How Internet of Things Influences Human Behavior Building Social Web of Services via Agent-Based Approach

Abstract

The paper discovers potential human interactions with growing amount of internet of things (IoT) via proposed concept of Social Web of Services (classical social web with smart things - daily life objects connected to the internet). To investigate the impact of IoT on user behaviour patterns we modelled human-thing interactions using agent-based simulation (ABM). We have proved that under certain conditions SmartThings, connected to the IoT, are able to change patterns of Human behaviour. Results of this work predict our way of living in the era of caused by viral effects of IoT application (HCI and M2M connections), and could be used to foster business process management in the IoT era.

Full text

HOW INTERNET OF THINGS INFLUENCES HUMAN BEHAVIOR
BUILDING SOCIAL WEB OF SERVICES VIA
AGENT-BASED APPROACH
Mikhail KOMAROV*1, Nikita KONOVALOV*2 and Nikolay KAZANTSEV*3
Abstract: The paper discovers potential human interactions with growing amount of
internet of things (IoT) via proposed concept of Social Web of Services (classical social web
with smart things - daily life objects connected to the internet). To investigate the impact of
IoT on user behaviour patterns we modelled human-thing interactions using agent-based
simulation (ABM). We have proved that under certain conditions SmartThings, connected to
the IoT, are able to change patterns of Human behaviour. Results of this work predict our
way of living in the era of caused by viral effects of IoT application (HCI and M2M
connections), and could be used to foster business process management in the IoT era.
Keywords: Internet of Things, Internet of Service, Web of Service, Social Web of
Service, business model, Agent-based theory, smart home, virality, digital transformation
1. Introduction
No one in 90s may even thought that in 20 years 4 billion IPv4 addresses will not be enough
to satisfy Internet connection demand. Already in 2008-2009 the number of Internet-
connected devices exceeded the number of people on the planet that we consider as a starting
point of structural changes described in this article. Cisco Internet Business Solutions Group
proclaimed it as a start of Internet of Things era [9]. Academia researchers have widely
studied these changes last years and coined more notions: Internet of Services (IoS) and Web
of Services (WoS). Even though these terms look very similar, they have noticeable
distinctions: they enhance, overlap and supplement the original IoT concept. We summarised
the main notions used in the article in the Table 1.
* Faculty of business and management, National Research University Higher School of
Economics
1 mkomarov@hse.ru
2 niks.konovalov@gmail.com
3 nkazantsev@hse.ru
FOUNDATIONS OF COMPUTING AND DECISION SCIENCES
Vol. 41 (2016)
DOI: 10.1515/fcds-2016-0012
No. 3
ISSN 0867-6356
e-ISSN 2300-3405
Unauthenticated
Download Date | 6/27/17 12:21 AM

Table 1. Main definitions used in paper
N
Notion
Definition
Source
1 Smart thing /
smart object
Objects that are called smart carry chunks of
application logic that let them make sense of
their local situation, intercommunicate and
interact with human users via IP protocol. Kortuem [15]
2 Internet of
Things (IoT)
t
he worldwide network of interconnected
objects uniquely addressable based on standard
communication protocols.
RFID group
[12]
3 Internet of
Services (IoS)
the global infrastructure that uses the Internet as
a medium for offering and selling services -
tradable goods. It r
epresents a model of IoT
(self-) organization.
[6], [13], [22]
4
Web of
Services
(WoS)
the network of interconnected services [6,13]
5 Social Web of
Services
(SWoS)
The network of shared services by people and
devices via social networks [6,13]
The main goal of our research is to check how the raising number of internet connected
devices influence users. We do it providing and testing a simulation model of SWoS as a
reasoning tool for enhancing IoT business models. Firstly, we analyse an existing benchmark
of IoT business model through the prism of Osterwalder Canvas framework [18] (chapter 2),
apply Agent based modelling simulation (chapter 3) and describe practical application of
models (chapter 4).
2. The raise of things
From its foundation, the Internet is constantly improving to be more and more
personalized especially during the last two years. Online environment is already about to
198
M. Komarov, N. Konovalov, N. Kazantsev
Unauthenticated
Download Date | 6/27/17 12:21 AM

aggregate each aspect of individual’s everyday life, for example, controlled via mobile
device. Business in its turn, goes digital as well: virtual workplaces and processes, digital
customers and digital “Things”. The whole picture is seen as a digital “Ecosystem”, where
enterprises are connected with its suppliers, distributors and customers in a common digital
network. Making business in such Ecosystems is more profitable for companies due to
economies of scale, as they are no longer acting as individual corporate entities. Accenture
Technology Vision 2015 names this trend “We-Economy” [1].
These concepts lead appearance of the new business models [6,13]. For instance, by
monitoring of IoT products lifecycle, companies are able to sell devices directly to the
customer at the very moment when such need occurs. By embedding the IoT into social
networks and providing mobile and ubiquitous access to its services, companies improve
their predictive analytics. Meanwhile Cloud technologies play a role of intermediate platform
to connect users, “Things” and Enterprises into a digital Ecosystem [11].
2.1. Smart home – a benchmark of Smart Thing business initiatives
The problem of unified control of all smart devices exercised the minds of such market
incumbents as Microsoft, Cisco and others, but their projects had little success. However,
during CES 2014 conference Samsung4 presented their new attempt. The developed app for
smartphones or TVs (with Android OS, apparently) let people control home appliances, for
example heating system or turn the lights on and off when they are not home with three main
functions: device control, home view and smart customer interface. Home View, another part
of the service, let people check up on their home using appliance cameras. Moreover, a
customer service component alerts people when their appliances need replacing or servicing
- something manufacturers would clearly enjoy.
A dedicated Smart Home software protocol (SHP) connects the related devices
(refrigerators, washers and dryers, ranges, microwaves, dishwashers, vacuums, smartphones,
TVs, gears, audio systems, cameras, computing devices and LED lighting)) and opens access
for other vendors, if they choose to get on board. The company plans to eventually expand
the Smart Home service to cover home-energy, secure home access, health care, and “eco
home applications” through the partnerships with third-party service providers.
There is no detailed information on how exactly operates their platform, and we suppose
that all third-parties devices could be included into this app only via agreements with
Samsung. The business model of its service is shown on (Fig.1):
4 Leading Korean company is a major manufacturer of home appliances like washing machines and
refrigerators, not to mention smartphones
How Internet of Things influences human behavior ...
199
Unauthenticated
Download Date | 6/27/17 12:21 AM

Figure 1. IoT influenced Smart Home Business model
For today, the main limitation for new market players is the range of supported devices
and their availability. It is crucially important to work with the most popular and affordable
smart devices in order to sweep the market. The biggest competition today is on US market,
where there are several competitors and maximal share of iOS devices than in other countries
[6]. At the same time, in other countries, Samsung has more power, because the accessibility
of its competitors’ devices is limited while Samsung products are readily available. However,
there is still an opportunity to beat the electronic giant, because of the type of supported
devices. While Samsung operates with its own appliances, which are rather massive and
expensive, it is possible to enter the market with smaller, cheaper and simpler devices like
sensors and power outlets offered by SmartThings [24] and Revolv [20].
The market of smart devices control application is still unsaturated, which leaves an
opportunity for new players. There are also several companies, which provide means of
communicating to any service via an encrypted P2P connection (Weaved) [25] or different
services for developers (Microsoft Home OS and others). However, these three described
company are the main pretenders for the unified method of controlling smart devices.
What happens when the number of smart things connected to Smart Home and mother
IoT-influenced business models will emerge? How that will affect human behaviour
patterns?
To investigate that in the next chapter we use Agent based modelling (ABM).
200
M. Komarov, N. Konovalov, N. Kazantsev
Unauthenticated
Download Date | 6/27/17 12:21 AM

3. Agent based model simulation
3.1. Agents
ABM simulation is a perfect solution for scrutinizing the aspects of complex systems,
which comprise interacting agents. Bonabeau [3] highlighted three main benefits of this
approach: it captures emerged phenomena, has a natural-language description and flexibility.
What is more, ABM models could be classified depending on their purposes and modelling
approach by Borshchev [5]. In our research we used simulation model to track the execution
through the discrete states, analyze dynamics and interactions within the algorithm suggested.
According to Macal and North [17], an agent should possess the following patterns:
•Rules to define agent behavior and decision-making
•Protocols to communicate with other agents within the system
•Have autonomous and self-directing functionality
•Be goal-oriented
Taking to consideration these patterns, we have highlighted two types of agents in the
model:
Type 1: a Human
Type 2: a Smart Thing (ST) - any electronic gadget with acting scripts, connected to the
Internet
As we proceeded with individual agent behavioural rules, the model could be classified
as a micro level simulation [4]. Both types of agents have their own attributes in order to
trigger behavioural patterns within current simulation process algorithm. Epstein [7] found
out that the understanding of social interactions lies in a deep process analytics from which
these interactions arise. In order to scrutinize the simulation’s results we found out 3 potential
outcomes:
1. Smart Things interaction is able to shape a fully viral behavioral trend under the
suggested default algorithm conditions.
2. Smart Things interaction is able to shape a partly viral behavioral trend under the
suggested default algorithm conditions.
3. Smart Things interaction, supported with human social effects is not able to shape a
behavioral trend under the suggested default algorithm conditions.
We have modelled the interaction between Human and ST according to the subjective
and average resource requirement. Resource Requirement (RR) is random and specific to
each Human agent numeric value, which shows optimal outcome of the Resource usage for
concrete Human agent. Either it could be subjective for the concrete Human with individual
machine settings or Human agent can accept a requirement trade off, offered by the ST, and
receive Average RR with default machine settings (Fig.2).
According to the model (Fig.1), it takes more time for Human to use Resource and get
subjective RR (i.e. making choice, machine tune-up, etc.), than to use default setting with
How Internet of Things influences human behavior ...
201
Unauthenticated
Download Date | 6/27/17 12:21 AM

average RR suggested by Smart Thing. This motivates Human for insignificant N% RR
tradeoff. In case the current average RR differs by less than N% from the subjective RR,
Human will use default average value instead of subjective one.
Figure 2. UML visualization of Smart Thing simulation
3.2. ABM description
We have pre-defined a list of conditions before simulation:
1. We assume that one Human has one Smart Thing
2. Humans interact with each other according to the simple random social graph
3. Smart Things are all connected into a one single net, so they can calculate the average
RR.
4. Social nature of Human agents influences on their decisions (Once three of four
socially connected agents start using the average RR mode, the fourth Human will switch to
this mode as well, no matter what the trade-off percentage is.)
Out of respect to the visual aspects, we did a simulation using random social network of
20 Human agents, using their 20 Smart Things (ST). Relations quantities between Humans
were randomly generated from 6 to 8 for each agent, representing a social graph (Fig.3, 4).
ST were interconnected into a single network.
202
M. Komarov, N. Konovalov, N. Kazantsev
Unauthenticated
Download Date | 6/27/17 12:21 AM

Figure 3. Social graph for Human agents
This model requires several stages of simulation:
1. On the first stage, all the humans have subjective RR (SRR) mode and randomly get
their value from 1 to 100. Smart Things, interconnected in the net, calculate the average value
of RR. We assume that the number of agents during the simulation is constant.
2. After that, STs choose Humans with N% difference between subjective and average
RR to make an trade off offer (Due to the sample size, trade-off percentage was set as 20%).
These Humans choose the average RR and will use it on further stages.
3. At the beginning of the second stage, each Human agent checks if any three of his
social friends switched to the average RR. In this case, this agent switches as well. Then,
Humans use their Resources according to the chosen RR mode. After that, SmartThings
calculate the new average RR and start looking for the new humans with possibility of 20%
trade-off.
4.On the further stages, the cycle repeats.
How Internet of Things influences human behavior ...
203
Unauthenticated
Download Date | 6/27/17 12:21 AM

.
Figure 4. Social graph for ST agents
3.3. Simulation result
In our simulation, different factors were evaluated coherently one by one to define the
extent of trend virality. These factors are trade off percentage level, number of social nets per
agent, value corridor for randomly generated subjective RR.
Each factor was divided into three value corridors, meanwhile other factors’ values were
fixed. The most effective way to create a fully viral trend appeared to be agent trade off
percentage level. It signifies that the more flexible decisions are made by an agent, the more
viral the trend is. Moreover, the number of social nets per agent plays a vital role as well.
According to the Appendix 1, fewer nets quantity could decrease trend virality drastically. It
took 27 attempts to reach a fully viral trend with 1 to 4 nets amongst 20 agents. Finally, value
corridor for subjective RR describes dispersion of agents’ preferences. It is easier to reach a
fully viral trend when Humans have similar preferences for resource requirement.
However, depending on the sample size, number of social relations and SRR limitations,
the trend may occur to be partially viral. In this case, not all of the Human agents will switch
to the AVG RR mode. It is proved by the number of simulation attempts, which were done
until fully trend were achieved and all agents were switched.
204
M. Komarov, N. Konovalov, N. Kazantsev
Unauthenticated
Download Date | 6/27/17 12:21 AM

Detail representation of simulation No.5 (see Appendix 1) is shown on the Figure 5.
Vertices of the graph are named according to the mask XXAVGYY, where:
XX – is the number of simulation stage, on which agent switched his mode,
AVG – average value for RR,
YY – ID of an agent, from 1 to 20.
Figure 5. Representation of Smart Thing simulation
Depending on the number of social relations between the Humans and their quantity, the
simulation leads to strengthening the trend. In our case, all the Human agents were influenced
by the ST proposal during the fourth attempt. On the larger samples, ST proposals may not be
that influential.
Overall, the SmartThings, connected to the IoT, are able to change patterns of Human
behavior (see Fig.6).
How Internet of Things influences human behavior ...
205
Unauthenticated
Download Date | 6/27/17 12:21 AM

Figure 6. Representation of Smart Thing simulation
Overall, we can conclude that that under the given simulation conditions and agents
quantity, two different types of trend were set: fully viral trend (all the agents switched their
mode) and partly viral trend. This means that options 1 and 2 are observed within given
simulation conditions. Absolutely viral trend (achieved at the first attempt) was reached with
the following simulation conditions:
1. Either trade off percentage level for subjective RR was set at abs. 50% from the
current stage average RR level,
2. Or each Human agent was acquainted with 40%-60% of other agents within the
whole net,
3. Or scatter of random subjective RR was narrowed to 30%-70% value corridor.
Lower levels of presented conditions lead to the option 2 - partly viral trend. Option 3 -
was not detected under given simulation conditions.
3.4. Outcomes
This IoT ABM simulation is good example of slight manipulation on the trends usage
through SmartThings’ mode setting. Companies can influence these trends by controlling the
IoT networks and setting default modes for their smart products and, as a result, sell advanced
settings for SmartThings separately. This business model is very similar to nowadays concept
of InApp Purchases (IAP) and paid Download Content (DLC) embedded in numerous mobile
applications and computer games. This enhancement allows setting a selling price at the
lower level and compensate costs by offering IAP, and paid DLC.
206
M. Komarov, N. Konovalov, N. Kazantsev
Unauthenticated
Download Date | 6/27/17 12:21 AM

4. Concept development and further research
After this work was presented in Sydney DataCom iThings 2015, we received multiple
questions where it could be applied except prediction analysis. There is an obvious area –
novel process management methodologies considering agents within social web of things.
Subject-oriented Business Process Management (S-BPM) is a relatively new approach
for the overall handling of work procedures in organizations from analysis to IT-based
execution including cognitive approach [10]. It focuses on the acting entities in processes
(people, software, information systems, services, etc.) and their interactions to achieve the
process goal. The explicit stakeholder and communication orientation makes it a promising
candidate to overcome major drawbacks of traditional BPM as there are deviations of lived
processes from their specification (model-reality divide), giving away opportunities for
improvement proposed by employees (lost innovation) and slow adaption of organization
and IT to changing requirements. With its easy-to-understand and easy-to-use notation based
on the Subject–Predicate–Object scheme of natural language, S-BPM facilitates semantic
and organizational integration of people in the design of their work procedures. On the other
hand, a clear formal semantic behind the graphical notation allows automatic code generation
for workflow execution at runtime. Hence, stakeholders can instantly test the models they
created and iteratively improve and complete them until they are considered ready for going
live and being executed by a workflow engine. This leads to seamless roundtrip engineering
based on a common understanding of both business and IT people and can significantly
increase organizational agility. S-BPM is not just another modelling language but also a
comprehensive methodology spanning the whole business process management life cycle. It
suggests a paradigm shift from the traditional control flow-based view to a stakeholder and
communication-oriented view on business processes. Subjects represent the active entities in
a process and behave in a certain way to accomplish the goal. Their behaviour includes
exchanging messages and performing activities on business objects. Hence, a subject-
oriented specification of a process follows the standard sentence semantics of natural
language consisting of subject, predicate and object. However, S-BPM is a successfully
implemented for business process management it is strongly required to have agent based
model simulation.
In this regard, it is necessary first to arrange simulation in order to understand how our
smart things will act as agents, become independent subjects in one process, and object in
another process. Furthermore, it is required to create an integration platform between IoT
vendors and data consumers.
5. Conclusion
The developed business model of Social Web of Services combines the idea of common
social web and usual service selling, increase the quantity of revenue streams and enhance
usual suggested ways of controlling smart devices. It may be the transition way to full Web
of Services and it can increase people's way of living by creating “smarter world”.
Provided ABM simulation showed that companies might manipulate human behavioural
patterns through ST cooperation and social effects. As a result, there are required usage
How Internet of Things influences human behavior ...
207
Unauthenticated
Download Date | 6/27/17 12:21 AM

setting trend could be created. This leads to the main enhancement of the developed business
model of Social Web of Services. A number of revenue streams is increased: enterprise is
able to sell additional settings as paid In-Machine Purchase or Download content (DLC)
where customer relationships are individualized due to subjective preferences.
Information security, information privacy and legitimacy are obstacles that we need to
avoid on the way of concept realization. We are sure, that the question of systematic resilience
should be the first thing solved before launching the platform. Moreover, we recommend the
ownership to be shared not only with private companies, but also with government officials,
and perhaps, while the service is international, its localizations in different countries should
be separated in order to protect users from different restrictions of the local government.
References
[1] Accenture Technology Vision (2015). IT Trends 2015, Accenture PLC
[2] Atzori, L. Iera, A., and Morabito, G., The Internet of Things: A survey, in Computer
networks, 54, 2010, 2787-2805,
[3] Bonabeau, E. Agent-based modeling: Methods and techniques for simulating human
systems, in Proceedings of the National Academy of Sciences, 99 (suppl 3), 2002,
7280-7287
[4] Borshchev, A., & Filippov, A.. From system dynamics and discrete event to practical
agent based modeling: reasons, techniques, tools. In Proceedings of the 22nd
international conference of the system dynamics society (Vol. 22), 2004
[5] ComScore research, February 2014:
http://www.comscore.com/Insights/Press_Releases/2014/4/comScore_Reports_Febr
uary_2014_US_Smartphone_Subscriber_Market_Share [Accessed 20, June, 2014]
[6] Efremov S., Pilipenko N., Voskov L., Komarov M. Cloud IoT Platforms: A Solid
Foundation for the Future Web or a Temporary Workaround?, in: Lecture Notes in
Computer Science Vol. 9247: Internet of Things, Smart Spaces, and Next Generation
Networks and Systems. Switzerland : Springer International Publishing, 2015. P. 47-
55.
[7] Epstein, J. M. Generative social science: Studies in agent-based computational
modelling, in Princeton University Press, 2006.
[8] ETP EPOSS. ETP EPOSS IOT Definition: http://old.smart-systems-
integration.org/internet-of-things/Internet-of-Things_in_2020_EC-
EPoSS_Workshop_Report_2008_v3.pdf/download [Accessed 20, June, 2014]
[9] Evans D., The Internet of Things. How the Next Evolution of the Internet Is Changing
Everything. Cisco Internet Business Solutions Group:
https://www.cisco.com/web/about/ac79/docs/innov/IoT_IBSG_0411FINAL.pdf ,
2011 [Accessed 20, June, 2014]
[10] Fleischmann, A. What Is S-BPM? S-BPM ONE – Setting the Stage for Subject-
Oriented Business Process Management, in Communications in Computer and
Information Science. Heidelberg: Springer, Berlin, 2010
[11] Gromoff A., Kazantsev, N., Kozhevnikov, D., Ponfilenok, M. and Stavenko, Y.
Newer Approach to Create Flexible Business Architecture of Modern Enterprise, in
Global Journal of Flexible Systems Management. 13(4), Springer, 2012, 207-215
208
M. Komarov, N. Konovalov, N. Kazantsev
Unauthenticated
Download Date | 6/27/17 12:21 AM

[12] Gubbi, Jayavardhana, et al. "Internet of Things (IoT): A vision, architectural elements,
and future directions." Future Generation Computer Systems 29.7 (2013): 1645-1660.
[13] IOT Definition: http://cordis.europa.eu/news/rcn/30283_en.html [Accessed 20, June,
2014]
[14] Komarov M.M., Khokhlova A.D. Development of a business model for Social Web
of Services, In Business Informatics. No.2 ,32, 2015, 20-29
[15] Kortuem, Gerd, et al. "Smart objects as building blocks for the internet of things."
IEEE Internet Computing 14.1, 2010, 44-51.
[16] M. Porter and D. Millar Information Technology as a strategic advantage, 1985
[17] Macal, C. M., & North, M. J.. Agent-based modeling and simulation, in Winter
simulation conference, 2009, 86-98
[18] Osterwalder A. and Pigneur Y.. Business Model Generation, in a handbook for
Visionaries, Game Changers, and Challengers. John Wiley & Sons Inc, 2010
[19] P Research. SAP IOT Definition: http://services.future-
internet.eu/images/1/16/A4_Things_Haller.pdf [Accessed 20, June, 2014]
[20] Revolv Company: http://revolv.com/ [Accessed 20, June, 2014]
[21] Samsung Smart Home: http://www.samsung.com/uk/home-appliances / [Accessed
20, June, 2014]
[22] Schroth, Christoph, and Till Janner. "Web 2.0 and SOA: Converging concepts
enabling the internet of services." IT professional 9.3, 2007, 36-41.
[23] SEEKDA: http://www.slideshare.net/dipina/towards-future-internet-web-30-
internet-of-services-internet-of-things# [Accessed 20, June, 2014]
[24] SmartThings Company:
https://www.kickstarter.com/projects/smartthings/smartthings-make-your-world-
smarter; https://smartthings.com [Accessed 20, June, 2014]
[25] Weaved Overview: http://www.yoics.com/overview [Accessed 20, June, 2014]
The paper is an extended version of the paper presented at the International
Conference on Big Data Intelligence and Computing (DataCom 2015),
Chengdu, China, December 19-21, 2015
How Internet of Things influences human behavior ...
209
Unauthenticated
Download Date | 6/27/17 12:21 AM

Appendix
Table 2. Aggregated examples of ABM simulation result.
No.
Attempts
till fully
viral
trend
achieved
Simulati
on stages
Agents
switched
Trade off
%
Social Nets
num.
SUBJECTI
VE RR
Value
Corridor
1
3
8
20
30
1-4
1-100
2
4
7
20
40
1-4
1-100
3
1
3
20
50
1-4
1-100
4
27
8
20
20
1-4
1-100
5
4
9
20
20
4-8
1-100
6
1
4
20
20
8-12
1-100
7
14
6
20
20
1-4
10-90
8
7
7
20
20
1-4
20-80
9
1
3
20
20
1-4
30-70
210
M. Komarov, N. Konovalov, N. Kazantsev
Unauthenticated
Download Date | 6/27/17 12:21 AM