The Use of a Modelling & Simulation Tier
by the EMULSION IoT Platform
IVAN GANCHEV
Department of Computer Systems,
University of Plovdiv “Paisii Hilendarski”,
24 Tsar Assen St., Plovdiv 4000,
BULGARIA.
&
Institute of Mathematics and Informatics,
Bulgarian Academy of Sciences,
Akad. G. Bonchev St., Block 8, Sofia 1113, BULGARIA.
&
Telecommunications Research Centre (TRC),
University of Limerick,
Plassey, National Technological Park, Co. Limerick,
IRELAND.
https://orcid.org/0000-0003-0535-7087
ZHANLIN JI
College of Artificial Intelligence,
North China University of Science and Technology,
Caofeidian, Tangshan City,
CHINA.
https://orcid.org/0000-0003-3527-3773
Abstract: - This paper presents some design aspects of the EMULSION IoT platform, developed as a typical
example of the horizontal IoT platforms. The architectural overview and multi-tiered structure of the platform
are described, with special attention being paid to its modelling & simulation tier as a novel architectural
element proposed for inclusion in similar IoT platforms. Used to model cyber-physical-social (CPS) objects
and IoT services, along with their attributes and temporal/spatial/event characteristics, this tier is also utilized to
simulate the actual provision of IoT services in order to determine the optimal configuration of the platform in
each particular use case, by solving complex optimization tasks. Examples of such tasks are presented in the
paper along with some results obtained to date.
Keywords: - IoT platform, horizontal-type, tiers, modelling, simulation.
Received: April 17, 2021. Revised: January 26, 2022. Accepted: February 18, 2022. Published: March 3, 2022.
1 Introduction
The provision of Internet of Things (IoT) services
requires the use of a proper IoT platform1 for
ensuring interconnection and interoperability of
different types of IoT things, objects, devices,
networks, hardware modules, software components,
etc., and providing the required functionalities for
telco-, device-, data-, application management, etc.
1 IoT platform – a type of a digital platform, used for
building and managing IoT solutions [1].
[2], with some freedom of use by consumers (along
with the required and customization and
personalization) [3], which in the cloud domain is
known as Platform as a Service (PaaS). IoT
platforms play an important role and are of major
importance for the entire IoT ecosystem [2]. All
companies considering the realization of potential
IoT scenarios and use cases should seriously
consider the use of a suitable IoT platform that can
provide a stable, secure, and flexible solution
meeting today’s architectural and business
requirements [1].
WSEAS TRANSACTIONS on SYSTEMS and CONTROL
DOI: 10.37394/23203.2022.17.15
Ivan Ganchev, Zhanlin Ji
E-ISSN: 2224-2856
133
Volume 17, 2022
The objective of this paper is to point out to the
need of using a dedicated modelling & simulation
tier in the IoT platforms for: (i) modelling the
utilized cyber-physical-social (CPS) objects and IoT
services, along with their attributes and
temporal/spatial/event characteristics; and (ii)
simulating the actual provision of IoT services as to
determine the optimal configuration of the
corresponding platform in each particular use case,
by solving complex optimization tasks.
The use of a modelling & simulation tier is
demonstrated here on the example of EMULSION
[4] – a generic multi-service cloud-based IoT
operational platform, which could be used as an
architectural foundation for creating ‘smart city’ IoT
systems [5] and for fast roll-out of related IoT
services of different types.
2 State of the Art in the Area of IoT
Platforms Development
Even though in the last two years there was no
increase in the number of IoT platform companies,
the size of the IoT platforms market increased
substantially and it is expected to grow from $5
billion in 2020 to $28 billion in 2026, which is a
33% compound annual growth rate (CAGR), as
shown in [2]. This growth seems driven by the fast
adoption of cloud-based IoT platforms and the
rising number of customers who chose to buy an
IoT platform rather than build their own. The same
source identifies 613 IoT platform providers
operating on the market at the end of 2021. The
dynamics of the current situation is also recognized
there by pointing out that, starting from 2019, many
of these platform providers have pivoted their
business models – either by no longer focusing on
IoT alone and offering a collection of modular, pre-
integrated artificial intelligence (AI) services (24%
of the IoT platform providers), or having moved
toward the sale of IoT application solutions (21%),
or even stopping to exist (26%). While the number
of IoT platform vendors in all other regions of the
world decreased during the 2019–2021 period, in
the Asia-Pacific region, and especially in China, the
number of IoT platforms increased strongly. In total,
182 new IoT platform companies emerged since
2019, and 66 of these operate in China [2].
Commercial IoT platforms could be classified
into five general types [2]:
• Telco management/connectivity platforms
(constituting 7% of the 613 IoT platforms
existed by the end of 2021, [2]) – used to
manage the connectivity and control the
traffic to/from IoT devices at scale, primarily
through 2G÷6G cellular networks, but also
via Low-Power Wide Area Networks
(LPWANs) such as Sigfox and LoRa, and
potentially fixed and satellite connections.
Typical functionalities include billing
management, connectivity orchestration,
connectivity management, and service
provisioning [1]. As part of the subscription,
add-ons may also be included, e.g., bill
analyzers, usage anomaly detectors, etc.
Many cellular operators are relying on this
type of IoT platforms (and probably will
continue to do so) to scale up their IoT
deployments and extend their services to new
verticals [6]. In addition, top IoT platform
vendors are continuing to support multiple
use cases across different verticals [7].
Examples of this type of IoT platforms
include Cisco’s Jasper, Ericsson’s DCP,
Huawei Connection Management Platform,
Verizon’s network + ThingSpace, etc.
• Device management/enablement platforms
(constituting 35%, [2]) – providing an ability
to remotely configure, monitor, control, and
manage IoT devices. Functionalities of these
include over-the-air firmware updates,
deployment configuration, device monitoring,
command and control, etc. [1].
• Data management platforms (constituting
43% of the 613 IoT platforms existed by the
end of 2021, [2]) – providing an ability to
ingest, store, and analyze data from IoT
devices. Typical functionalities include data
storage support (data bases, data lakes, data
warehouses), data analysis (rules engines /
event management, data preparation,
extraction, transformation, load – ETL, data
analytics, AI / machine learning), and
southbound data ingest/egress (data
acquisition drivers and interfaces, IoT hubs,
IoT device software development kits, data
brokers) [1].
• Application management/enablement
platforms (constituting 58% of the 613 IoT
platforms existed by the end of 2021, [2]) –
providing an ability to rapidly develop, test,
verify, validate, and manage IoT applications
of different types. Standard functionalities
include application management (IoT
application marketplace and application
lifecycle management), application
development (digital twins, integrated
development environments – IDEs),
northbound data ingest/egress (application
WSEAS TRANSACTIONS on SYSTEMS and CONTROL
DOI: 10.37394/23203.2022.17.15
Ivan Ganchev, Zhanlin Ji
E-ISSN: 2224-2856
134
Volume 17, 2022
programming interface (API) management,
alert/notification services) [1]. IoT platform
vendors, as well as some third parties,
monetize applications built on top of the IoT
platforms of this type. Typical examples
include the Siemens’ Closed-Loop
Foundation application for MindSphere and
the Edge2Web’s Director application for
MindSphere, [1].
• IoT-based Infrastructure-as-a-Service (IaaS)
– these constitute only 3%2 of the 613 IoT
platforms existed by the end of 2021,
according to [2]. However, cloud hyperscalers
(e.g., Microsoft, AWS, Google) realize IaaS
revenue also by hosting the IoT platforms of
other companies on their infrastructures.
Examples include MachineMetrics on AWS
IoT Core, Uptake on Microsoft’s Azure, and
Oden Technologies on Google Cloud, [1].
Most IoT platform companies today are now
offering vertical solutions alongside their platforms
[2]. This contrasts with the horizontal design
principles for achieving easy integration and full
interoperability that holds much potential for
creating new business opportunities and associated
IoT services, and for eliminating the duplicate
solutions, thus simplifying the existing IoT
environments. Equipped with this new approach,
each service, application, or network provider can
supply a complete horizontal-slice solution,
applicable to multiple IoT domains, with greater
possibility to easy and timely adjust its operation to
new emerging IoT scenarios and use cases, and with
efficient operation, administration, and management
(OAM) of the IoT ecosystem throughout the entire
lifetime.
For non-hyperscaler IoT platform vendors, it is
extremely difficult nowadays to compete with cloud
hyperscalers, which capture an increasing quantity
of revenue with respect to platforms (via PaaS)
and/or computing infrastructures (via IaaS), [1]. So,
non-hyperscaler IoT platform vendors are
progressively focusing on more vertical purpose-
built specific applications (e.g., GE Digital),
services (e.g., Accenture), or solutions (e.g.,
Siemens), [1]. As the platform layer itself becomes
less differentiated, many companies seem
increasingly offering more vertical or use-case
specific solutions (both hardware and software) by
leveraging some underlying IoT platform, [1]. Large
multinationals and big enterprises have selected at
2 The total number of percentages is greater than 100%
because some of the platforms are identified in [2] as
being of more than one type.
early stage the IoT platform(s) for their own usage.
For instance, Walmart selected a single IoT platform
(Azure) for use, but Volkswagen selected several
IoT platforms (Siemens, AWS, and Azure), [1].
However, most small and medium enterprises
(SMEs) cannot afford the use of big-vendor
platforms. Following the horizontal trend,
EMULSION is developed as a horizontal IoT
platform of a combined (hardware and software)
type as to satisfy the needs of SMEs.
3 EMULSION IoT Platform
Building a robust and reliable, but at the same time
low-cost, IoT platform is a difficult task, taking into
account the fact that each such platform comprises
multiple heterogeneous hardware elements (e.g.,
semiconductors, sensors, actuators, monitoring
stations, controllers, guards, single-board
computers, communication modules, etc.) and
software components (e.g., embedded operating
systems, distributed message queues, producer-
consumer subsystems, machine-learning APIs,
applications, tools, utilities, databases,
dews/fogs/clouds, etc.). With millions of
repositories of open-source software components
and hardware modules available on GitHub3,
choosing the right hardware elements and software
code in numerous technical routes is a big research
and development (R&D) challenge.
EMULSION is being built by means of low-cost
electronic modules and open-source software
components, by utilizing the multi-tier IoT
architecture shown in Figure 1 and Figure 2. The
sensors (S), location trackers (T) [8], and
monitoring stations (MS), deployed in the sensor
tier for detecting and reporting changes occurring in
the physical world, communicate with the
information centers, located in the cloud tier, by
means of data/remote transfer units (D/RTUs) [9],
which support different wireless communication
standards for 2G÷6G cellular, LoRaWAN, WLAN
(Wi-Fi), Bluetooth Low Energy (BLE), etc.,
communications, and through smart gateways,
which ensure the required interoperability between
the involved heterogeneous IoT things, objects,
devices, nodes, etc. Additionally, wireless sensor
networks (WSNs) are established, where needed, for
extending the communication range and reaching
the corresponding communication gateway(s). This
way the information centers in the cloud tier can
collect and analyze the data coming from the sensor
3 https://github.com/
WSEAS TRANSACTIONS on SYSTEMS and CONTROL
DOI: 10.37394/23203.2022.17.15
Ivan Ganchev, Zhanlin Ji
E-ISSN: 2224-2856
135
Volume 17, 2022
tier as to come up with appropriate decisions and
recommendations that are sent back as a
configuration information and/or commands to the
respective actuators (A), controllers (C), and guards
(G), this way enforcing the necessary OAM actions
in the sensor tier and realizing the desired/required
changes in the physical world.
S
A
T
MS
Communication tier Cloud tier
2G÷6G/LoRa/Wi-Fi/BLE
IoT Nodes
D/RTU
Mobile APPs Web APPs
Client tierSensor tier
D/RTU
D/RTU
WSN
C
G
D/RTU
Smart
Gateways
Service tier
Figure 1. The IoT multi-tier architecture of EMULSION.
EMULSION comprises seven tiers, which are
depicted in Figure 2 and briefly described below:
1) Sensor tier – encompassing two groups of IoT
nodes: (i) sensors, monitoring stations, location
trackers [8], etc., for detecting and notifying about
the changes happening in the physical world; and
(ii) actuators, controllers, guards, etc., for imposing
required changes in the physical world.
2) Communication tier – comprising a variety of
D/RTUs [9] and smart communication gateways,
facilitating communication between IoT nodes in
the sensor tier, accessible via different types of
communication networks, and information centers
in the cloud tier, and ensuring the required
interoperability.
Sensor
tier
Middleware
tier
Cloud
tier
Service
tier
Communication
tier
Modelling &
simulation tier Client
tier
Fog
Gateways
Figure 2. The 7-tier structure of EMULSION.
3) Modelling & simulation tier – used for: (i)
modelling of CPS objects and IoT services (and
their inherent attributes and temporal/spatial/event
characteristics); and (ii) simulating the actual
provision of IoT services. This is a new type of tier,
proposed here for inclusion in the IoT platforms.
More details about this are provided in the next
section.
4) Client tier – containing different smart clients
and personal assistant applications, this tier
facilitates the consumers’ access to the services
supplied by the platform, by considering all the
current context in order to use the 'best' service
instances at any given moment, in conformance to
the Always Best Connected and best Served
(ABC&S) communication paradigm [10].
5) Service tier – providing two main types of
services: (i) regular IoT services, mostly intended
for the ‘smart environment control’ and ‘smart
health’ [11] [12] IoT domains. Each such service is
delivered in a highly personalized, customized, and
contextualized ABC&S way to each particular
consumer. The Internet of Services (IoS) [16] is
another principle taken on board for facilitating the
integration of different (atomic) IoT services into
composite services in such a way that allows for
their easy discovery, interlinking, mash up,
aggregation, and composition into a single service
WSEAS TRANSACTIONS on SYSTEMS and CONTROL
DOI: 10.37394/23203.2022.17.15
Ivan Ganchev, Zhanlin Ji
E-ISSN: 2224-2856
136
Volume 17, 2022
bundle; and (ii) supplementary services, used to
discover and recommend to each consumer the 'best'
instances of the regular IoT services, preferred by
him/her. A prototype recommendation service is
being developed for consumers to help them plan
pro-actively in advance, and change dynamically
afterwards if needed, the routes to follow in their
movement through poor air-quality/polluted areas,
in accordance with the consumers’ personal health
status. Such a novel health-related criterion (and
corresponding service) for route generation and
recommendation to consumers could (and should)
be included in the existing navigation systems and
applications alike. For achieving intelligent service
recommendations, suitable models, algorithms, and
techniques were elaborated [13] [14] for use by
EMULSION.
6) Middleware tier – operating as a medium
between all other tiers for ensuring their
interoperability [15] and as a central manager of the
IoT devices, this tier is concerned also with the
proper OAM of the IoT devices and their integration
into EMULSION. Moreover, this tier deals with
issues related to programmability, scalability, and
mapping of the available IoT devices to the supplied
IoT services. Applications scheduling is also done
by this tier [16].
7) Cloud tier – furnished with a Data
Management Platform (DMP) core, this tier deals
with ‘big data’ processing and analytics issues, and
converting the raw sensing data gathered from the
physical world into actionable analytic datasets that
could be properly used by the supplied IoT services.
For creating an efficient and effective DMP, the
Lambda architecture [17] was selected for
utilization (Figure 3).
Serving layer (HBase)
Batch layer
(Hadoop)
Speed Layer
Data stream
Stream procesing
Realtime view
All data
Precompute views
Batch view Batch view
Query
Figure 3. The utilized Lambda architecture.
A distributed, Kafka4-based, publish-subscribe
messaging module was designed for use in the data
stream part for achieving a high-throughput data
processing (Figure 4). A second Kafka-based
module was designed for the data subscription and
archival of messages in the query part, by means of
ETL batched consumption operations. For fetching
the available topics and storing these on HDFS5, a
Camus6-based pipeline between Kafka and HDFS is
used. For the DMP functions encapsulation, a
shared interface class was elaborated.
Figure 4. The elaborated DMP architecture.
Figure 5 illustrates the ETL operation within
DMP, involving: (i) consumption of topics from
Kafka; (ii) byte stream formatting with a relevant
Avro schema; and (iii) Kafka messages’
serialization with corresponding topics to HDFS.
Kafka
filtered-
logs-topic
Classifi
-cation
load resource
to memory
HDFS
HTTP
mobile
keyword-topic
tag-topic
Statistics-
topic
camus job
camus job
keyword
tag
Statistics
default
shopping
……
default
app
……
……
…… Redis
Figure 5. The ETL operation within DMP.
4 https://kafka.apache.org/
5 https://hadoop.apache.org/docs/r1.2.1/hdfs_design.html
6 https://docs.confluent.io/1.0/camus/docs/intro.html
WSEAS TRANSACTIONS on SYSTEMS and CONTROL
DOI: 10.37394/23203.2022.17.15
Ivan Ganchev, Zhanlin Ji
E-ISSN: 2224-2856
137
Volume 17, 2022
4 Modelling & Simulation Tier of
EMULSION
The goal of using this tier is twofold: (i) to model
CPS objects and IoT services, supplied by the
EMULSION platform, along with their inherent
attributes, and temporal/spatial/event characteristics.
The Generalized Nets modelling [18] [19] is applied
here to optimize the queuing disciplines used for
serving the incoming requests for IoT services
provided by EMULSION; (ii) to simulate the actual
provision of IoT services as to determine the
optimal platform configuration in each particular
use case, by solving complex optimization tasks, as
described in the next subsections.
4.1 Determination of the optimal schemes
for the deployment of sensors and/or
monitoring stations
This task aims at determining the optimal number of
sensors and/or monitoring stations (in the sensor
tier) needed to cover a particular area of interest as
to minimize the capital expenditure (CAPEX) and
operating expenditure (OPEX) while maintaining
high key performance indicators’ (KPIs) values. For
instance, in the ‘smart environment control’ IoT
domain, which is initially targeted by EMULSION,
an important service relates to the continuous
monitoring of the air quality index (AQI) in the
target area(s) for the purposes of daily live reporting
based on geo-grid positioned AQI monitoring
stations, as shown in Figure 6.
AQI monitoring station
Existing road/track
Area of interest
Figure 6. Using a geo-grid of AQI monitoring stations
for covering a target area: the initial positioning of
stations.
In addition, the AQI data obtained by such geo-
grid are used also for short/long-term forecasting of
the future atmospheric environment conditions in
the target area(s), for the purposes of smart pro-
active route planning, performed by the
corresponding pilot prototype application, on behalf
of consumers, as to find and recommend to them the
‘healthiest’ possible route to follow through that
area(s).
However, taking into account that people in their
movement, either on foot or by bike/vehicle, follow
the tracks or roads existing in the area, an
adjustment of the position of the AQI monitoring
stations is required as to cover the existing
track/road map, as shown in Figure 7.
AQI monitoring station
Existing road/track
Area of interest
Figure 7. Using a geo-grid of AQI monitoring stations
for covering a target area: the final positioning of
stations after adjustment made by the EMULSION’s
modelling & simulation tier by keeping the same
number of stations in the area.
4.2 Determination of the optimal schemes
for the deployment of communication
gateways
This task aims at determining the optimal schemes
for the deployment of (short-range) communication
gateways (and associated fog nodes), given the
actual/existing position of the sensors and/or
monitoring stations already deployed in the sensor
tier, as to minimize the CAPEX and OPEX while
maintaining high KPI values of EMULSION. In the
‘smart environment control’ IoT domain, which is
initially targeted by EMULSION, the key factor in
this task is the maximum communication range
supported by the corresponding wireless
communication standard, utilized by the AQI
monitoring stations. Figure 8 illustrates the
solution for the sample use case presented in Figure
7. As can be seen from Figure 8, just 5 short-range
communication gateways (of a particular type) are
WSEAS TRANSACTIONS on SYSTEMS and CONTROL
DOI: 10.37394/23203.2022.17.15
Ivan Ganchev, Zhanlin Ji
E-ISSN: 2224-2856
138
Volume 17, 2022
needed to serve all monitoring stations deployed in
the target area. In addition, it is advisable to allow
the positioning of stations, if possible at all, in the
serving area of not just one but multiple
communication gateways for achieving higher
reliability of operation. In other words, the AQI
monitoring stations, located in the cross sections of
the serving areas of two (or more) gateways, have
the advantage of using secondary gateway(s) in the
case of any fault, malfunctioning, or
misconfiguration of the primary gateway they have
been initially assigned to. The primary gateway in
this case is determined in such a way as to achieve
(almost) equal load for all gateways deployed in the
area. This is shown in Figure 8, by using the same
color for all monitoring stations assigned to a
primary gateway of the same color. In this use case,
each communication gateway gets equal load of
serving 7 monitoring stations.
AQI monitoring station
Existing road/track
Area of interest
Communication gateway
with its serving area
Figure 8. The optimal locations of the communication
gateways, serving the AQI stations in a target area,
as determined by the modelling & simulation tier of
EMULSION.
In addition, for easy access and management, it
could be preferable to have a communication
gateway mounted at the same point where a
monitoring station is positioned. In this case, some
additional adjustment of the communication
gateways’ locations might be needed as shown in
Figure 9. However, this may lead to imposing a
non-equal load to communication gateways. For
instance, in the use case presented in Figure 9, the
light-green colored gateway gets the least load (of 6
monitoring stations), whereas the blue-colored
gateway gets the highest load (of 8 monitoring
stations), compared to the remaining mid-loaded
gateways (each serving 7 monitoring stations).
AQI monitoring station
Existing road/track
Area of interest
Communication gateway
with its serving area
Figure 9. The final locations of the communication
gateways, co-positioned with some of the AQI
monitoring stations in the area.
4.3 Determination of the optimal cloud
configuration
This task aims at determining the optimal
configuration of the platform’s cloud tier as to
achieve an efficient provision of the corresponding
IoT services. The results of performing this task
with respect to determining the optimal (combined)
configuration of the Kafka Brokers (KBs) and
Storm Supervisors (SSs) in the cloud tier, are shown
on Figure 10.
Figure 10. The number of messages per second,
processed by different KB/SS cloud configurations
of EMULSION.
The presented results were obtained for the case
of sending 10,000,000 messages from the sensor tier
to the cloud tier, by utilizing different number of
concurrency threads – starting with only 10 threads
(with 1,000,000 messages processed by each thread)
and ending up with 200 threads (with 50,000
WSEAS TRANSACTIONS on SYSTEMS and CONTROL
DOI: 10.37394/23203.2022.17.15
Ivan Ganchev, Zhanlin Ji
E-ISSN: 2224-2856
139
Volume 17, 2022
messages processed by each thread). In the case
considered here, the most efficient operation of the
platform was achieved by using a 3-KBs/3-SSs
cloud configuration, utilizing 100 threads (with
100,000 messages processed by each thread).
The presented results of the performance of the
Kafka-Storm-Kafka part of the DMP core of the
cloud tier clearly demonstrate that determining the
KB/SS appropriate numbers for use is of vital
importance for achieving optimal operation of the
EMULSION platform in real use case scenarios.
5 Conclusion
Elaborated as a typical example of the horizontal
IoT platforms, the presented EMULSION IoT
platform is able to meet the today’s requirements for
achieving multi-dimensional flexibility, scalability,
interoperability, and easy adjustment to new-
emerging IoT scenarios and use cases. By
considering the current consumer/service/network
context, the designed platform is able to supply
highly personalized, customized, and contextualized
IoT services by utilizing distributed real-time ‘big
data’ processing and analytical techniques in the
cloud. This way, EMULSION can convert the
collected sensing data and information about the IoT
service activities of consumers into rich analytic
datasets, which are utilized for the proactive real-
time recommendation of the 'best' service instances
for use by each individual consumer in conformance
to the ABC&S communication paradigm [10].
While focusing primary on the ‘smart environment
control’ and ‘smart healthcare’ IoT domains,
EMULSION has the capacity for the supply of
services in other IoT domains as well.
Specific attention in this paper has been paid to
presenting some aspects of the modelling &
simulation tier of this platform, which is a novel
architectural element, proposed here for inclusion in
similar IoT platforms. The utilization of such tier
allows to perform different optimization tasks in
order to obtain the optimal platform configuration in
each particular use case and thus achieve an
efficient provision of IoT services and the best
quality of experience (QoE) delivered to consumers
anytime-anywhere-anyhow.
Future R&D work will be focused on the
elaboration of effective models of advanced IoT
services for provision by EMULSION.
References:
[1]. Wopata, M., 5 Things to Know About the IoT
Platforms Market. 2021, IoT Analytics.
[2]. Wegner, P., IoT Platform Companies
Landscape 2021/2022: Market consolidation
has started. 2021, IoT Analytics.
[3]. Lueth, K.L., IoT Platform Companies
Landscape 2019/2020: 620 IoT Platforms
globally. 2019, IoT Analytics.
[4]. Ganchev, I., Z. Ji, and M. O’Droma. A Generic
Multi-Service Cloud-Based IoT Operational
Platform - EMULSION. in 2019 International
Conference on Control, Artificial Intelligence,
Robotics & Optimization (ICCAIRO). 2019.
[5]. Ganchev, I., Z. Ji, and M. O'Droma. A generic
IoT architecture for smart cities. in Irish
Signals & Systems Conference 2014 and 2014
China-Ireland International Conference on
Information and Communications Technologies
(ISSC 2014/CIICT 2014). 25th IET. 2014. IET.
[6]. Yaici, K., IoT forecast: connectivity
management platforms 2019–2028, in IoT
Platforms and Technology. 2020.
[7]. Kontos, P., IoT platform enterprise contracts
tracker 2H 2020, in IoT Platforms and
Technology. 2021.
[8]. Ganchev, I., Z. Ji, and M. O’Droma, Designing
a Low-Cost Location Tracker for Use in IoT
Applications, in XXXIII General Assembly and
Scientific Symposium of the International
Union of Radio Science (URSI GASS 2020).
2020, IEEE. p. x1-x2.
[9]. Ganchev, I., Z. Ji, and O.D. M., Low-cost and
ultra-low-power consuming RTUs for use in
IoT systems. Int. J. Circ., Syst. and Sig.
Process., 2020. 14: p. 76–82.
[10]. Ganchev, I., User-Centric and Context-Aware
ABC&S. International Journal "Information
Technologies & Knowledge", 2015. 9(3): p.
203-215.
[11]. Jing, E., et al., ECG Heartbeat Classification
Based on an Improved ResNet-18 Model.
Comput Math Methods Med, 2021. 2021: p.
6649970.
[12]. Wang, Y.F., et al., RG Hyperparameter
Optimization Approach for Improved Indirect
Prediction of Blood Glucose Levels by
Boosting Ensemble Learning. Electronics,
2021. 10(15): p. 1797.
[13]. Zhou, D., et al., Novel SDDM Rating
Prediction Models for Recommendation
Systems. IEEE Access, 2021. 9: p. 101197-
101206.
[14]. Zhang, H., et al., FeatureMF: An Item Feature
Enriched Matrix Factorization Model for Item
Recommendation. IEEE Access, 2021. 9: p.
65266-65276.
WSEAS TRANSACTIONS on SYSTEMS and CONTROL
DOI: 10.37394/23203.2022.17.15
Ivan Ganchev, Zhanlin Ji
E-ISSN: 2224-2856
140
Volume 17, 2022
[15]. Razouk, W., D. Sgandurra, and K. Sakurai, A
new security middleware architecture based on
fog computing and cloud to support IoT
constrained devices, in Proceedings of the 1st
International Conference on Internet of Things
and Machine Learning. 2017, Association for
Computing Machinery: Liverpool, United
Kingdom. p. Article 35.
[16]. Lee, H., et al., A novel service-oriented
platform for the internet of things, in
Proceedings of the Seventh International
Conference on the Internet of Things. 2017,
Association for Computing Machinery: Linz,
Austria. p. 1-8.
[17]. Perera, S. and S. Suhothayan. Solution patterns
for realtime streaming analytics. in
Proceedings of the 9th ACM International
Conference on Distributed Event-Based
Systems. 2015. ACM.
[18]. Andonov, V., S. Poryazov, and E. Saranova,
Generalized net representations of control
structures in service systems theory. Advanced
Studies in Contemporary Mathematics
(Kyungshang), 2020. 30(1): p. 49-60.
19]. Andonov, V., S. Poryazov, and E. Saranova,
Generalized net representations of elements of
service systems theory. Advanced Studies in
Contemporary Mathematics (Kyungshang),
2019. 29(2): p. 179-189.
Ivan Ganchev carried out the conceptualization,
methodology, project administration and
supervision, funding acquisition, writing of the
original draft and editing of the final paper.
Zhanlin Ji carried out data curation, formal analysis,
investigation, validation, software
development, and resource management.
This publication has emanated from joint research
conducted with the financial support of the
Bulgarian National Science Fund (BNSF)
under the Grant No. KP-06-IP-CHINA/1 (КП-
06-ИП-КИТАЙ/1) and the S&T Major Project
of the Science and Technology Ministry of
China, Grant No. 2017YFE0135700.
Creative Commons Attribution License 4.0
(Attribution 4.0 International, CC BY 4.0)
This article is published under the terms of the Creative
Commons Attribution License 4.0
https://creativecommons.org/licenses/by/4.0/deed.en_US
WSEAS TRANSACTIONS on SYSTEMS and CONTROL
DOI: 10.37394/23203.2022.17.15
Ivan Ganchev, Zhanlin Ji
E-ISSN: 2224-2856
141
Volume 17, 2022