Enabling Distributed Intelligence in the Internet of Things using the

IOTA Tangle Architecture

Tariq Alsboui, Yongrui Qin and Richard Hill

School of Computing and Engineering, University of Huddersﬁeld, U.K.

Keywords:

Internet of Things, Distributed Intelligence, IOTA, Mobile Agent.

Abstract:

It is estimated that there will be approximately 26 to 30 billion Internet of Things (IoT) devices connected

to the Internet by 2020. This presents research challenges in areas such as data processing, infrastructure

scalability, and privacy. Several studies have demonstrated the beneﬁts of using distributed intelligence to

overcome these challenges. This article reviews existing state-of-the-art distributed intelligence approaches in

IoT and focuses on the motivations and challenges for distributed intelligence in IoT. We propose a potential

solution based on IOTA (Tangle), a platform that enables highly scalable transaction-based data exchange

amongst large quantities of smart things in a peer-to-peer manner, together with mobile agents to support

distributed intelligence. Challenges and future research directions are also discussed.

1 INTRODUCTION

The Internet of Things (IoT) is a mature ﬁeld of re-

search that was brought to attention by Auto-ID cen-

tre, when they used Electronic Product Code (EBC)

along with Radio Frequency Identiﬁcation (RFID) to

automatically identify the and track the itinerary of

items in supply chain (Ashton, 2009). IoT is consid-

ered as a novel paradigm that connects physical ob-

jects to the Internet. The basic idea behind it is to

connect physical objects to the virtual world and al-

low them to sense and modify the environment by us-

ing sensors and actuators (Atzori et al., 2010). Con-

necting the physical world to the Internet plays a cru-

cial role in enhancing our lives by turning cities into

smart cities (Perera et al., 2017), homes into smart

homes (Doan et al., 2018), and campuses into smart

campuses (Angelis et al., 2015). According to sev-

eral research reports, it is estimated that there will be

approximately 26 to 30 billion devices connected to

the Internet in 2020 (Gartner, 2013; Research, 2013).

Consequently, this in turn brings many challenges in

a number of areas, such as data processing, saving re-

sources, and scalability (Esposito et al., 2017).

One of the key technologies being explored for

overcoming many of the challenges associated with

the growing number of connected IoT devices is dis-

tributed intelligence (Byers and Wetterwald, 2015).

Distributed intelligence is deﬁned as a system of en-

tities e.g., smart sensors, working together to reason,

plan, and solve problems (Lynne, 2007). The main

aim of such technology is to enable entities (smart

objects) in an IoT system to cooperate at optimal efﬁ-

ciency to achieve desired goals.

In the context of IoT, according to (Van den

Abeele et al., 2015), distributed intelligence is de-

ﬁned as Cooperation between devices, intermediate

communication infrastructures (local networks, ac-

cess networks, global networks) and or cloud sys-

tems in order to optimally support IoT communication

and IoT applications. As stated in (Van den Abeele

et al., 2015) in order to enable distributed intelligence,

communication and computation capability should be

placed at the right place.

Based on the above deﬁnition, this paper presents

a new scalable, and energy efﬁcient distributed in-

telligence approach for IoT. We propose the utiliza-

tion of IOTA Tangle architecture (Serguei, 2017) and

Mobile Agent (Lepp

anen et al., 2014) to enable dis-

tributed intelligence. IOTA is an emerging platform

that is particularly designed for the Internet of Things

to overcome the problems of scalability, transaction

fees, and mining of the blockchain technology. The

main component of IOTA is the Tangle, which is

based on the concept of a Directed Acyclic Graph

(DAG)(Serguei, 2017). The IOTA platform provides

a potential and highly scalable solution to enable

distributed intelligence. The mobile agent technol-

ogy can provide cooperation and information shar-

ing among different types of nodes (Lepp

anen et al.,

392

Alsboui, T., Qin, Y. and Hill, R.

Enabling Distributed Intelligence in the Internet of Things using the IOTA Tangle Architecture.

DOI: 10.5220/0007751403920398

In Proceedings of the 4th International Conference on Internet of Things, Big Data and Security (IoTBDS 2019), pages 392-398

ISBN: 978-989-758-369-8

2014). A description about the architecture is pre-

sented in Section 4.

The remainder of this paper is structured as fol-

lows: Section 2 identiﬁes the motivation and chal-

lenges behind the need for distributed intelligence in

IoT. Section 3 presents an overview of the recent dis-

tributed intelligence approaches in IoT. In Section 4

a summary on future research direction is discussed.

Finally, Section 5 concludes the paper.

2 MOTIVATIONS AND

CHALLENGES

In this section, we present the need for distributed in-

telligence in the Internet of things (IoT) by enlisting

some of the key factors that dictate the challenges in

IoT related research. We then brieﬂy describe how

IOTA platform can be utilized to realize distributed

intelligence.

2.1 Saving Resources

It is generally expected that IoT would produce a mas-

sive amount of data, which when sent to a central lo-

cation results in larger consumption of network re-

sources. IoT consists of nodes that has limited re-

sources such as power consumption (battery limits),

computational capability and maximum memory stor-

age, which makes distributed intelligence a challeng-

ing task. IOTA Qubic protocol (Foundation, 2016b)

saves resources by outsourcing intensive computa-

tions to an external more powerful nodes. This is

can be achieved through Qubic-enabled IOTA nodes

(Q. Nodes). Qubics are inserted as messages in IOTA

transactions. It consists of instructions, called meta-

data responsible for deciding how and when to pro-

cess data.

2.2 Scalability

Scalability refers to the ability of the network to deal

with the growing amount of work needed when the

network grows. It can be divided into two parts: Hor-

izontal scaling and Vertical scaling. In Horizontal

scaling, the network is expected to grow by adding

more nodes to it. On the other hand, Vertical Scal-

ing is to equip the existing devices in the network

by adding more (CPU, RAM, power) (Bondi, 2000).

IoT is constantly changing and developing to ﬁt in

environmentally in order to deal constantly with en-

larging demands and in accordance with the predic-

tions provided in (Gartner, 2013; Research, 2013).

Hence, potential solutions should be highly scalable

to deal with billions of smart objects, which will be

soon connected to the Internet.IOTA Tangle (Serguei,

2017) can provide a valuable solution to accommo-

date the fast growth of interconnected things. IOTA

Tangle scales well when the number of Tangle nodes

increases.

2.3 Privacy

Privacy refers to the capability of a system to keep

information/data private, e.g, to make sure that if any-

one has accessed the data will be unable to make sense

of it. Moreover. Information leakage is generally the

ultimate user concern, especially relating to sensitive

data, such as location, and movement trajectory in-

formation. Potential solutions should identify in what

form the data should be, and who can get access to it.

Consequently, the IOTA Masked Authenticated Mes-

saging (MAM) protocol (Foundation, 2016a) can be

utilized for achieving privacy. An example where pri-

vacy is of concern for users is in health care applica-

tions where information about patients is sensitive.

2.4 Ofﬂine Capability

Ofﬂine Capability is also known as resiliency and is

often deﬁned as the capability of the system, to work

in emergency cases, such as Internet connection not

reachable. This indicates that if the internet connec-

tion goes down, the system will not function. There-

fore, there is no need for a network to be connected to

the Internet all the time. IOTA tasks can be done on

an ofﬂine network. IOTA Tangle offers this capabil-

ity, but the transactions have to be re-attached to the

main tangle if further processing is needed. In such a

manner, distributed intelligence and processing is de-

sirable and well supported.

3 EXISTING DISTRIBUTED

INTELLIGENCE APPROACHES

IN IoT

Over the last few years, distributed intelligence has

started to gain attention from many researchers in

the ﬁeld of IoT (Van den Abeele et al., 2015; Byers

and Wetterwald, 2015; Sahni et al., 2017; Rahman

and Rahmani, 2018). Most of these research efforts

are to deal with problems relating to data process-

ing, data management, scalability, and privacy. Re-

cently, the authors in (Van den Abeele et al., 2015)

introduced the concept of Sensor Function Virtualiza-

tion (SFV) as a potential technique to support dis-

Enabling Distributed Intelligence in the Internet of Things using the IOTA Tangle Architecture

393

Table 1: Comparison Among Distributed Intelligence Approaches in IoT.

Distributed Intelligence Approaches Saving Resources Scalability Privacy Ofﬂine Capability

(Van den Abeele et al., 2015) High High Low High

(Byers and Wetterwald, 2015) High High Low Low

(Sahni et al., 2017) High High Medium Medium

(Rahman and Rahmani, 2018) High Low Low Low

tributed intelligence in IoT. The basic idea is to en-

able distributed processing of certain functionalities

by ofﬂoading them from constrained devices to un-

constrained infrastructure such as a virtualized gate-

way, the cloud and other in-network infrastructure.

SFV focuses on the three main points including, scal-

ability, heterogeneity of the IoT, and transparency. To

achieve scalability, the approach relies on cloud in-

frastructure by allowing part of SFV functionalities

to run on the cloud beneﬁting from the elasticity pro-

vided by the cloud, and tired design. this handles the

increased load, when devices are joining the network.

The second point is related to heterogeneity of IoT

in terms of resources constraints devices, i.e., lim-

ited power, limited processing, infrastructure and it

should shift the user from the low level details of the

devices. The ﬁnal point is related to transparency in

which any virtual functions that are added to the de-

vices must build on top of existing communication in-

terfaces and that changes to protocols running on end

devices must be minimal and preferably non-existent.

The advantages of their approach are reduction in en-

ergy consumption, scalability, ﬂexibility, and trans-

parency. However, security and privacy issues are

brieﬂy acknowledged in their approach. Moreover,

implementation and evaluation of the proposed ap-

proach is not provided in which they outline as part

of their future work.

The work by (Byers and Wetterwald, 2015)

presents fog computing architecture as a solution to

enable distributed intelligence in IoT. The proposed

approach described fog nodes in terms of hardware

architecture as well as software architecture. From

a hardware point of view, fog nodes can be imple-

mented as ancillary functions on traditional network

elements such as gateways, edge devices, and appli-

ances or as stand-alone fog boxes. From a software

point of view, fog nodes are highly virtualized ma-

chines with multiple VMs running under a highly ca-

pable hypervisor. The beneﬁts of using fog nodes are

to enhance reliability, bandwidth, and security. How-

ever, fog computing still has issues regarding security,

privacy (Esposito et al., 2017; Yi et al., 2015; Gillam

et al., 2018). In addition to that, implementation and

evaluation of the proposed approach is not provided.

More recently, a new computing paradigm, called

Edge Mesh that aims to enable distributed intelligence

in IoT is proposed in (Sahni et al., 2017). The pro-

posed paradigm distributes the decision-making tasks

among edge devices within the network rather than

transferring all the data to a centralized server for

further processing. In Edge Mesh, all the computa-

tion tasks and data are shared using a mesh network

of edge devices and routers. Edge Mesh architec-

ture consists of four main types of devices. First,

end devices are mainly used for sensing and actuat-

ing. Second, edge devices are used for processing

and connecting with end devices. Third, routers are

used for transferring data among edge devices. Fi-

nally, cloud is used to perform big data analytics on

historical data. The advantages of edge mesh are, dis-

tributed processing, low latency, fault tolerance, bet-

ter scalability, better security, and privacy. However,

they have component for achieving security and pri-

vacy, but how privacy can be achieved is not consid-

ered. Furthermore, implementation and evaluation is

not provided.

Different from the above, the work presented

in (Rahman and Rahmani, 2018) proposed an

AI based distributed intelligence assisted approach

named as Future Internet of Things Controller

(FITC). The proposed approach uses both edge and

cloud based to distribute intelligence. In particular,

edge controller is used to provide low-level intelli-

gence and cloud based controller to provide high-level

intelligence, which they refer to as distributed intelli-

gence. The beneﬁts of their work are to reduce re-

sponse time and loosen the requirements for rules.

However, the approach lack of a mechanisms that en-

ables privacy, and ofﬂine capability.

Table 1 provides a comparison of the reviewed

distributed intelligence approaches according to the

challenges provided in Section 2. Overall there are

many pieces of solutions to enable distributed intelli-

gence in IoT. We compare the approaches and ranked

as High, Medium, and Low based on potentiality of

tackling the identiﬁed technical challenges. A ﬁeld

in the table is given a rank of High if the approach

satisﬁes the challenge corresponding to that column.

The approach is ranked with Medium if it supports the

challenge, but not providing a way of how to achieve

it. Low is given to the approach if it does not address

IoTBDS 2019 - 4th International Conference on Internet of Things, Big Data and Security

394

the challenge at all.

3.1 Limitations of Prior Work

From the above we can see that most of the existing

approaches to enabling distributed intelligence in IoT

suffer from inherent problems. Firstly, they rely on

centralized architecture for processing data (Gillam

et al., 2018), which introduces a high cost and de-

lay that is not acceptable for distributed applications.

Such examples include health monitoring, emergency

response, autonomous driving, and so on. In addi-

tion to that, it would consume much network band-

width (Perera et al., 2017). Besides, solutions based

on fog computing still have issues regarding security

and privacy (Esposito et al., 2017). Moreover, there is

a need for a standardized way for describing the data

generated by IoT, such as the one promised by IOTA

Identity of Things (IDoT) (Foundation, 2016a), which

will also help secure the network. Another problem is

the lack of a mechanism to describe in what form the

data should be, and who can get access to it (multi-

party authentication scenarios), all of which are re-

lated to privacy. Finally, only a few of the approaches

facilitate an implementation and evaluation of their

proposed approach.

4 A NOVEL ENABLING

APPROACH FOR

DISTRIBUTED INTELLIGENCE

IN IoT

The problems and limitations presented above lead

to future research opportunities. Possible solutions

for enabling distributed intelligence in IoT can be

achieved through the use of the IOTA platform (Ser-

guei, 2017) and mobile agent technology (Lepp

anen

et al., 2014).

4.1 Fundamental Tools and Techniques

IOTA (a.k.a. distributed ledger (Serguei, 2017)) is an

emerging platform that is particularly designed for the

internet of things to overcome the problems of scala-

bility, transaction fees, and mining, which is consid-

ered as a resource extensive task that other crypto-

currency lacks by utilizing the blockchain technol-

ogy (Nakamoto et al., 2008). The main component

of IOTA is the Tangle, which is based on the concept

of a Directed Acyclic Graph (DAG) (Serguei, 2017).

Tangle is the protocol, or the data structure used in

IOTA in order to store the transactions, which has col-

Table 2: Node Types in IOTA Network.

Node Type Storage Validation

Full Node Full Tangle last Snapshot Yes

Light Node None No

PermaNode Full Tangle Permanently Yes

lection of nodes (also called as Sites or Vertices) and

arrows (also called as Edges). All the vertices or sites

which hold data (transactions) are connected to one

another using edges, and these edges are used to val-

idate the transaction and to check whether it is valid

transaction, in order to achieve approval or conﬁrma-

tion of the transaction eventually. Edges can range

from a minimum of two to a maximum of many and

are called as Parent. If there is a site with less than two

edges, it represents that the actions are unconﬁrmed,

and these are called as Tips of the tangle. Genesis is

the unique site or the very ﬁrst site which do not have

any previous site or parents (Serguei, 2017). The Tan-

gle architecture through several nodes QubicNodes,

Full Nodes, Light Node, and Masked Authentication

Messaging. As discussed in Section 2 will be utilized

to achieve distributed intelligence in IoT.

Table 2 Describes the features of the participant’s

nodes of IOTA network. the following paragraphs de-

scribes each participants node in the network

The concept of IOTA Full Node (Foundation,

2016a), which can be deﬁned as node within the tan-

gle architecture that is capable of ﬁnding neighbours

and communicate with them, attaching data to the tan-

gle, bundling and signing, tip selection, validation,

PoW, and attaching data to the tangle. IOTA full

nodes have high computational capacity and are re-

sponsible for doing the PoW on behalf of the end

nodes taking into consideration that end nodes have

limited resources.

The concept of Light Node (Foundation, 2016a)

that participates in the network and can be deﬁned as

a node within the tangle architecture that relies on the

full node to interact with the Tangle; it distinguishes

itself from other nodes in the sense that it does not

store a copy of the tangle, and does not either validate

transactions or communicate with neighbors. It has

been speciﬁcally designed as a lightweight node for

resource constrained nodes.

The concept of Qubic protocol (Foundation,

2016b), which is under development and is deﬁned as

a protocol that describes IOTA’s solution for quorum-

based computations. Qubic focuses on three types of

computations including: oracle machines, outsource

computations, and smart contracts. We are mainly

concerned with the outsourcing part to save resources

of IoT devices. Qubic is optimized for IoT and

makes it possible to ofﬂoad computations function-

Enabling Distributed Intelligence in the Internet of Things using the IOTA Tangle Architecture

395

alities from nodes with battery-limits to an external

more powerful nodes, which in turns reduces energy

consumption. Qubic are inserted as messages in IOTA

transactions. It consists of instructions, called meta-

data responsible for deciding how and when to pro-

cess them.

The concept of Permanodes (Foundation, 2016a),

which is under development and is deﬁned as node

within the IOTA tangle architecture that has features

including: ﬁnding a neighbouring nodes and commu-

nicate with them, attaching data to the tangle, tip se-

lection, and do the Proof-of-Work (PoW). The PoW is

a short computational operation compared to mining

in blockchain. Furthermore, these nodes are distin-

guishable from other nodes by having the capability

of storing the whole tangle data permanently. This is

beneﬁcial for some of the IoT applications in which

access to the full data history is required.

In order to ensure privacy, IOTA developed a

protocol called Masked Authenticating Messaging

(MAM) (Handy, 2017). In the tangle, each transac-

tion carries a message, which allows these messages

to be exchanged between the nodes. This indicates

that anyone can view the messages in the network.

MAM utilizes the Merkle tree based signature scheme

and enables privacy by encrypting data. Encryption

occurs through three techniques including: private

mode, public mode, restricted mode, thereby enabling

privacy.

The mobile agent (MA) technology can provide

cooperation and information sharing among different

types of nodes (Lepp

anen et al., 2014). Mobile agent

is deﬁned as a piece of software that performs data

processing autonomously while moving from node

to node in the network (Alsboui et al., 2017). The

agent can collect local data and perform any neces-

sary data aggregation. Mobile agents can make de-

cision autonomously without user input. They pro-

vide ﬂexibility in terms of decision making, and reli-

ability in terms of node failure. (Lange and Oshima,

1999) listed seven good reasons for using MA such as,

reducing network load, they adapt dynamically, and

They execute asynchronously and autonomously.

4.2 A New Enabling Approach

Based on the above tools and techniques, we pro-

pose an approach that enables distributed intelligence

in IoT while taken into consideration the challenges

identiﬁed in Section 2.

As the number of the Internet of Things nodes are

expected to reach 30 billion devices in 2020, any pro-

posed solution should be highly scalable, and energy

efﬁcient to deal with billions of smart objects that will

be connected to the internet. Aiming at this, two im-

portant points are to be noted. Firstly, by ofﬂoading

functionality using IOTA Qubic from resource con-

straints nodes i.e., battery limit to a more powerful un-

constrained infrastructure, we expect our solution to

take advantage of the mechanisms provided by these

environments to save resources i.e., power consump-

tion, and storage. Secondly, towards a scalable ap-

proach, IOTA scales well when the number of nodes

are added. This requires only the veriﬁcation of two

previous transactions by signing nodes with a private

keys, apply tip selection using Random Walk Monte-

carlo Algorithm (RWMC), and Proof-of-Work (PoW)

to solve the cryptographic puzzle. If the PoW is heavy

on nodes with limited processing, the PoW can be di-

rected to the IOTA full node.

Another ﬁnal impotent point is to ensure pri-

vacy by employing Masked Authenticated Messaging

(MAM). MAM is offered by IOTA as an extension

module, which acts as a second layer data communi-

cation protocol to encrypt or mask data. This means

that IoT edge nodes running the MAM client are able

of transmitting encrypted sensor data by using this

communication protocol. This fulﬁlls a crucial need

in IoT applications, i.e., health care in which access

and privacy meet.

Figure 1: A New Distributed Intelligence Approach for IoT.

Having in mind the requirements of the previous para-

graphs, our proposed approach is presented in Figure

1. Our approach consists of three layers. In the ﬁrst

layer, which comprises end nodes running an IOTA

light client and will act as an end points to the IOTA

network. Since there will be no interactions among

nodes in the network running IOTA light clients, we

employ MA to facilitate cooperation among nodes in

IoTBDS 2019 - 4th International Conference on Internet of Things, Big Data and Security

396

the network and by cooperation we mean data shar-

ing. To do so, MA will carry transactions that con-

tains the data and aggregate the data. Furthermore,

by dispatching MA, we reduce the amount of sensory

data by eliminating redundancy. For example, nodes

that are placed in proximity of each other are likely to

generate redundant data. Ultimately, data aggregation

is required to reduce the data trafﬁc in the network.

Second Layer, comprise of a less unconstrained

devices running the IOTA full Node. Finally, the ap-

plication layer utilize from the proposed approach in

terms of energy efﬁciency, scalability, and privacy. To

this end, a number of questions should be tackled in

the future: How is data ofﬂoaded? What kind of prim-

itive mode will be used? How is the PoW outsourced

to IOTA full Node? How MA will be routed between

nodes in an energy efﬁcient manner?

5 CONCLUSIONS

IOTA Tangle architecture has the potential of enabling

distributed intelligence in IoT, which will be beneﬁ-

cial for a wide-range of applications, such as smart

cities, and healthcare. IOTA Tangle allows for dis-

tributed computation, making it suitable for enabling

distributed intelligence in IoT. In this article, we have

discussed the need for distributed intelligence in IoT

as well as how the IOTA platform can be utilized to

enable it. We have presented a review of the recent

state of the art distributed intelligence approaches in

IoT. Also, we have discovered that there is a need for

a a lightweight solution for enabling distributed intel-

ligence based on the identiﬁed limitations of existing

approaches. We have discussed the challenges as well

as future research directions in developing a new dis-

tributed intelligence approach and we believe that the

integration of IOTA and mobile agent would solve the

problems. Finally, we outline pathways to solutions

to the problems identiﬁed and envision that an IOTA

Tangle architecture will facilitate distributed intelli-

gence in IoT.

In the authors opinion, MA migrates between

nodes in the network and facilitates cooperation be-

tween them. This leads to several advantages, such as

reduction in the network load by moving MA carrying

data instead of sending data to a central location for

further processing, and overcomes network latency.

REFERENCES

Alsboui, T., Alrifaee, M., Etaywi, R., and Jawad, M. A.

(2017). Mobile agent itinerary planning approaches in

wireless sensor networks- state of the art and current

challenges. In Maglaras, L. A., Janicke, H., and Jones,

K., editors, Industrial Networks and Intelligent Sys-

tems, pages 143–153, Cham. Springer International

Publishing.

Angelis, E. D., Ciribini, A., Tagliabue, L., and Paneroni,

M. (2015). The brescia smart campus demonstrator.

renovation toward a zero energy classroom building.

Procedia Engineering, 118:735–743.

Ashton, K. (2009). That ’Internet of Things’ Thing. RFID

Journal.

Atzori, L., Iera, A., and Morabito, G. (2010). The internet of

things: A survey. Computer Networks, 54(15):2787–

2805.

Bondi, A. B. (2000). Characteristics of scalability and their

impact on performance. In Workshop on Software and

Performance, pages 195–203.

Byers, C. C. and Wetterwald, P. (2015). Fog computing dis-

tributing data and intelligence for resiliency and scale

necessary for iot: The internet of things (ubiquity

symposium). Ubiquity, 2015(November):4:1–4:12.

Doan, T. T., Safavi-Naini, R., Li, S., Avizheh, S., K., M. V.,

and Fong, P. W. L. (2018). Towards a resilient smart

home. In Proceedings of the 2018 Workshop on IoT

Security and Privacy, IoT S&P ’18, pages 15–21,

New York, NY, USA. ACM.

Esposito, C., Castiglione, A., Pop, F., and Choo, K. R.

(2017). Challenges of connecting edge and cloud

computing: A security and forensic perspective. IEEE

Cloud Computing, 4(2):13–17.

Foundation, I. (2016a). Iota development roadmap. (1).

(visited on 2-01-2019).

Foundation, I. (2016b). The qubic protocol. (1). (visited on

2-1-2019).

Gartner (2013). Gartner says the internet of things installed

base will grow to 26 billion units by 2020. (1).

Gillam, L., Katsaros, K., Dianati, M., and Mouzakitis,

A. (2018). Exploring edges for connected and au-

tonomous driving. In IEEE INFOCOM 2018 - IEEE

Conference on Computer Communications Workshops

(INFOCOM WKSHPS), pages 148–153.

Handy, P. (2017). Introducing masked authenticated mes-

saging. (1). (visited on 6-01-2019).

Lange, D. B. and Oshima, M. (1999). Seven good reasons

for mobile agents. Commun. ACM, 42(3):88–89.

Lepp

anen, T., Riekki, J., Liu, M., Harjula, E., and Ojala,

T. (2014). Mobile Agents-Based Smart Objects for

the Internet of Things, pages 29–48. Springer Inter-

national Publishing, Cham.

Lynne, P. (2007). Distributed intelligence: Overview of the

ﬁeld and its application in multi-robot systems. In The

AAAI Fall Symposium Series. AAAI Digital Library.

Nakamoto, S. et al. (2008). Bitcoin: A peer-to-peer elec-

tronic cash system.

Perera, C., Qin, Y., Estrella, J. C., Reiff-Marganiec, S., and

Vasilakos, A. V. (2017). Fog computing for sustain-

able smart cities: A survey. ACM Comput. Surv.,

50(3):32:1–32:43.

Rahman, H. and Rahmani, R. (2018). Enabling distributed

intelligence assisted future internet of things con-

Enabling Distributed Intelligence in the Internet of Things using the IOTA Tangle Architecture

397

troller (ﬁtc). Applied Computing and Informatics,

14(1):73 – 87.

Research, A. (2013). More than 30 billion devices will wire-

lessly connect to the internet of everything in 2020.

(1).

Sahni, Y., Cao, J., Zhang, S., and Yang, L. (2017). Edge

mesh: A new paradigm to enable distributed intelli-

gence in internet of things. IEEE Access, 5:16441–

16458.

Serguei, P. (2017). The tangle. (1).

Van den Abeele, F., Hoebeke, J., Teklemariam, G. K., Mo-

erman, I., and Demeester, P. (2015). Sensor function

virtualization to support distributed intelligence in the

internet of things. WIRELESS PERSONAL COMMU-

NICATIONS, 81(4):1415–1436.

Yi, S., Li, C., and Li, Q. (2015). A survey of fog computing:

Concepts, applications and issues. In Proceedings of

the 2015 Workshop on Mobile Big Data, Mobidata 15,

pages 37–42, New York, NY, USA. ACM.

IoTBDS 2019 - 4th International Conference on Internet of Things, Big Data and Security

398