is valid then incentive
will be given. If service is invalid a pre-deﬁned amount of
cryptocurrency is deducted from edge servers account. An-
other feedback mechanism is also implemented in proposed
system. In this mechanism end user give rating to edge server
for their service provided.Feedback mechanism and incentive
mechanism can determine the validity state of the edge server.
There are some security assumption in our proposed work
such as feedback given by LWC are always depend on validity
state of service codes. There will be no malacious activity in
term of feedback. In our proposed system incentive will be
given after delivering of valid service codes to IoTs. However,
there is no proper fair payement system which ensures proper
incentive given mechanism. In proposed system there is no
security mechanism for authentication of LWCs. We assume
that all LWC nodes are not a part of malacious network.
A. Blockchain Entities
There are two main entities of blockchain which are
described below: -
1) Cloud Nodes (CNs): CNs are nodes with sufﬁcient
resources used in blockchain process. The CNs are privileged
nodes in the blockchain network which maintain the dis-
tributed ledger with smart contract records. CNs are respon-
sible for adding block in blockchain, validation of transaction
and execution of smart contract. Cloud servers act as CNs. All
the CNs follows PoA consensus mechanism. After reaching at
51% majority rule position, transaction is added in blockchain
through cloud servers.
Fig. 1: Blockchain based Secure Service Providing Mechanism for IoTs.
2) Lightweight Node (LN): LN are less privileged nodes in
network and only read the information in the blockchain and
check the validity of the edge service provider and service
codes. Each LWC also acts as a LN in proposed blockchain.
B. WorkFlow of Proposed Sytem Model
In proposed blockchain architecture, LWCs requests a
service code to near placed edge server. Frequently used
service codes are stored in cache memory of edge servers.
Edge servers entertain the request by checking their cache
memory. If required service codes are not present in cache
memory then edge server request to cloud servers. Blockchain
is implemented among cloud servers. Cloud servers perform
PoA consensus mechanism and provide secure service codes
to edge servers. Validity and reputation rating of edge servers
is stored in blockchain. Reputation rating is provided by end
users. For protecting LWC from untrusted edge server smart
contract is triggered and ﬁgure out the validity of edge servers.
To determine the validity of service codes, LWC communi-
cate with cloud servers and get the hash against service codes.
After this, LWC generate a hash of service codes which is
given by edge servers. Now LWC compare the hashes of both
service codes and validate the service codes as well as edge
server. If both hashes are same then transaction is valid and
edge servers are also not a part of malicious network. All
the validated states of edge servers are updated in blockchain
network. After achieving valid results, LWC give some in-
centive to edge servers. Incentive will be given to encourage
the participation rate of edge servers. Incentive is given in the
form of cryptocurrency. The incentive given to edge servers
can be exchangeable to local currency. Incentive amount is
depending on type of service required. Service rating is also
stored in blockchain given by end users. Depending on the
rating and validated state of edge server, cloud servers can
take decision whether particular edge server is part of network
or not. With the help of blockchain, security is achieved with
low performance overhead. The proposed model is shown in
IV. RES ULT S AN D DISCUSSIONS
In this section, we discuss the outcomes of our pro-
posed system. In blockchain gas is the unit to calculate
how much transactions are executed. Transaction is set of
action performed is ethereum environment. Every operation
performed in the blockchain have some gas consumption.The
transaction which is more resource consuming can have high
gas consumption than normal.There are some pre deﬁned gas
consumption rates mentioned in ethereim yellow paper .
1 gas unit = 4 gwei (1 ETH = 1000000000 gwei).
In the ﬁg.2 we plot various function and their gas con-
sumption accordingly.As in Fig.2 shows register edge server
in a network can cost more than others.This is due to
more network bandwidth usuage and more information again
each edge server is required.We introduce new concept in a
network design.Incentive given to edge servers for their par-
ticipation in a network and providing services to LWCs.The
second most gas consuming fuction is service validation
function. In this function the services given to LWC by edge
servers and services present in cloud servers are validated.
Whether the service provided by the edge servers are valid or
not. In table 2 the name of the events are mentioned with their
gas consumed values. In ﬁg.2 we used name of the events as
alphabets however, “a” is used for register edge devices,“b”
is used for service request,“c” is used for service response,
“d” is used for service validation, “e” is for Incentive given
and rating is shown by “f”.
Fig. 2: No. of events executed with respect to time taken.
In the ﬁg 3, we perform the comparison between two
parameters: Time for service request mined by the CNs and
number of transactions executed. As ﬁg 3 shows, as the
number of transaction increases, service request validation
time also increases linearly. The time taken to execute and
mined transaction is increases with type of service demand.
The service which is more resource consuming can require
TABLE I: No. of events executed with respect to time taken.
Event Gas Consumed
Register edge server 148107
Service request 43109
Service response 65457
Service validation 112399
Fig. 3: No. of transaction mined and executed with respect
TABLE II: No. of transaction mined and executed with
respect to time.
No. of Transaction Time (sec)
more time to perform execution and mining process. When
LWC request edge servers for the services, edge server can
check their cache for required service response, if service
is present than response time is minimum and execution of
service takes minimum time. Moreover, edge servers can
communicate with cloud servers for service response. In this
process more time is required to fulﬁll the end user require-
ment. To minimize the delay and enhance the efﬁciency there
is need of more edge servers to fulﬁll the requirements. In
table 3 we present results in tabular form.
In the ﬁg 4, we perform analysis between the gas consump-
tion of various services. We take 10 transaction having differ-
ent services required by the end nodes. The gas consumption
Fig. 4: Gas Consumed with respect to Service Response and
of services depends on two main factors: one is the service
size and other is difﬁculty level of hash. As the required
service is resource consuming and utilize more bandwidth
then gas consumption is also high than normal value. The
second one is hash difﬁculty, if the hash generated by hashing
algorithm have high complexity than more mining power is
required and execution time also increases. Depending on
these two factors gas consumption is based. In table 4, we
present the gas consumption in tabular form. In Fig. 5, we
TABLE III: Gas consumption with respect to Request and
Response of Services.
No. of Transaction Service Request Service Response
T1 43493 65393
T2 28942 31299
T3 33452 35134
T4 21678 27451
T5 34567 36717
T6 37810 38967
T7 19781 23980
T8 34598 39098
T9 49876 53245
T10 32980 35678
evaluate the comparison between number of edge devices
participate in a network and their gas consumption. At the
start of network when there is small number of edge device in
network, their gas consumption is not much high. We simulate
our proposed system up to 100 edge devices. The result
shows that, there is linear change in the gas consumption.
Thus, proposed system is scalable enough. With scalability
our system also achieve security. In table 5, we present the
V. C ONCLUSIONS
In this paper, we proposed a secured service providing
mechanism for IoT devices. In the proposed system we also
Fig. 5: Gas Consumed with respect to edge devices.
TABLE IV: Gas Consumption with respect to Edge devices.
No. of Edge Devices Gas Consumption
consider computing technologies such as cloud computing
and edge transparent computing. To protect the IoT devices
from the malicious edge servers we introduce blockchain
technology. The validity of edge servers is maintained by the
use of smart contract. Service provided by the edge servers is
also veriﬁed by the IoT devices. By considering the resource
constrained IoT devices we design blockchain system with
minimum resources consumption. Incentive mechanism and
reputation system is introduced for edge service providers.
Depending on rating and cryptocurrency account value, valid-
ity states of edge servers are determined. The rating given by
the end nodes is considered as a feedback of service provided
by edge servers. From results, it is to be concluded that our
proposed system is suitable for lightweight devices.However,
the gas consumption is depending on the size of transaction.
In our proposed system model there are some limitations,
such as MAC spooﬁng attacks are not considered for egde
service providers. We proposed a secure mechanism at both
layers cloud and egde server layer. Moreover, there is also
need of security mechanism at LWC layer. This will also
verify the reputation system given by end users.
 Middleton, Peter, Peter Kjeldsen, and Jim Tully. ”Forecast: The internet
of things, worldwide, 2013.” Gartner Research (2013).
 Triantaﬁllou, Peter, Nikos Ntarmos, S. Nikoletseas, and P. Spirakis.
”NanoPeer networks and P2P worlds.” In Proceedings Third International
Conference on Peer-to-Peer Computing (P2P2003), pp. 40-46. IEEE,
 Ali, Muneeb, and Zartash Afzal Uzmi. ”CSN: A network protocol for
serving dynamic queries in large-scale wireless sensor networks.” In
Proceedings. Second Annual Conference on Communication Networks
and Services Research, 2004., pp. 165-174. IEEE, 2004.
 Krco, Srdjan, David Cleary, and Daryl Parker. ”P2P mobile sensor
networks.” In Proceedings of the 38th Annual Hawaii International
Conference on System Sciences, pp. 324c-324c. IEEE, 2005.
 Nakamoto, Satoshi. ”Bitcoin: A peer-to-peer electronic cash system.”
 Xu, Yang, Guojun Wang, Jidian Yang, Ju Ren, Yaoxue Zhang, and Cheng
Zhang. ”Towards Secure Network Computing Services for Lightweight
Clients Using Blockchain.” Wireless Communications and Mobile Com-
puting 2018 (2018).
 Sharma, Pradip Kumar, and Jong Hyuk Park. ”Blockchain based hybrid
network architecture for the smart city.” Future Generation Computer
Systems 86 (2018): 650-655.
 Sharma, Pradip Kumar, Saurabh Singh, Young-Sik Jeong, and Jong
Hyuk Park. ”Distblocknet: A distributed blockchains-based secure sdn
architecture for iot networks.” IEEE Communications Magazine 55, no.
9 (2017): 78-85.
 Novo, Oscar. ”Scalable Access Management in IoT using Blockchain: a
Performance Evaluation.” IEEE Internet of Things Journal (2018).
 Xu, Chenhan, Kun Wang, Peng Li, Song Guo, Jiangtao Luo, Baoliu
Ye, and Minyi Guo. ”Making big data open in edges: A resource-
efﬁcient blockchain-based approach.” IEEE Transactions on Parallel and
Distributed Systems 30, no. 4 (2019): 870-882.
 Kushch, Sergii, and Francisco Prieto-Castrillo. ”A Rolling Blockchain
for a Dynamic WSNs in a Smart City.” arXiv preprint arXiv:1806.11399
 Zhang, Guozhen, Tong Li, Yong Li, Pan Hui, and Depeng Jin.
”Blockchain-based data sharing system for ai-powered network oper-
ations.” Journal of Communications and Information Networks 3, no. 3
 Lin, Di, and Yu Tang. ”Blockchain Consensus Based User Access
Strategies in D2D Networks for Data-Intensive Applications.” IEEE
Access 6 (2018): 72683-72690.
 Li, Jiao. ”Data Transmission Scheme Considering Node Failure for
Blockchain.” Wireless Personal Communications 103, no. 1 (2018): 179-
 Ren, Yongjun, Yepeng Liu, Sai Ji, Arun Kumar Sangaiah, and Jin Wang.
”Incentive Mechanism of Data Storage Based on Blockchain for Wireless
Sensor Networks.” Mobile Information Systems 2018 (2018).
 Wood, Gavin. ”Ethereum: A secure decentralised generalised transac-
tion ledger.” Ethereum project yellow paper 151 (2014): 1-32.
 Samaniego, Mayra, and Ralph Deters. ”Blockchain as a Service for
IoT.” In 2016 IEEE International Conference on Internet of Things
(iThings) and IEEE Green Computing and Communications (GreenCom)
and IEEE Cyber, Physical and Social Computing (CPSCom) and IEEE
Smart Data (SmartData), pp. 433-436. IEEE, 2016.
 Sharma, Pradip Kumar, Mu-Yen Chen, and Jong Hyuk Park. ”A soft-
ware deﬁned fog node based distributed blockchain cloud architecture
for IoT.” IEEE Access 6 (2017): 115-124.
 Rahman, Md Abdur, Md Mamunur Rashid, M. Shamim Hossain, Elham
Hassanain, Mohammed F. Alhamid, and Mohsen Guizani. ”Blockchain
and IoT-based cognitive edge framework for sharing economy services
in a smart city.” IEEE Access 7 (2019): 18611-18621.
 Pan, Jianli, Jianyu Wang, Austin Hester, Ismail Alqerm, Yuanni Liu, and
Ying Zhao. ”Edgechain: An edge-iot framework and prototype based on
blockchain and smart contracts.” IEEE Internet of Things Journal (2018).