Advantages of IoT-Based Geotechnical Monitoring
Systems Integrating Automatic Procedures for Data
Acquisition and Elaboration
Andrea Carri 1, Alessandro Valletta 2 ,* , Edoardo Cavalca 2, Roberto Savi 1and Andrea Segalini 2
Citation: Carri, A.; Valletta, A.;
Cavalca, E.; Savi, R.; Segalini, A.
Advantages of IoT-Based
Geotechnical Monitoring Systems
Integrating Automatic Procedures for
Data Acquisition and Elaboration.
Sensors 2021,21, 2249. https://
Aime’ Lay-Ekuakille and
Received: 9 December 2020
Accepted: 19 March 2021
Published: 23 March 2021
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional afﬁl-
Copyright: © 2021 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
1ASE—Advanced Slope Engineering S.R.L., Via Robert Koch 53/a, Fraz. Pilastrello, 43123 Parma, Italy;
firstname.lastname@example.org (A.C.); email@example.com (R.S.)
2Department of Engineering and Architecture, University of Parma, Parco Area delle Scienze 181/a,
43124 Parma, Italy; firstname.lastname@example.org (E.C.); email@example.com (A.S.)
*Correspondence: firstname.lastname@example.org; Tel.: +39-0521-905952
Monitoring instrumentation plays a major role in the study of natural phenomena and
analysis for risk prevention purposes, especially when facing the management of critical events.
Within the geotechnical ﬁeld, data collection has traditionally been performed with a manual ap-
proach characterized by time-expensive on-site investigations and monitoring devices activated by
an operator. Due to these reasons, innovative instruments have been developed in recent years in
order to provide a complete and more efﬁcient system thanks to technological improvements. This
paper aims to illustrate the advantages deriving from the application of a monitoring approach,
named Internet of natural hazards, relying on the Internet of things principles applied to monitoring
technologies. One of the main features of the system is the ability of automatic tools to acquire
and elaborate data independently, which has led to the development of dedicated software and
web-based visualization platforms for faster, more efﬁcient and accessible data management. Ad-
ditionally, automatic procedures play a key role in the implementation of early warning systems
with a near-real-time approach, providing a valuable tool to the decision-makers and authorities
responsible for emergency management. Moreover, the possibility of recording a large number
of different parameters and physical quantities with high sampling frequency allows to perform
meaningful statistical analyses and identify cause–effect relationships. A series of examples deriving
from different case studies are reported in this paper in order to present the practical implications of
the IoNH approach application to geotechnical monitoring.
Keywords: landslide; monitoring; Internet of things; real time; early warning systems
Nowadays, monitoring systems play a key role in the study and description of natural
phenomena, together with the analyses for what concerns risk prevention. This allows
obtaining useful information for continuous and optimal management of infrastructures,
landslides and critical events in general [
]. The geotechnical ﬁeld has traditionally been
characterized by a manual approach regarding both data collection and on-site investiga-
tions. This methodology can be quite complex and time-consuming [
], particularly for
environments and sites difﬁcult to access directly, especially during hazardous events. Due
to these reasons, innovative tools integrating different sensors have been developed in the
last two decades thanks to technological improvements [
]. The main goal is to produce a
complete and efﬁcient system featuring automatic procedures and characterized by higher
accuracy, reliability and durability compared to traditional devices.
One of the major advancements introduced by innovative instrumentations is the
possibility to overcome the so-called “manual philosophy”, thanks to the integration
of Internet of things (IoT) technologies in the design process of automated monitoring
Sensors 2021,21, 2249. https://doi.org/10.3390/s21062249 https://www.mdpi.com/journal/sensors
Sensors 2021,21, 2249 2 of 18
]. When it comes to landslide monitoring, the integration of IoT-based devices
results in a multilayered structure, where each component has a key role in guaranteeing
the system functionality and efﬁciency [
]. While the architecture could vary from one
case study to another, it is possible to identify some essential components for IoT-based
The perception layer, consisting of a network of connected sensors responsible for
collecting the data and transmit them to the elaboration center;
The data layer, which entails dataset storage and elaboration. This is an essential
component in an IoT-based system since traditional approaches are usually unable to
manage the amount of data generated by an automatic system;
- The presentation layer, where monitoring outcomes are represented and made avail-
able for the end-user through dedicated mobile applications or web-based services.
This approach has found successful applications in the geotechnical monitoring ﬁeld,
with the integration of both traditional and innovative sensors for monitoring and early
warning purposes. Some examples include the multi-device wireless monitoring of a
sloped surface [
], as well as sensors columns for the detection of the formation of a slip
], and optical instrumentation to acquire images and detect potentially critical
]. In all these conﬁgurations, the main advantage relies on the possibility of
achieving a real-time or near-real-time monitoring approach through the integration of
automated processes and multiparameter tools. Moreover, IoT-based monitoring systems
have been employed in underground excavation works, speciﬁcally for early warning
purposes. For example, multiparameter approaches have been introduced in subway
construction sites, where traditional EWS based on single-sensor data and simple models
are not suitable for such complex frameworks [
]. In fact, more recent systems are able
to acquire information from multiple sensors thanks to the integration of wireless networks
to connect every device installed on-site [16–18].
The implementation of IoT-based procedures is the core component of the Internet of
natural hazards (IoNH) approach [
]. IoNH represents a design concept where different
system components can interact and exchange data thanks to Internet connectivity. This is
achieved by exploiting a bidirectional remote control of different sensors and control units
(CU) placed in a speciﬁc monitoring site, taking advantage of a cloud-based database (DB),
software and representation platform. Equipped with appropriate technology, network-
connected devices can be remotely controlled and accessed, thus creating an integrated
and automated monitoring system with improved efﬁciency and performance.
The ability of IoT tools to automatically acquire and elaborate monitoring data has
laid the basis for the development of dedicated self-learning or machine-learning automatic
software, able to analyze the dataset and apply self-check controls in order to provide a
preliminary validation of raw data, excluding outliers and sensors errors. The statistical
validation is possible thanks to the high sampling rate available with modern technologies
(e.g., CU and DB services). IoNH also relies on the possibility to record a large number of
different parameters and physical quantities, obtaining a more accurate description of the
ongoing phenomenon. This aspect permits to deﬁnition cause–effect relationships or failure
predictions, sometimes only based on numerical data. Moreover, an automatic procedure
would permit monitoring even rapid and impulsive events that manually operated tools
would not be able to identify due to their lower sampling frequency.
The management process requires the design and application of web-based represen-
tation platforms, making the monitoring results easily and quickly accessible for further
analyses and interpretation [
]. This aspect should not be underestimated since the
timely evaluation of sensor outcomes leads to effective intervention strategies, e.g., people
evacuation from an area at risk or road closure before the occurrence of a critical event.
Additionally, automatic systems based on solid calculation algorithms are a key component
in the design of landslide early warning systems (EWS), which need to be able to detect
landslide activities as soon as possible .
Sensors 2021,21, 2249 3 of 18
This paper will focus on the features and applications of the IoNH system, which
is intended to provide a new methodology to overcome the traditional approach that
characterizes the geotechnical monitoring ﬁeld. Particular attention will be dedicated to
the different advantages resulting from the integration of these IoT-based components into
automatic monitoring systems. Moreover, a series of examples taken from different case
studies are reported in order to underline the methodology value in different aspects of the
monitoring activity, e.g., power supply control, data elaboration procedures, cause–effect
correlations, etc. Additionally, the following sections report a description of the IoNH
structure and components, also exploring some details regarding the algorithms involved
in the elaboration process.
2. Materials and Methods
The IoNH approach is based on the experience acquired along 7 years of automatic
geotechnical monitoring through a modular underground monitoring system (MUMS),
from on-site sensors acquisition to results representation. MUMS was designed as an
automatic inclinometer composed of several nodes (links) located at custom distances
along a single chain, linked by an aramid ﬁber and a quadrupole electrical cable. The
array functionality is automated, collecting multiparametric data through a remotely
controlled datalogger and RS485 protocol [
]. Since its ﬁrst application as a device to
monitor landslide displacements, MUMS technology has been developed in order to adapt
to different geotechnical scenarios, and it is nowadays the core of several tools intended
for other applications (e.g., underground excavations, civil and geotechnical structures,
rockfall and debris ﬂow barriers, and geothermal plants). The IoNH working principle is
common to all these applications and is summarized in Figure 1.
Internet of natural hazards (IoNH) approach applied to modular underground monitoring
system (MUMS) system, with data collection, transmission, database storage, automatic elaboration,
results representation, alarms activation.
2.1. Control Unit
CU queries sensors at the speciﬁed sample period and applies the ﬁrst ﬁlter on
the raw dataset, performing 64 readings and saving the mean value in an external SD
card. Moreover, the datalogger is able to record any kind of traditional sensors featuring
an analog output signal. This particularly useful feature permits the management of
complex and diversiﬁed monitoring systems with a single logger, DB, software and web
platform, with relevant advantages in terms of data management and visualization. The
CU power supply comes from a lead battery recharged through a solar panel or, if possible,
directly connected to the electrical line. Both these conﬁgurations are intended to provide
a reliable power source, which is an absolute necessity for any acquisition system [
Sensors 2021,21, 2249 4 of 18
Moreover, the elaboration software analyses the voltage and automatically sends warning
messages, while the battery level is available in the web-platform. This aspect represents a
signiﬁcant upgrade with respect to traditional automated, non-remotely controlled devices
since the inability to verify the control unit functionality could lead to potential losses of
As anticipated in previous paragraphs, one of the main features of the IoNH approach
is the possibility to integrate a wide selection of different sensors. This characteristic is
intended to provide a ﬂexible and customizable system, able to adapt to different contexts
according to the relevant parameters to be monitored. The most common conﬁguration,
designed to measures displacements in landslides and underground constructions, re-
lies on microelectromechanical system (MEMS) sensors, composed of an accelerometer,
a magnetometer, and a thermometer. In those cases where a higher resolution is needed
(e.g., monitoring of structures and buildings), it is possible to integrate an electrolytic
tilt sensor. It is worth mentioning that MEMS have a 360
measuring range, while elec-
trolytic cells feature a full-scale value of
. Additionally, the structure of MUMS-based
arrays permits the exploitation of different sensor typologies in the same array, such as
piezometers to detect water level variations [
] or high-accuracy thermometers to have
a detailed description of the underground temperature at different depths [
IoNH allows the acquisition and elaboration of monitoring data sampled by traditional
instrumentation featuring an analog output signal (e.g., extensometers, crack meters, load
cells, etc.), which can also be integrated into a monitoring system, including digital output
tools. This particular conﬁguration can usually be observed when dealing with tunnels
and underground excavations, where the monitoring activity involves a large number of
different elements of the structure and the surrounding environment .
2.3. Data Transmission and Storage
The data transmission generally relies on 4G, 3G or GPRS lines with a UMTS router,
while in some cases, it is also possible to use ﬁber optics leading to a local elaboration
center. Raw data (electrical signals) are stored in a cloud DB and always available for a
back analysis or subsequent elaborations by exploiting new algorithms or statistical ﬁlters.
In the case of monitoring systems based on the IoNH structure, the elaboration process
takes into account several variables, resulting in a quite complex and potentially expensive
procedure from a computational point of view. For this reason, the system is designed to
transmit raw data and calculate the results at the elaboration center or, alternatively, run
on a speciﬁc elaboration center on-site.
2.4. Elaboration and Self-Check Controls
The elaboration algorithm, speciﬁcally designed for each MUMS-based application,
converts electrical signals into physical units, removes spike noises and accidental errors,
and calculates the ﬁnal results. Self-check data controls are a fundamental aspect when
dealing with the management of automated tools. In particular, MUMS devices exploit a
multilevel automated control procedure implemented in the elaboration algorithm, relying
on the veriﬁcation of speciﬁc physical entities listed below :
I. Recognition of spike noise;
II. Coherence of gravity acceleration vector;
III. Variability of gravity acceleration vector;
IV. Coherence of magnetic ﬁeld vector;
V. VVariability of magnetic ﬁeld vector;
VI. Coherence of node temperature;
VII. Running average;
VIII. Recognition of instrumental noise;
IX. Recognition of not-operating sensors.
Sensors 2021,21, 2249 5 of 18
2.4.1. Recognition of Spike Noise (I)
Spike noise (or impulse noise) is a sporadic and impulsive electrical disturb, related to
sharp and sudden disorder in the signal, often related to external factors, which heavily
contaminate the acquired data [
]. The identiﬁcation of spike noise exploits statistical tools
over a speciﬁc dataset. For example, the MUMS algorithm analyses the statistical dispersion
evaluating the variability of a univariate sample of quantitative data with the scaled median
absolute deviation (MAD). MAD represents a measure of the statistical dispersion, and
it is deﬁned as the median of the absolute deviations from the data’s median [
in Equation (1). The removal of accelerometer spike noise is particularly important in
some applications, such as tunneling and underground excavations in general, where the
vibrations induced by works could highly inﬂuence the single data-point, resulting in a
well-identiﬁable outlier :
√2∗er f c−13
er f c(x)=2
xe−t2dt =1−er f (x)(3)
er f (x)=2
•Xiis the data to be evaluated;
•er f c(x)is the complementary error function;
•er f (x)is the error function .
The recognition of spike noises shows high performances if the dataset is centered [
For this reason, when dealing with real-time data transmission, the calculation of the last
part of the data sample is not completely validated until its time window is fully centered.
The immediate consequence is a time delay for the validation of the results, strictly related
to the number of points considered, the sample period, and the data transmission frequency
(as in Equation (5)):
•outiis the identiﬁcation process of outlier i;
•tsp is the sampling period;
•tf t is the frequency of data transmission;
•dwrepresents the dimension of the dataset window.
parameters should be deﬁned accordingly to the monitoring needs.
The sampling period is related to the most probable phenomenon evolution, with a higher
sampling rate linked to a rapid displacement (Equation (6)). Thanks to the improved
connection between elements included in IoNH systems, it is possible to update the
value during the monitoring process, with an automated or a manual procedure performed
by remote. Moreover,
is strongly dependent on the dataset window dimension (related
to the robustness of the statistical analysis) and electrical power (ep), as in Equation (6):
The frequency of data transmission is an important parameter because it determines
the data elaboration and consequently the time period (
) related to the providing of ﬁnal
results (Equation (7)). This parameter should coincide with the sampling period if the mon-
Sensors 2021,21, 2249 6 of 18
itoring has early warning system purposes in order to have timely information regarding
the phenomenon evolution. Therefore, this process usually requires more electrical power
supply than the sampling procedure, especially for remote areas where the 3G phone line
is not well covered. These parameters should be considered at the monitoring design stage
in order to have a proper power supply (where it is possible):
tf t =fep,v,dw,tr(7)
Finally, regarding the dimension of the dataset window (
), it should be considered
that a reduced dataset validates the raw data in a shorter time but could not be statistically
]. A large dataset strengthens the statistical processes and recognition of instru-
mental and accidental noise, but it comes with an increase in validation and calculation
time. Therefore, a monitoring system involving these procedures should be described as
a “near-real-time” application instead of “real time”, due to all processes needed for data
acquisition and elaboration .
2.4.2. Coherence and Variability of Gravity Acceleration and Magnetic Field Vector (II, III,
IV and V)
MUMS system usually relies on 3D MEMS sensors featuring three main compo-
nents, namely an accelerometer, a magnetometer, and a thermometer. Thanks to its three-
dimensional functioning principle, it is possible to reconstruct the magnetic ﬁeld vector,
represented by gravity acceleration g. Moreover, the static nature of these measures allows
controlling the coherence and variation of instrumental data at every step. The coherence,
as in Equation (8), identiﬁes sensors with relevant malfunctions that would undermine
their calibration. The variation (Equation (9)) could be related to instrumental noise that
characterizes every electronic device, and it is eligible under a certain threshold (
should be deﬁned accordingly to sensor technical features. Automated ﬁlters identify
uncorrected values and adjust the related results, giving a preliminary software validation
to the dataset. Since they exploit the resultant vector in a 3D space, previously mentioned
controls are not possible using a 2D MEMS:
2.4.3. Running Average (VII)
Running average process is complementary to the recognition of spike noise, and it is
applied to identify and reduce instrumental noise by eliminating ﬂuctuations smoothing
the analyzed dataset [
]. The operation is fundamental when the expected displacements
have the same order of magnitude as instrumental sensitivity and repeatability [
gether with the mentioned advantages in terms of the improved sensor performances,
this operation should be avoided when rapid and impulsive displacements are expected.
In this speciﬁc case, the acceleration phase is highly anticipated, consistently reducing
the displacement rate and the correct phenomenon identiﬁcation. A possible solution
could be obtained by increasing the sampling period (with consequently incremented
2.5. Web Platform and Alarms
Outcomes of interest are represented through a dynamic and intuitive web platform
with secure, controlled access. This passage is of fundamental importance since the in-
stallation of a huge number of instruments with multiparametric data requires adequate
Sensors 2021,21, 2249 7 of 18
data management (following this approach, the design process should aim to simplify and
optimize the visualization and interpretation of results, leaving time-expensive operations,
such as organization and representation processes to automated systems.
Web-based environments provide several advantages in terms of accessibility and
performances, thanks to the implementation of a remote-user-friendly interface, which
allows interacting with a continuously updated system [
]. Moreover, the integration of
automatic processes gives the possibility to customize the platform, allowing the users to
interact easily with monitoring data according to speciﬁc needs (e.g., selecting a speciﬁc
time interval, downloading data plots in several formats, etc.) Finally, this approach
permits to conﬁgure automated reports, which periodically summarize the site conditions.
The main objective is to highlight speciﬁc events like the increasing of displacement rates or
pore pressure, together with other conditions diverging from the historical data trends. The
management of a large number of multiparameter data with high sampling frequencies
sometimes permits the establishment of signiﬁcant cause effects relationships, with direct
correlations between rainfalls, water levels and displacements over speciﬁc periods.
A smart IoNH management system should automatically change CU conﬁguration
parameters, like sampling period, data transmission frequency, etc., according to monitor-
ing outcomes. In this way, it is possible to overcome some issues related to the elaboration
process, e.g., the disadvantages related to the application of running average, as mentioned
before. In addition, an increase in the sampling period highly improves the performances
of failure forecasting models for early warning applications [
]. On the other hand, if the
battery voltage is reaching critical levels, the management system should automatically
reduce sampling and transmission frequency.
The activation of alert or alarm procedures is another key element that composes an
IoNH system. Traditional methods involve the deﬁnition of thresholds and, consequently,
the activation of predeﬁned procedures at their overcome. This philosophy requires the def-
inition of the landslide mechanical model or, in general, the monitored phenomenon [
The approach could be characterized by multilevel or single-level solutions, involving a
very clear distinction without any intermediate level, and its effectiveness depends on
several critical factors. First, it is necessary to carry on-site investigations in order to create
the landslide geological and geotechnical model. This should be followed by a mechani-
cal numerical model that tries to reproduce the monitoring outcomes [
] starting from
materials characteristics. In general, the procedure requires several months of monitoring
data and the employment of well-prepared technicians. This leads to signiﬁcant time and
cost requirements that should be taken into account since their availability should not
be taken for granted. Other methods exploit only monitoring data without taking into
account any mechanical description of the phenomenon. These approaches use failure
forecasting models, like the Inverse Velocity Method [37,38], in order to deﬁne thresholds
or activate warning procedures related only to displacement rate and acceleration data.
These approaches involve some hypotheses and assumptions in order to be applied and are
not able to guarantee a completely accurate representation of every possible phenomenon.
Therefore, it is usually recommended not to apply these models in isolation to avoid
possible misinterpretations .
Finally, another application takes advantage of trigger or shock sensors. These in-
struments can be mechanical or electronic devices and are designed to detect impulsive
phenomena like the impact of a mass against a barrier or a debris ﬂow occurrence [
Depending on the device, these sensors usually consist of a steel cable, a dynamic MEMS
or a geophone. In the ﬁrst case, the cable could be wrapped up in the barrier, activating
a trigger signal when the rope is pulled. A dynamic MEMS works as a shock sensor
and issues warnings at the overcoming of a predetermined acceleration threshold. The
geophone has a very similar working principle as a dynamic MEMS sensor and records
vibrations or seismic waves that could anticipate the arrival of a critical event. The trigger
activation could directly lead to alert procedures, or it could be integrated within cross-
check controls. These processes analyze other relevant physical entities like steel posts
Sensors 2021,21, 2249 8 of 18
tilt or load exerted on ropes. As an example, an application of IoNH to rockfall barriers
monitoring is represented in Figure 2. A rock mass hits the barrier, pulling a mechanical
device. This one triggers the CU, which starts to read every sensor connected with a local
smart mesh network and activates a photoshoot captured with a small camera. Information
is sent to the elaboration center, where a self-learning software analyses the sensor results.
This operation could be tackled in different ways by using absolute thresholds, relative
thresholds or by comparing the last data received with the previous data trend. According
to predetermined conditions or warning levels, the procedure could send emails (with
the photoshoot attached) and/or SMS to the authorities in charge while activating sirens
and/or trafﬁc lights placed on the site access roads at the same time.
Sensors 2021, 21, x FOR PEER REVIEW 8 of 18
Finally, another application takes advantage of trigger or shock sensors. These in-
struments can be mechanical or electronic devices and are designed to detect impulsive
phenomena like the impact of a mass against a barrier or a debris flow occurrence [40–42].
Depending on the device, these sensors usually consist of a steel cable, a dynamic MEMS
or a geophone. In the first case, the cable could be wrapped up in the barrier, activating a
trigger signal when the rope is pulled. A dynamic MEMS works as a shock sensor and
issues warnings at the overcoming of a predetermined acceleration threshold. The geo-
phone has a very similar working principle as a dynamic MEMS sensor and records vi-
brations or seismic waves that could anticipate the arrival of a critical event. The trigger
activation could directly lead to alert procedures, or it could be integrated within cross-
check controls. These processes analyze other relevant physical entities like steel posts tilt
or load exerted on ropes. As an example, an application of IoNH to rockfall barriers mon-
itoring is represented in Figure 2. A rock mass hits the barrier, pulling a mechanical de-
vice. This one triggers the CU, which starts to read every sensor connected with a local
smart mesh network and activates a photoshoot captured with a small camera. Infor-
mation is sent to the elaboration center, where a self-learning software analyses the sensor
results. This operation could be tackled in different ways by using absolute thresholds,
relative thresholds or by comparing the last data received with the previous data trend.
According to predetermined conditions or warning levels, the procedure could send
emails (with the photoshoot attached) and/or SMS to the authorities in charge while acti-
vating sirens and/or traffic lights placed on the site access roads at the same time.
IoNH approach applied to a rockfall event and its impact on protection barriers, followed by
trigger activation, data collection and transmission, database storage, data processing, and activation
of warning procedures.
3. Results and Discussion
As discussed above, a major task of an IoNH system is the monitoring of CU battery
voltage. Due to their nature, rechargeable lead–acid batteries lose their functionality when
the level goes down a critical value, causing a potential loss of data. For this reason, the
automated software is able to send warning messages to monitoring responsible for the
occurrence of critical situations in order to take timely actions and restore the system
functionality. Figure 3presents an example of this procedure, displaying a case study
where the battery voltage monitoring allowed to identify relevant reductions of the charge
Sensors 2021,21, 2249 9 of 18
level. In particular, the system involved an array of 23 high-resolution thermometers for
the monitoring of soil temperature at different depths. The instrumentation was set on a
10-min sampling frequency, while a lead–acid battery connected to a solar panel provided
the power supply to the monitoring system. After the tool installation, a prolonged period
of bad weather in the area hindered the ability of the solar panel to properly recharge the
battery. This event, coupled with the particularly high sampling frequency, generated a
decreasing trend of energy supply. Thanks to alert messages issued in correspondence to
these events, it was possible to address the problem with appropriate on-site operations,
thus avoiding more serious malfunctions. In particular, after the substitution of the battery
and the upgrading of the solar panel with a more powerful one, at the beginning of
December, the system reached a stable conﬁguration and was able to continue regularly
the monitoring activity.
Figure 3. Battery voltage monitoring over time, with three different charge levels.
The removal of outliers and spike noise is a relevant topic when dealing with EWS,
even if the process causes a time-delay related to the complete dataset validation. For
example, Figure 4reports a monitoring dataset characterized by a sample period of
a statistical analysis that considers 11 elements and a data transmission frequency to
the elaboration center of 30 min [
]. Because of the delay between data acquisition
and elaboration, this speciﬁc case study qualiﬁes as a “near-real-time” monitoring setup,
according to the observations reported in the previous section. When the information to be
evaluated reaches the database, the calculation process could apply only a left-weighted
de-spiking process, using a dataset, which starts 5 points before (green lines in Figure 4). If
the data to be evaluated does not follow the previous trend, the algorithm could potentially
Sensors 2021,21, 2249 10 of 18
identify two scenarios: the occurrence of a spike (Figure 4, case a) or the beginning of a
new movement trend (Figure 4, case b). The software must perform a preliminary real-
time analysis that interprets one of the two mentioned cases, accordingly to the deviation
from the median. This information has a 50% reliability [
]. After 30 min a new data
transmission takes place (blue dashed line), adding 3 new information to the dataset.
The new elaboration takes advantage of these elements, and the algorithm identiﬁes the
occurrence of a spike (Figure 4, case a) or a displacement (Figure 4, case b). The net result
has higher conﬁdence (80%) than the previous one, keeping a small uncertainty due to the
incompleteness of the dataset (2 data are still missing). The following data transmission
(occurring 30 min later and represented by red dashed lines in Figure 4) completes the
dataset (orange lines in Figure 4) and conﬁrms the outcomes previously obtained with a
level of conﬁdence equal to 100%. Finally, the consequence of spike noise detection and
removal is the time delay related to the acquisition of the complete centered dataset (1 h
and 15 min for the example in Figure 4).
Raw data analysis, focusing on a spike event (
) and an actual displacement (
), was performed by using an
11-element data window centered on the continuous red line points, ranging between the green line and the orange line.
Data transmissions are represented by red, dashed blue and dashed red lines .
As stated before, multiparameter devices are one of the most relevant advantages
of an innovative monitoring system. Similarly, it is possible to include different sensors
measuring the same parameters in a single monitoring element in order to obtain a redun-
dant measure and improve data reliability. Following this idea, the MUMS system permits
the integration of two different tilt sensors at the same depth, giving redundancy to ﬁnal
outcomes. The double system (deﬁned Tilt Link HR 3D) exploits a 3D MEMS and a 2D
electrolytic cell (Figure 5), monitoring a wide range of differential displacements thanks
Sensors 2021,21, 2249 11 of 18
to the high-resolution that characterizes the electrolytic cell and the 360-degree range of
MEMS. In this conﬁguration, results are obtained using the tilt value provided by the
MEMS accelerometer, or the information of electro-level and the direction of movement
recorded by the magnetometer. The instrumental axes of the two sensors coincide.
Tilt Link HR 3D sensor, equipped with 3D microelectromechanical system (MEMS) and
2D electrolytic cell, placed on the same electronic board, with instrumental axes aligned on the
Moreover, the redundancy measures provided by this approach play a major role
in data analysis and validation processes [
]. Figure 6shows a case history where the
MUMS system was applied as an EWS to control a landslide threatening some houses.
Speciﬁcally, the slope monitoring device included an automatic inclinometer (Vertical
Array) integrating a total of 29 Tilt Link HR 3D sensors spaced 1 m one from each other
and a piezometer to measure the water table level. Additionally, two tiltmeters and two
crack meters were installed on the house walls to identify any effect induced by the slope
movements on the buildings. In particular, the tiltmeters integrated a MEMS sensor and
electrolytic tilt sensor, in a similar conﬁguration to the one used in Tilt Link HR 3D. On the
other hand, the two crack meters were analog devices featuring an mV/V output signal.
All these tools were connected to a single control unit powered through the connection
to the electrical line, while data were sent to the elaboration center via UMTS connection.
On 19 November 2019, MEMS sensors recorded a rapid displacement of 6 mm localized
at 13 m of depth that occurred over a time interval of 12 h. The comparison of local
differential displacements recorded by MEMS (Figure 6a) and electrolytic cells (
along maximum grade direction gave a comfortable result, allowing the complete validation
of the ongoing phenomenon. The evolution of displacement recorded at
over time highlights the improved accuracy and stability that distinguish electro-levels
) with respect to MEMS (Figure 6d) sensors, while the impulsive event was
effectively recorded by both devices.
Sensors 2021,21, 2249 12 of 18
Comparison between local differential displacements recorded by (
) MEMS and (
) electrolytic cells along
maximum grade direction and their evolution along time at a depth of 13 m (c,d), respectively).
As previously discussed, the management of a large number of multiparameter
data with proper sampling frequencies could bring to the identiﬁcation of signiﬁcant
cause-effects relationships, such as direct correlations between rainfalls, water levels and
displacements over speciﬁc time periods. Figure 7represents a case history in northern
Italy where a MUMS-based automatic inclinometer was installed in order to monitor an
active landslide that interacts with the construction site of a new road tunnel. Speciﬁcally,
the monitoring tool is composed of 20 Tilt Link HR 3D V sensors, installed between
of depth, and one piezometer. This sensor is designed to measure the
absolute pore water pressure over time. Thanks to the presence of a barometer on-site,
it is possible to remove the atmospheric component from the measure, thus obtaining
the relative pore pressure. Then, the conversion of this value with previously computed
calibration parameters allows for the evaluation of the water level variation over time. The
instrumentation is still active and has monitored displacements, water level variations, and
rainfall height, showing signiﬁcant and interesting cause–effect correlations. In particular,
the water level has been highly inﬂuenced by rainfalls, leading to a variation of 5 m in about
two weeks and a stabilization that occurred later, during a period of heavy precipitations.
Local differential displacements recorded on the maximum grade direction at a depth that
ranges between 7.5 and 6.5 m show a strong correlation with the water level trend. It is
also possible to observe the interaction of both water level and displacements, which tend
to increase during the ﬁrst signiﬁcant impulsive events, reaching a situation that does not
change even though the occurrence of continuous and relevant precipitations.
Sensors 2021,21, 2249 13 of 18
Comparison between rainfall height, water level variations and displacements recorded by
a MUMS-based automatic inclinometer on a landslide in northern Italy.
As mentioned in the previous paragraph, a trigger activation should lead to cross-
check controls that analyze outcomes of related sensors. An application of this idea is
displayed in Figure 8, showing a case history where a rockfall barrier is monitored through
a rockfall safety network (RSN) MUMS system. RSN is a monitoring network composed
of a main control unit radio connected with modules (deﬁned BPM), each one equipped
with a 3D MEMS, a 2D electrolytic cell and a load cell. Each single BPM module is
installed directly on a different steel post of the barrier in order to monitor its rotation
and inclination over time (thanks to MEMS and electro-level sensors) while the load cell
controls the mountain brace load. The datalogger is powered by a lead–acid battery and a
solar panel, while BPM modules exploit lithium-thionyl chloride batteries. A mechanical
trigger connected to the control unit completes the monitoring system. When the trigger
activates, the CU queries every device connected to the smart-mesh network and sends
data to the elaboration center thanks to a UMTS connection. There, the software correlates
the trigger to its related barrier, associating the BPM installed in correspondence with
the potential impacts. The algorithm analyses the monitoring outcomes with a two-step
procedure in order to check if the trigger activation corresponds to an actual event. In the
ﬁrst phase, the software deﬁnes a speciﬁc threshold for each sensor, established by a mean
value computed from a dataset of monitoring outcomes recorded during previous days.
If a predeﬁned percentage of sensors overcomes these values, the algorithm activates the
following step. At this stage, monitoring data are compared to reference parameters, as
service energy level (SEL) or maximum energy level (MEL), that are deﬁned by speciﬁc
tests performed on the barrier. If the elaboration returns a positive result, then an alarm
message is issued since the trigger is activated in correspondence with an exceptional event.
Sensors 2021,21, 2249 14 of 18
On the other hand, no alarm is disseminated if one of these conditions is not veriﬁed since
it means that the trigger sensors recorded a non-critical event. This procedure is intended
to reduce the probability of issuing a false alarm, which can be identiﬁed if no variation
is recorded by BPM modules at the trigger activation. An example of this occurrence is
reported in Figure 8, where it is possible to observe how data trends recorded by MEMS,
electro-level and load cell sensors do not show any variation despite several activations of
the trigger installed on the barrier (activations are marked as A1 to A5 in Figure 8). In all
these cases, the software automatically recognized the false alarm and did not disseminate
any alert message. Further conﬁrmation derived from on-site investigations, which veriﬁed
the absence of material on the barrier after the trigger activations.
Trigger activations and related steel post tilt values recorded by MEMS and electro-level
sensors, together with the mountain brace load identiﬁed by load cell sensor.
Another relevant advantage of the correlation between various physical entities is
to provide more information in order to avoid interpretation errors and identify the phe-
nomenon behavior. Figure 9reports tilt data recorded by a tiltmeter placed on a building
in order to identify possible landslide interactions with structural stability. The case study
is the same previously described in Figure 5. It is possible to observe a trend, which could
be interpreted as a slow and continuous tilt, related to a period of continuous rainfall.
However, thanks to the thermometer integrated into the sensor, it was possible to observe
a similar trend of temperature and tilt data. Therefore, thanks to the installation of two
sensors, it was possible to ﬁnd a relation between tilt variation and thermal effects, thus
providing a more reliable monitoring data interpretation.
Sensors 2021,21, 2249 15 of 18
Comparison between tilt data and temperature recorded by MEMS sensor placed in a
The enhancement of connection between different devices has been one of the most
inﬂuential technological improvements that emerged in recent years. Thanks to Internet
connectivity, it has been possible to improve the remote interaction of various technologies,
thus creating the Internet-of-things (IoT) concept. As a result of their wide applicability
and the several advantages deriving from its implementation, IoT principles have been
integrated into a signiﬁcant number of different processes and ﬁelds. In particular, the
development of new automated and innovative monitoring tools has brought to the appli-
cation of IoT to the geotechnical sector, leading to the design of advanced systems with the
improved connection between different components.
One of the systems deriving from this new approach is the deﬁned Internet of natural
hazards (IoNH) and allows the remote collection and management of information that
are fundamental for risk prevention purposes. As described in the present paper, the
IoNH concept focuses on integrating IoT-based technologies to permit a constant check
of instrumentation functionality, both in terms of sensor operativity and control unit
conditions. Another advantage regards the possibility to control a large number of critical
areas from a single control room thanks to completely automated procedures for data
acquisition and elaboration. Moreover, the high sampling frequencies achievable by devices
connected to the IoNH system give the possibility to perform statistical analyses to validate
numerical results. Other features are the establishment of reliable cause–effect relationships
related to multiparameter information, the activation of alert procedures with warning
messages to the authority in charge of the direct activation of trafﬁc lights and/or sirens.
This paper presented a series of examples acquired from different case studies in order to
underline the positive effects of the IoNH integration in a geotechnical monitoring system.
Sensors 2021,21, 2249 16 of 18
Finally, one of the most relevant topics related to automatic monitoring and early
warning systems concerns risk communication strategies and procedures. In fact, when
dealing with systems that can autonomously control the closure of roads, the preliminary
evacuation of landslide areas, etc., it is necessary to implement speciﬁc actions that should
be followed by the authority in charge and by the population. It is fundamental to keep
in mind that technology should not be intended as a substitution of human decisions
and evaluations, but it represents a valid tool for the proper and timely management of
potentially hazardous situations.
Conceptualization, A.C., A.V., E.C., R.S. and A.S.; Data curation, A.C., A.V.
and R.S.; Formal analysis, A.C., E.C. and R.S.; Funding acquisition, A.S.; Investigation, A.C., A.V.,
E.C. and R.S.; Methodology, A.C., A.V., E.C. and R.S.; Project administration, A.S.; Resources,
A.S.; Software, A.C. and A.V.; Supervision, A.C. and A.S.; Validation, A.C., A.V., E.C. and R.S.;
Visualization, A.C. and R.S.; Writing—original draft, A.C. and A.V.; Writing—review and editing,
A.V. and A.S. All authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement:
The data presented in this study are available on request from the
Conﬂicts of Interest: The authors declare no conﬂict of interest.
Dunnicliff, J.; Marr, W.A.; Standing, J. Principles of geotechnical monitoring. In ICE Manual of Geotechnical Engineering: Volume II;
Burland, J., Chapman, T., Skinner, H., Brown, M., Eds.; Thomas Telford Ltd.: London, UK, 2012; pp. 1363–1377. [CrossRef]
Zhang, Y.F.; Cai, M.F. Geotechnical Engineering Intelligent Monitoring and Controlling System and its Application in Pit
Engineering. Appl. Mech. Mater. 2011,105-107, 1561–1566. [CrossRef]
Valletta, A.; Carri, A.; Segalini, A. Innovative monitoring instruments as support tools for natural risks management. Rendiconti
Online della SocietàGeologica Italiana 2019,48, 76–83. [CrossRef]
Giorgetti, A.; Lucchi, M.; Tavelli, E.; Barla, M.; Gigli, G.; Casagli, N.; Chiani, M.; Dardari, D. A Robust Wireless Sensor Network
for Landslide Risk Analysis: System Design, Deployment, and Field Testing. IEEE Sensors J. 2016,16, 6374–6386. [CrossRef]
El Moulat, M.; Debauche, O.; Mahmoudi, S.; Brahim, L.A.; Manneback, P.; Lebeau, F. Monitoring System Using Internet of Things
For Potential Landslides. Procedia Comput. Sci. 2018,134, 26–34. [CrossRef]
Lo, H.-C.; Hsu, S.-M.; Chou, P.-Y.; Ke, C.-C. Mitigating drought and landslide simultaneously for mountain tribes of Taiwan:
Hydrogeological investigation, modelling, and development of an intelligent hazard prevention system. Nat. Hazards
Abraham, M.T.; Satyam, N.; Pradhan, B.; Alamri, A.M. IoT-Based Geotechnical Monitoring of Unstable Slopes for Landslide Early
Warning in the Darjeeling Himalayas. Sensors 2020,20, 2611. [CrossRef]
Bai, D.; Tang, J.; Lu, G.; Zhu, Z.; Liu, T.; Fang, J. The design and application of landslide monitoring and early warning system
based on microservice architecture. Geomat. Nat. Hazards Risk 2020,11, 928–948. [CrossRef]
Oguz, E.A.; Robinson, K.; Depina, I.; Thakur, V. IoT-Based Strategies for Risk Management of Rainfall-Induced Landslides: A
Review. In Proceedings of the 7th International Symposium on Geotechnical Safety and Risk (ISGSR 2019), Taipei, Taiwan, 11–13 December
2019; Research Publishing Services: Singapore, 2019; pp. 733–738.
Chaturvedi, P.; Thakur, K.K.; Mali, N.; Kala, V.U.; Kumar, S.; Yadav, S.; Dutt, V. A low-cost iot framework for landslide prediction
and risk communication. In Internet of Things A to Z; Hasan, Q., Ed.; John Wiley & Sons: Hoboken, NJ, USA, 2018; pp. 593–610.
Butler, M.; Angelopoulos, M.; Mahy, D. Efﬁcient IoT-enabled landslide monitoring. In Proceedings of the 2019 IEEE 5th World
Forum on Internet of Things (WF-IoT), Limerick, Ireland, 15–18 April 2019; pp. 171–176.
Terzis, A.; Anandarajah, A.; Moore, K.; Wang, I.-J. Slip surface localization in wireless sensor networks for landslide prediction.
In Proceedings of the 5th International Conference on Information Processing in Sensor Networks, Nashville, TN, USA, 19–29 April 2006;
Association for Computing Machinery: New York, NY, USA, 2006; pp. 109–116.
Aggarwal, S.; Mishra, P.K.; Sumakar, K.V.S.; Chaturvedi, P. Landslide Monitoring System Implementing IOT Using Video Camera.
In Proceedings of the 3rd International Conference for Convergence in Technology (I2CT), Prune, India, 6–8 April 2018; Institute of
Electrical and Electronics Engineers (IEEE): Piscawatay, NJ, USA, 2018; pp. 1–4.
Xie, Y.; Liu, J. Analysis of early-warning threshold for metro construction collapse risk based on D-S evidence theory and rough
set. Wuhan Univ. J. Nat. Sci. 2017,22, 510–516. [CrossRef]
Sensors 2021,21, 2249 17 of 18
Wang, M.; Xiao, J.; Rong, X. Remote Monitoring and Early Warning System for Subway Construction. Pavement Geotech. Eng.
Transp. 2013, 135–141. [CrossRef]
Ding, L.Y.; Zhou, C.; Deng, Q.X.; Luo, H.B.; Ye, X.W.; Ni, Y.Q.; Guo, P. Real-time safety early warning system for cross passage
construction in Yangtze Riverbed Metro Tunnel based on the internet of things. Autom. Constr. 2013,36, 25–37. [CrossRef]
Wang, L.; Xu, S.; Qiu, J.; Wang, K.; Ma, E.; Li, C.; Guo, C. Automatic Monitoring System in Underground Engineering Construction:
Review and Prospect. Adv. Civ. Eng. 2020,2020, 1–16. [CrossRef]
Liu, Q.; Lu, G.; Huang, J.; Bai, D. Development of tunnel intelligent monitoring and early warning system based on micro-service
architecture: The case of AnPing tunnel. Geomatics Nat. Hazards Risk 2020,11, 1404–1425. [CrossRef]
Segalini, A.; Carri, A.; Valletta, A.; Cavalca, E. Internet-of-Things principles applied to geotechnical monitoring activities: The
Internet of Natural Hazards (IoNH) approach. In Proceedings of the 3rd ICITG—International Conference on Information
Technology in Geo-Engineering, Guimaraes, Portugal, 29 September–02 October 2019.
Huang, J.; Huang, R.; Ju, N.; Xu, Q.; He, C. 3D WebGIS-Based Platform for Debris Flow Early Warning: A Case Study. Eng. Geol.
2015,197, 57–66. [CrossRef]
Reid, M.E.; Lahusen, R.G.; Baum, R.L.; Kean, J.W.; Schulz, W.H.; Highland, L.M. Real-Time Monitoring of Landslides. Report No.
2012–3008, Fact Sheet; USGS Publications: Reston, VA, USA, 2012. [CrossRef]
Segalini, A.; Chiapponi, L.; Pastarini, B.; Carini, C. Automated Inclinometer Monitoring Based on Micro Electro-Mechanical
System Technology: Applications and Veriﬁcation. In Proceedings of the Landslide Science for a Safer Geoenvironment; Sassa, K.,
Canuti, P., Yin, Y., Eds.; Springer: New York, NY, USA, 2014; pp. 595–600.
Reid, M.E.; Baum, R.L.; LaHusen, R.G.; Ellis, W.L. Capturing landslide dynamics and hydrologic triggers using near-real-time
monitoring. In Landslides and Engineering Slopes—From the past to the future. Proceedings of the 10th International Symposium on
Landslides and Engineered Slopes; Chen, Z., Zhang, J.-M., Ho, K., Wu, F.-Q., Li, Z.-K., Chen, Z., Zhang, J.-M., Ho, K., Wu, F.-Q., Li,
Z.-K., Eds.; Taylonr and Francis: Abingdon, UK, 2008.
Segalini, A.; Giovannelli, R.; Chiapponi, L.; Carri, A. Integrated Analysis of Data Collected by an Innovative Monitoring System
and Results of a 3D Numerical Model in Boschetto Landslide. In Landslides and Engineered Slopes. Experience, Theory and Practice:
Proceedings of the 12th International Symposium on Landslides, Napoli, Italy, 12–19 June 2016; Aversa, S., Cascini, L., Picarelli, L.,
Scavia, C., Eds.; CRC Press: Boca Raton, FL, USA, 2016; pp. 579–585.
Tinti, F.; Carri, A.; Kasmaee, S.; Valletta, A.; Segalini, A.; Bonduà, S.; Bortolotti, V. Ground temperature monitoring for a coaxial
geothermal heat exchangers ﬁeld: Practical aspects and main issues from the ﬁrst year of measurements. Min. Geol. Pet. Eng. Bull.
2018,33, 47–57. [CrossRef]
Lunardi, P.; Gatti, M. Tunnel monitoring system—A contribution for the preparation of guidelines. In Proceedings of the
ITA—AITES 2010 World Tunnel Congress—WTC “Tunnel vision towards 2020”, Vancouver, Canada, 14–20 May 2010.
Carri, A. Innovative Monitoring Instrumentations and Methods for Landslide Risk Management and Mitigation. PhD. Thesis,
Department of Engineering and Architecture, University of Parma, Parma, Italy, 2019. Available online: https://hdl.handle.net/
1889/3734 (accessed on 18 December 2017).
Zhang, H.; Wu, Y. Removing spike noise from track alignment irregularities measures using an iteration ﬁlter. In Proceedings of the
2017 5th International Conference on Computer-Aided Design Manufacturing Modeling and Simulation (CDMMS 2017); AIP Publishing:
Melville, NY, USA, 2017; Volume 1834, p. 30007.
Hoaglin, D.C.; Mosteller, F.; Tukey, J.W. Understanding Robust and Exploratory Data Analysis; John Wiley & Sons: Hoboken, NY,
USA, 2006; p. 404. ISBN 0-471-09777-2.
30. Glaisher, J. XXXII. On a class of deﬁnite integrals. Lond. Edinb. Dublin Philos. Mag. J. Sci. 1871,42, 294–302. [CrossRef]
Allasia, P.; Manconi, A.; Giordan, D.; Baldo, M.; Lollino, G. ADVICE: A New Approach for Near-Real-Time Monitoring of Surface
Displacements in Landslide Hazard Scenarios. Sensors 2013,13, 8285–8302. [CrossRef]
Storm, K. Introduction to construction statistics using Excel. In Industrial Process Plant Construction Estimating and Man-Hour
Analysis; Gulf Professional Publishing: Houston, TX, USA, 2019; pp. 1–21.
Frigerio, S.; Schenato, L.; Bossi, G.; Cavalli, M.; Mantovani, M.; Marcato, G.; Pasuto, A. A web-based platform for automatic and
continuous landslide monitoring: The Rotolon (Eastern Italian Alps) case study. Comput. Geosci. 2014,63, 96–105. [CrossRef]
Carlà, T.; Intrieri, E.; Di Traglia, F.; Nolesini, T.; Gigli, G.; Casagli, N. Guidelines on the use of inverse velocity method as a tool for
setting alarm thresholds and forecasting landslides and structure collapses. Landslides 2017,14, 517–534. [CrossRef]
Bicocchi, G.; D’Ambrosio, M.; Rossi, G.; Rosi, A.; Tacconi-Stefanelli, C.; Segoni, S.; Nocentini, M.; Vannocci, P.; Tofani, V.
Geotechnical in situ measures to improve landslides forecasting models: A case study in Tuscany (Central Italy). In Landslides and
Engineered Slopes. Experience, Theory and Practice, Proceedings of the 12th International Symposium on Landslides, Napoli, Italy, 12–19
June 2016; Aversa, S., Cascini, L., Picarelli, L., Scavia, C., Eds.; Taylor and Francis: Boca Raton, FL, USA; CRC Press: Abingdon,
London, UK; pp. 419–424.
Krkaˇc, M.; Arbanas, S.M.; Arbanas, Ž.; Bernat, S.; Špehar, K.; Watanabe, N.; Nagai, O.; Sassa, K.; Marui, H.; Furuya, G.; et al.
Review of monitoring parameters of the Kostanjek landslide (Zagreb, Croatia). In Landslide Science for a Safer Geoenvironment;
Sassa, K., Canuti, P., Yin, Y., Eds.; Springer: Cham, Switzerland, 2014; pp. 637–643.
Fukuzono, T. A new method for predicting the failure time of a slope. In Proceedings of the 4th International Conference and Field
Workshop on Landslides, Tokyo, Japan, 23–31 August 1985; Tokyo University Press: Tokyo, Japan, 1985; pp. 145–150.
38. Voight, B. A method for prediction of volcanic eruptions. Nat. Cell Biol. 1988,332, 125–130. [CrossRef]
Sensors 2021,21, 2249 18 of 18
Rose, N.; Hungr, O. Forecasting potential rock slope failure in open pit mines using the inverse-velocity method. Int. J. Rock Mech.
Min. Sci. 2007,44, 308–320. [CrossRef]
Zimmer, V.L.; Sitar, N. Detection and location of rock falls using seismic and infrasound sensors. Eng. Geol.
Hajovsky, R.; Pies, M.; Velicka, J. Monitoring the condition of the protective fence above the railway track. IFAC Pap. OnLine
52, 145–150. [CrossRef]
Yan, Y.; Li, T.; Liu, J.; Wang, W.; Su, Q. Monitoring and early warning method for a rockfall along railways based on vibration
signal characteristics. Sci. Rep. 2019,9, 1–10. [CrossRef] [PubMed]
Segalini, A.; Valletta, A.; Carri, A.; Cavalca, E. Monitoring of a retaining wall with innovative multi-parameter tools. In Proceedings
of the 4th Regional Symposium on Landslides in the Adriatic—Balkan Region, Sarajevo, Bosnia and Herzegovina, 23–25 October 2019;
Uljarevi´c, M., Zekan, S., Salkovi´c, S., Ibrahimovi´c, D., Eds.; Geotechnical Society of Bosnia and Herzegovina: Sarajevo, October,