The end of IoT Enthusiasm or the Beginning of a new IoT Vision

Abstract

The article presents a new vision of IoT/Smart Home configurations. Smart Home can be regarded as an IoT lab. The experience acquired working on Smart Home solutions can be transferred to IoT space. The new IoT vision is presented using Smart Home space. The new vision related to Limited Natural Language Communication, Smart Home-learning and above all, to the transition to a new architecture - Semantic Network Based Architecture is reasoned. According to this paradigm, Smart Home configurations will employ Semantic Oriented (real time) Programing.

Full text

Spring Scientific Session of FMI 2016
Page 1
THE END OF IOT ENTHUSIASM OR THE BEGINNING OF A NEW IOT VISION
Lyubomir Blagoev, General Manager, USW Ltd, Bulgaria, lyubo@usw.bg
Kamen Spassov, Head of e-Government Laboratory, Bulgarian Academy of Sciences, Bulgaria,
kspassov@gmail.com
Abstract
The article presents a new vision of IoT/Smart Home configurations. Smart Home can be
regarded as an IoT lab. The experience acquired working on Smart Home solutions can be
transferred to IoT space. The new IoT vision is presented using Smart Home space. The new
vision related to Limited Natural Language Communication, Smart Home-learning and above all,
to the transition to a new architecture - Semantic Network Based Architecture is reasoned.
According to this paradigm, Smart Home configurations will employ Semantic Oriented (real
time) Programing.
Keywords: Smart Home, Internet of Things (IoT), Semantic networks, Limited Natural Language, Machine
learning.
1 INTRODUCTION
Last year (2015) unpleasantly surprised us with a sharp drop in IoT sales and above all of Smart Home
equipment [1]. From the point of view of Technology Adoption Life Cycle (TALC), Smart Home market is at
the TALC stage defined as chasm. The early Smart Home market ended but the mass Smart Home market
did not come yet.
First participants in early Smart Home market - technology enthusiasts (often referred to as ‘innovators’)
were not disappointed. They got linked ‘things’. At the one end of the link were terminal devises (most often
Smartphones) or big Data Bases. At the other end were devices of all kinds which intelligence was limited
to the capability to support connectivity to different transmission platforms and protocols.
Enthusiasts were happy but visionaries did not receive any more advanced functionality than that offered
by the achieved connectivity. The IoT platforms and the based on them configurations of connected and
interacted autonomic devices were supposed to provide expected functionality. Anticipations of the
visionaries were not met and it turned out that the understanding of IoT platform needs reconsideration [2].
Intensive discussions on this topic did not happen. Potential participants in such discussions fall into the
situation of people who don’t know where they are heading to, so they don’t bother to make decisions
which road to go [3].
This way we came to the suspicion that the TALC chasm is not marketing but technological.
2 SMART HOME SEGMENT IS AT THE BOTTOM OF THE TALC CHASM
Indeed the Smart Home shows the biggest delay in its development in comparison with the other sectors of
the IoT space. Large and useful Smart City systems were created with clear understanding of the cost and
benefits. Creators of a new generation of smart vehicles recorded remarkable progress, although smart
vehicles are more intelligent than linked to one another. Similar observations are valid for telemedicine,
agriculture and other IoT sectors.
The reasons for the lag in the Smart Home sector are two:
- Smart Home environment is very specific. The consumer value of each IoT-implementation is
determined mainly by quite uncertain and dependent on various factors sense of home comfort.
- Smart Home is an environment whose functionality is similar largely to the functionality of control
systems. At the same time, a Smart Home solution and a control system are hardly compatible
because of the need to create a sense of home environment comfort.
As it is mentioned in [4]:
“Obviously the more complex the future Smart Home is evolving the more complicated
becomes its control. Therefore, the inhabitants will need stronger engagement with the
control. Ergo from served by an intelligent environment with a high degree of automation
Smart Home inhabitants could turn into servants of a system of smart devices”

Spring Scientific Session of FMI 2016
Page 2
3 SMART HOME SPACE AS AN IOT LAB
The experience shows that achieving any form of home comfort is possible only applying Human Oriented
Approach. The difference between Smart Home solutions and classical control systems discussed earlier is
not the only one. Getting back home usually, we ask people at home the question: ‘What happened today?’
This is the beginning of a simple conversation in which the exchanged information is not so important. The
psychological effect of such communication process is much more important. The result can be described
as ‘entrance/registration into the home environment’, ‘inclusion in the home environment based on sharing
what happened in it’, etc.
It does not matter how we describe that result, it cannot be achieved by reading log records from the
screen about actions taken by home devices or committed by the occupants of the home. The replacement
of these log records with texts representing semantics of the log by means of natural language would be a
solution precisely corresponding to the Human Oriented Approach. However, such a solution should not be
sought in the vast field of artificial intelligence, not only because it will be very expensive, but also because
household devices and possible actions they perform can be presented with a limited semantics, i.e. they
are ‘low intelligent’.
We defined Smart Home features that ensure Human Oriented Approach:
3.1 Smart Home feature N1: Ability for Limited Natural Language Communication
Constrains of this Natural Language Communication are derived naturally and they are well defined. The
lexical volume of communication is defined by:
 Lexical units representing the semantics of the devices connected in a Smart Home configuration
 Lexical units representing the semantics of the possible acts of interaction between the connected
devices.
The grammar of the natural language that serves such a communication will also be simple and with the
lexical set do not need to be artificially extended and complicated. Variability, beauty, grace of expression,
etc. are not needed. The addition of a new device in Smart Home-configuration essentially means adding
new semantics with clear volume and composition.
Adding a new device does not mean overcoming protocol differences only, but requires an extension of
interactions between people and devices in the Smart Home environment. Adding a new device requires an
extension of the existing semantics in a particular Smart Home configuration. It is not recommended every
time the same device is added to different Smart Home configurations to make new definitions of this
semantics, because essentially they are equivalent. This way we come to
3.2 Smart Home feature N2: New vision for „smart device configuration“
To create a new Smart device it is not enough to produce a device with certain functionality that could be
tagged as “smart”. Producers should include in Smart Home device configuration along with standard
protocols drivers for data exchange also semantics definitions that allow integration in any Smart Home
environment.
To make it happen it is necessary to solve the problem of standardization or at least the unification of the
semantic representation of devices, interactions between them, and their interaction with the occupants of
the Smart Home. The semantic representation must be in different languages, as the manuals of the
devices are available in different languages.
On the other hand it is desirable the unification of this presentation not to impose lexical restrictions, i.e. to
allow synonymous representations. A prerequisite for solving this problem is the availability of WordNet [5],
which might help to include new languages. However, this is not enough and we should point out the need
for:
3.3 Smart Home feature N3: Capability for Speech Communication
Writing and reading of long (albeit clever) texts with a laptop or a smartphone or a tablet will completely
erase the benefits achieved by introduction of Limited Natural Language Communication. Seems, the
provision of voice communication in the Smart Home-environment is a must.
Currently the implementation of text-to-speech conversion is not a technical problem. Adding such
functionality to a Smart Home configuration, i.e. IoT- configuration, is not so expensive. The available
solutions allow to adjust the speed, timbre, etc. of the voice reproduction and thus to enrich the voice
communication within the Limited Natural Language at a low cost.

Spring Scientific Session of FMI 2016
Page 3
The reverse process, i.e. the speech recognition, will also be not too expensive because of the use of
Limited Natural Language and the clear semantic constrains - far away from complexity of the Artificial
Intelligence solutions and related costs.
3.4 Smart Home feature N4: Ability for learning instead for tuning
The functionality of a Smart Home environment can be represented by various scenarios of interaction of
interconnected devices and people communicating with them. This interaction becomes more complicated
with the increase of the number of devices.
For example, the reaction to a door of the refrigerator left open in a Smart Home configuration can be
constrained to limit the operation of the vacuum cleaner near the refrigerator. It can be assumed that after
some time defrosting will result in water leakage. But, if to such Smart Home-configuration we add a
module to control the contents of the fridge, this module should estimate after some time that the food is
spoiled and could launch a scenario for new food delivery.
Adding a new device to a Smart Home configuration requires constructing its interactions with existing
devices. In general, we should assess the all possible interactions, i.e. the possible connections between
devices based on the use of the same semantics. For example, in the control scenario “refrigerator’s door
is open” and in the control scenario “food contained in the refrigerator” the semantics of "refrigerator’s door
open” is included. The same semantics is not present in the scenario “a cat near to the baby’s bed”, but it is
present in the scenario “a cat near to unauthorized places at home”.
The best way to solve this problem is to provide Smart Home tools to connect new devices not by changing
settings but by self-learning for interaction between new and existing devices in the Smart Home
environment. Learning should be led in interactive mode via voice communication with terminology
available to the owners of Smart Home, which do not have special training and experience.
3.5 Smart Home feature N5: Support for (local) Semantic interoperability
Semantic representation of interoperating devices and people in Smart Home space requires establishing
and maintaining a sustainable Semantic interoperability. It can be said that this is a fundamental
requirement, which is the foundation for the realization of the aforementioned characteristics.
The experience from previous works on the creation of communities of developers that use the same
definitions of concepts, data and processes that form a standard environment of Semantic Interoperability
shows that it is difficult. At least at this stage to create a standard environment with coverage similar to the
coverage of Internet standards is challenging. Pragmatism shows that it is more likely to establish and
maintain different environments with semantic interoperability formed, for example, by different
manufacturers Smart Home equipment suppliers or described as Smart Home Digital Environment – SHDE
[4].
It is hard to imagine the IoT-space as a unified environment with established and maintained sustainable
Semantic interoperability. More realistic is to see this space as consisting of local environments with
Semantic interoperability that exchange data within some level of semantic connectivity. This connectivity
will be very different from the connectivity established by currently available protocols to exchange data
with other devices.
3.6 Smart Home feature N6: New semantic oriented (real time) programing
The widespread program languages utilize the universal If-Then-Else (ITE) scheme and its derivatives to
describe arbitrarily complex structures of logical processing. High-level algorithmic languages support
complex data structures presented as mathematical abstractions that can be loaded with different
meanings. Thus arises the problem of organization and control of "the meeting between mathematical
abstraction and semantics of things from real life."
The best solution to this problem is, if possible, to avoid solving it at all. For this to happen it is necessary to
start working in an environment that presents the semantics of things from real life and semantics of their
interaction instead of starting from the linguistic tools to semantics.
This approach is a further development of the tools for establishing and maintaining Semantic
interoperability. Creating models of concepts, data and processes is an advancement in models of real
objects from Smart Home environment. Every model alongside with meta-descriptions will contain program
code that ensures the maintenance of the semantics of its instances.
An important feature of semantic oriented programming is the total decentralization of the program code.
The definition of a program fragment is determined solely in relation with the required processing of the

Spring Scientific Session of FMI 2016
Page 4
content maintained in the instance of a model. The model contains a program fragment as a part of
maintaining its content.
The decentralized program code will work in real time in accordance with classical procedures for sharing
computing resources of Smart Home environment. In this case, the organization of the work will be
significantly eased due to the existing opportunity to identify the shared resources. The arbitration the
access to these resources will be embedded as a description in their models, which in turn are subject of
shared use.
For this purpose, it is necessary to have
3.7 Smart Home feature N7: New semantic oriented architecture
The models of objects (members of SHDE according [4]) and their instances will exist in a homogeneous
environment that will have the structure of a graph. The nodes of the graph will represent the semantics of
models and their instances, and connecting arcs will also provide semantics, including the semantics of
interaction.
This architecture has got some specifics and the following ones are good to be pointed out:
 The environment is homogeneous and its components are distinctive in semantics, which they
represent.
An object as an instance of another object playing the role of a model in itself can act as a model of
respective instances. The depth of such semantic embedding should not be restricted by the
environment but by the ontology that is created.
 Dynamic configuration of sets of objects that are involved in currently executed scenario
For another scenario another set of interacting objects will be formed. Objects from the previous set
may be present in the new scenario. The definition of the functional modules in this environment is
not only impossible, but also unnecessary. Adding a new object in the environment can alter the
composition and function of a part of or all previous sets of objects and scenarios in which they are
included. It is possible to perceive graph nodes as modules, but there will be difficulties with
connecting arcs because it is hard to perceive them as interface modules in all of the cases.
 There is a lack of functional layers
The lack is predetermined by the lack of layers in the existing semantics in the real world of the
inhabitants of the Smart Home. The very concept of "layered model" serves the human mindset for
a presentation of connected components of a complex system with a simplified presentation by
defining layers in its construction. If someone prefers to work with functional layer, it is possible.
However, in this environment it does not make much sense.
The described unified graph based environment of models and their instances is essentially a semantic
network. Therefore, it is more properly to call such a semantic oriented architecture Semantic Network
Based Architecture - SNBA.
4 CONCLUSIONS
We created an IoT Lab in a Smart Home environment following the definitions formulated earlier. A
software realization of SNBA is created with an initial set of models that form the core of Scenario
Description Framework-SDF. It was tested with in real time with few scenarios to prove the new IoT vision
and architecture.
It is planned to create new SDF resources and to improve the existing ones to enable Limited Natural
Language Communication and test it.
5 ACKNOWLEDGMENTS
The work is a result of joint efforts of the team of INATO Ltd with the experts from the e-Governance
Laboratory at Bulgarian Academy of Sciences.
References
[1] John Greenough, The US smart home market has been struggling — here's how and why the market
will take off, Business Insider, (http://www.businessinsider.com/the-us-smart-home-market-report-
adoption-forecasts-top-products-and-the-cost-and-fragmentation-problems-that-could-hinder-growth-
2015-9)

Spring Scientific Session of FMI 2016
Page 5
[2] Knud Lasse Lueth, Jan Kotzorek, White paper „IoT platforms-The central backbone for the Internet of
Things“, IOT ANALYTICS, Nov.2015 (http://blog.mydevices.com/wp-content/uploads/2015/11/White-
paper-IoT-platforms-The-central-backbone-for-the-Internet-of-Things-Nov-2015-vfi4.pdf)
[3] Lewis Carroll, Alice's Adventures in Wonderland, 1865
[4] Blagoev L., Spassov K., Smart home as a Digital Environment, Sofia University, Spring Scientific
Session of Faculty of Mathematics and Informatics, 28.03.2015, Sofia, Bulgaria,
(https://www.researchgate.net publication/275032974)
[5] Christiane Fellbaum (1998, ed.) WordNet: An Electronic Lexical Database. Cambridge, MA: MIT Press.