Received 28 June 2024, accepted 5 July 2024, date of publication 12 July 2024, date of current version 23 July 2024.

Digital Object Identifier 10.1109/ACCESS.2024.3427386

IoT-Driven Microseismic Sensing System and

Monitoring Platform for Landslide
Detection
P. K. INDUKALA , U. G. GOSH, AND MANEESHA VINODINI RAMESH , (Senior Member, IEEE)
Center for Wireless Networks & Applications (WNA), Amrita Vishwa Vidyapeetham, Amritapuri, India
Corresponding author: P. K. Indukala ([email protected])
This work was supported in part by the Ministry of Earth Sciences (MoES), Government of India, and Sikkim State Disaster Management
Authority (SSDMA) under the project titled ‘‘Advancing Integrated Wireless Sensor Networks for Real-Time Monitoring and Detection of
Disasters;’’ and in part by Amrita Vishwa Vidyapeetham.

ABSTRACT Landslides pose a significant threat to human life and infrastructure, causing extensive
damage and fatalities. Effective monitoring and dissemination of early warnings of imminent landslides
are constrained by a lack of precise spatial and temporal information on landslide triggers and uncertainties
of the factors that lead to such events. This paper addresses these issues by presenting an Internet of Things
(IoT)- driven platform designed to capture microseismic vibrations in landslide-prone areas. The proposed
system aims to provide insights into the onsets of hazardous landslides, particularly those stimulated by heavy
rainfall and earthquakes. This treatise utilizes a microseismic smart sensing system with geophone sensors
(SM-s nodes) which continuously records and transmits real-time data on seismic activities associated with
potential landslides, enabling timely propagation of early warnings. The proffered system’s ability to acquire
and characterize microseismic signals was systematically validated through an integrated set of landslide
laboratory experiments, outdoor field trials, and real-world deployment in Chandmari located in the State of
Sikkim, India, situated in the North-Eastern Himalayan region. Furthermore, the paper provides an in-depth
analysis of the historical microseismic activities, differentiating them from ambient noises such as pedestrian
and vehicular movements and slope instabilities triggered by rainfall and earthquakes. The system’s
performance was evaluated during three real-world events: two earthquakes and an instance of rainfall
precipitation. This study explored the time and frequency characteristics as well as the variations of ground
motion parameters during recorded slope instabilities. A comparative analysis of existing microseismic
monitoring approaches was also conducted to assess the effectiveness of the proposed system. The insights
gained from this work were instrumental for the development of decision models capable of identifying
precursory microseismic activities precedent to imminent landslides, towards safeguard of lives and property
damage.

INDEX TERMS Early warning systems, geophones, IoT platform, landslides, microseismic signals, system
design, signal processing, wireless sensor networks.

I. INTRODUCTION staggering impact of landslides, with over 72,000 fatalities

Landslides pose significant threats to both human lives and and USD 11.5 billion in damages, globally, recorded between
infrastructure, as they involve the unpredictable disruptive 1900 and 2022 [2], [3]. Almost 75 % of these deadly incidents
movements of soil, rock, and other materials under the transpired in Asian nations, with a notable concentration in
force of gravity [1]. Recent research has pinpointed the the Himalayan regions of northern India [4].
Landslide events are triggered by various factors, such
The associate editor coordinating the review of this manuscript and as heavy rainfall, earthquakes, and anthropogenic activities,
approving it for publication was Hosam El-Ocla . severally or in tandem [5], [6]. Intense rainfall saturates
2024 The Authors. This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License.
For more information, see https://creativecommons.org/licenses/by-nc-nd/4.0/
VOLUME 12, 2024 97787
 P. K. Indukala et al.: IoT- Driven Microseismic Sensing System and Monitoring Platform

soil and destabilizes natural slopes whereas seismic events triggers. These challenges highlight the critical need for
weaken soil shear strength that culminate in frequent non-invasive techniques to understand subsurface changes of
instances of rapid landslides. Human endeavors such as the underlying landslide phenomena effectively. Microseis-
mining, road and dam construction can disrupt landscapes or mic monitoring, prompted by geophysical sensors, can detect
weaken soil structures that precipitate landslides. Recurrent faint earth tremors effectuating broad spatial coverage and
stick-slip motion along ruptured surfaces of steep slopes, non-invasive sensing of deep subsurface changes, precluding
breakage of soil pipes, strains and displacements due to the need for drilling, thereby minimizing environmental
progressive failure, sudden changes in load distribution along disturbances [15].
the shear surface due to brittle soil, and seismic forces furnish Microseismic signals precedent to landslide events afford
additional pointers to impending landslides [7], [8]. pragmatic precursors apropos slope instabilities or failures,
The Himalayan regions, typified by recent tectonic land- onset by water infiltration, seasonal variations in pore
slides, linked to rock mass fracturing and uncontrolled road pressure within the soil mass, expansion of micro-cracks, and
cuttings are characterized by intricate geological composi- seismic activities [23], [24], [25], [26], [27], [28], [29], [30].
tions, steep gradients, and high topographical ruggedness. Identification of pre-trigger signals is inevitable as landslides
These factors, amplified by seismic activity and heavy can occur abruptly. However, the microseismic signals,
deluge significantly increase the susceptibility to coseismic symptomatic of impending landslides, are relatively weak and
landslides [9], [10], [11]. Given these risks, implementation vulnerable to noise interferences. Noise interjections creep in
of early warning systems is crucial for the propagation of from divergent sources- surface or subsurface movements of
timely alerts, propitious to minimized property damage and cracks, terrestrial faults, water infiltration, and anthropogenic
enhanced communal welfare. excursions. These disturbances obscure targeted seismic
Most early warning systems for rainfall-induced land- signals impeding effective characterization of seismic events
slides rely on precipitation thresholds, with limited studies associated with landslides, and the site-specific noises [21],
investigating the microseismic precursory signals responsible [22].
for landslides, especially coseismic landslides [9], [12]. Microseismic sensors integrated with Internet of Things
In-situ landslide monitoring entails direct measurement of (IoT) technology augment the monitoring of landslides
soil moisture, pore water pressure, and ground deformation by enabling real-time data collection, which bolster pre-
using geophysical instruments. Such setups facilitate pre- dictive capabilities of monitoring systems, propitious to
cise, localized data and real-time monitoring or geological enhanced safety and resilience of communities in landslide-
transitions [13], [14], [15]. However, the preceding options prone regions. All the same, development of real-time IoT
are pricey, afford limited spatial coverage, and rely on landslide monitoring systems are riddled with challenges.
communication technologies with lower data rates, which are Conventional IoT gadgets are constrained in their ability
ill-suited for detailed microseismic monitoring [1], [16], [17]. to acquire, process, and transmit sensor data of higher
Affordable low-power sensors offering spatial information resolution. These limitations spawn increased development
with high resolution and minimal latency, are the preferred costs, higher energy consumption, and a dependance on
attributes of an ideal intelligent Landslide Early Warning centralized data centers [14]. Therefore, there is an exigent
system [1], [18]. need for the development of cost-effective, non-invasive real-
Traditional geodetic surveying methods are inefficient and time monitoring solutions that acquire multiple microseismic
require skilled workers whereas remote sensing techniques triggers of landslides, to effectuate early warning capability.
lack real-time resolution, obligatory for automated data Such systems in hazardous Himalayan terrains that enable
analysis [19]. Geophysical monitoring methods furnish the understanding of subsurface changes over large areas is
valuable information on subsurface slope activities and pre- undubitable. The key requirements for such a system are
cursory failure conditions [15], [20], [21], [22]. Standardized detailed below.
automated techniques for large-scale spatial analysis using Microseismic events are typically within the range of
these approaches is challenging in fragile Himalayan terrains. natural seismic activity falling below 100 Hertz [31], unlike
Nevertheless, these methods are expensive and mandate human/vehicle movements that produce vibrations in higher
specialized equipment and domain expertise, exacerbating frequency ranges based on the activity, and environment
implementation in disaster-prone areas [20]. characteristics [20], [32], [33], [34]. There is a need
The Himalayan region is characterized by shearing rocks to capture the microseismic activity beneath the earth’s
and highly jointed geologic formations, undergoing multiple surface, where landslides and earthquakes originate, through
phases of deformation, continuing to which move at rates precursory signals. The seismic signals associated with
ranging from a few millimeters to several centimeters per landslides often fall under the (0-10) Hz frequency range [20],
year [9]. Additionally, drilling activities in these regions [35], [36]. For accurate capture of microseismic signals,
could aggravate the vulnerability to landslides and other a data collection system should follow the Nyquist-Shannon
mass earthly movements. Deep drilling methods are par- sampling theorem, which calls for a sampling rate of at least
ticularly unsuitable due to the risks of borehole col- twice the maximum frequency of the monitored signal [37].
lapses, destabilization of hill slopes, and potential landslide However, a much higher sampling rate, typically ten times

97788 VOLUME 12, 2024

P. K. Indukala et al.: IoT- Driven Microseismic Sensing System and Monitoring Platform

the maximum signal frequency, is recommended to capture II. MICROSEISMIC SIGNAL MONITORING PLATFORM
subtle microseismic activities. Contemporary ealy warning This section discusses the proposed IoT-driven microseis-
systems are often designed for real-time lower data rates mic signal monitoring platform, suitable for large-scale
processing and transmission [20], [38]. Therefore, a system spatiotemporal monitoring, real-time data acquisition, and
is needed that can acquire microseismic signals in real-time, classification of microseismic signals to identify imminent
at high sampling rates from multiple triggers, sustain harsh landslides. The key components of the proposed monitoring
environments, with an extended service lifetime. platform are detailed herein.
Based on these requirements, the key research problem
can be stated as- how to effectively capture early triggers of A. SENSORS AT EDGE
landslides in real-time in resource-constrained regions, using The proposed system was conceived to provide a low-cost
microseismic monitoring approaches to save lives and abate solution for effective capture of microseismic activity beneath
property damage. the earth’s surface, which emanate landslide pre-triggers.
Major contributions of this investigation are summarized Extant sensing instruments listed in Table 1 are capable of
as follows: detecting subtle microseismic signals. The table acquaints
• Introduction of an IoT-driven platform with tailor-made parameters such as bandwidth capability, number of axes,
components for acquisition and seamless data transmis- weight, power requirements, and cost of these seismic
sion of subterranean microseismic vibrations in the face instruments. Notably, the listed seismometers are active,
of heavy rainfall and earthquakes, across multiple edges heavy, and expensive, while the moderately priced active or
and cloud. passive accelerometers, are capable of only detecting surface
• Development of a Smart Microseismic Sensing system, movements. Geophones stand out as the optimal choice due
incorporating geophone sensors, referred to as SM-s to their resilience to harsh environments and their ability to
nodes, for incessant capture, record and transmission of detect ultra-low frequency bands and low-magnitude vibra-
real-time data from landslide-prone areas. tions [39]. Low-cost, passive, and lightweight geophones
• Empirical assessment, validation and verification, via were selected for this study, for their ease of deployment
laboratory set-ups and field deployment of the opera- and maintenance. To further enhance their functionality,
tional capabilities of the proffered smart microseismic geophones can be outfitted with IoT capabilities, enabling
sensing nodes. real-time data transmission to the cloud, augmenting their
• Implementation of the IoT-driven microseismic smart effectiveness in capturing and relaying valuable seismic
sensing system in the Himalayan region characterized by information.
active tectonics and frequent heavy rainfall, where both
TABLE 1. List of seismic sensors.
landslides and earthquakes are common occurrences,
facilitating the continuous real-time transmission of the
earth’s vibrational data.
• Comprehensive analysis of historical microseismic data,
from multiple SM-s nodes deployed at the Chandmari
site, focused on slope instabilities actuated by precipita-
tion and seismic tremors.
This monitoring system was deployed at Chandmari, in the
East Sikkim district of the State of Sikkim, within the
North-eastern Himalayan region [subsection III-D].

A. PAPER OUTLINE
The rest of this paper is organized as follows: Section II,
delves into architectural details of the microseismic signal
monitoring platform for landslide detection and early warning
systems. Section III surmises the methods used for vali- The proposed smart microseismic sensing system
dation of the proposed microseismic sensing system, via employed geophone sensors to continuously capture, record,
laboratory setups, implementation in real-world operational and transmit real-time vibrational data on microseismic
scenarios and data processing methods for microseismic events occurring in landslide-prone regions, as shown in
signal characterization. Section IV describes the results Figure 1. A geophone, functioning as a ground motion
and discussions on the analysis of microseismic signals transducer, generates an analog differential voltage output
acquired from the proffered system. Section V provides a across its terminals, which is directly proportional to the
comparison of the proposed system with existing systems velocity of ground vibrations induced by the surrounding
pertinent to microseismic landslide monitoring. Conclusions medium. These compact, self-powered sensors offer a flat
and directions of future research are covered in section VI. frequency response in velocity, within a specific frequency

VOLUME 12, 2024 97789

 P. K. Indukala et al.: IoT- Driven Microseismic Sensing System and Monitoring Platform

FIGURE 1. Microseismic signal monitoring architectural framework enabling the capture and transmission of vibrational data from Smart microseismic
IoT Devices across heterogeneous data networks, facilitating detection and monitoring of microseismic activity in landslide-prone regions.

range above their resonant frequencies and attenuate sensors as shown in Figure 1. These sensors operate in
frequencies below them, rendering them well-suited for the synergy with supporting components like microcontrollers,
measurement of ground velocity. signal conditioning boards, communication modules, power
For recording the earth’s ground vibrations, multiple supply units, batteries, solar panels, among other peripher-
tri-axial geophones equipped with three terminals were als. This system enables local data processing and initial
utilized: Horizontal-1 (H1), Horizontal-2 (H2), and Vertical decision-making, instead of centralized processing to ensure
(V3) axes, corresponding to waveform propagation in the X swift responses. The strategically deployed system collects,
(North-South), Y (East-West), and Z (Up-Down) directions, stores, transmits, and analyzes ground vibrations related to
respectively [14], [22]. Geophones with resonant frequencies microseismic activities, in landslide prone areas.
of 10 Hz and 30 Hz were strategically placed at locations
mentioned in subsection III-D, to examine signal patterns,
associated spectral range of gradual subsurface movements, 1) DATA ACQUISITION HARDWARE PLATFORM
and rapid fractural/ slip surface movements within the A custom data acquisition hardware was designed for the
landslide mass, taking into account the trade-offs discussed precise acquisition of signals from the geophone sensors,
in section I. facilitating accurate measurement of microseismic activities,
Standard tipping bucket gauges within a rain-gauge system as represented in Figure 2. This hardware was constituted
for the measurement of rainfall intensity were also employed, of the following components: a pre-conditioning filter unit,
in tandem with the geophone sensors. This strategic integra- an amplification and differential-to-single-ended conversion
tion enabled detailed comparison of rainfall variations and a unit, and a power supply unit for essential support. The
study of the pretrigger patterns recorded by the geophones, terminals originating from the geophone sensors were
in landslide-prone regions, during periods of intense rainfall configured differentially. Paired outputs from each geophone
and earthquake tremors. terminal were initially directed into a pre-conditioning unit,
comprised of preamplifiers and filtering systems, as shown in
Figure 2.
B. SMART MICROSEISMIC MONITORING EDGE (SM-S This stage serves to amplify the delicate seismic vibra-
NODES) tions detected by the geophones and effectively eliminate
The Smart Microseismic Monitoring Edge forms the heart Radio Frequency (RF) and other high-frequency noise
of the architecture, featuring highly specialized microseismic sources. Noise reduction was accomplished through the

97790 VOLUME 12, 2024

P. K. Indukala et al.: IoT- Driven Microseismic Sensing System and Monitoring Platform

entire system transforms into a cost-effective IoT-enabled

configuration known as a Smart Microseismic Sensing
and Monitoring edge or SM-s nodes within the proposed
sensing system. The proposed design adopted solar-powered
renewable energy sources via solar panels, to conserve
energy. The power system was comprised of renewable
energy sources such as solar panel, which is the main source,
and the local power grid (AC socket power), both of which
were interfaced to a battery. The charge controller unit
ensures that power was properly fed to energize lead acid
FIGURE 2. Geophone data acquisition process.
batteries. Context aware energy management techniques, for
effective power consumption in edge sensing systems were
incorporated as outlined in [40], which extends the system
lifetime.
Sensitivity of the proffered microseismic data acquisition
system using a N-bit ADC (here,N=12) with a full-scale range
can be calculated as shown in eqn.1:
VADC VADC
SADC = N ≈ 12 ≈ 0.000805 V/bit. (1)
(2 − 1) (2 − 1)
where VADC is the output ADC value. Given that the
sensitivity of the geophone is SSensor , the overall sensitivity
of the system is defined in eqn.2:
SADC 0.000805 V/bit
SSystem = ≈ ≈ 2.8 × 10−5 m/s/bit.
SSensor 28.8 V/(m/s)
FIGURE 3. Microseismic sensing node for geophones. (2)

2) FIRMWARE DEVELOPMENT
implementation of a first-order Resistor-Capacitor- Low Pass The geophone data acquisition firmware was developed
Filter (RC-LPF) with an anti-aliasing cutoff at 156 Hz. within the open-source Arduino Integrated Development
The first-order passive Low Pass Filter (LPF) was chosen Environment (version 1.6.13), for in-depth exploration of
for its cost-effectiveness and durability in challenging subsurface movements within Sikkim’s challenging terrain,
environments. This decision was in tune with the objective and capture of microseismic signals. The effective capture of
of developing an affordable sustainable solution for the microseismic signals, mandate employment of high sampling
continuous acquisition and monitoring of microseismic rates, at least ten times the maximum signal frequency,
signals. Increasing the filter order would impose greater as discussed in section I. Lower data rates pertain to frequency
circuit complexity that potentially calls for expensive active of acquiring the data at longer intervals, such as one sample
filters, validating the choice of a passive first-order LPF. per minute or every five minutes, in lieu of incessant
Following the geophone signal’s pre-conditioning phase, acquisition. This can curb the timeliness and resolution of the
the filtered signals are routed into instrumentation amplifiers, microseismic data, potentially reducing the effectiveness of
outfitted with variable gain adjustment options. The precision monitoring and early warning systems.
of this gain was finely tuned to the desired level via Therefore, acquisition of geophone signals at higher
a network of resistors, controlled by a Printed Circuit data rate of 1 kHz (across three axis or channels) from
Board (PCB)-mountable switch. Subsequent to the fine- multiple locations, mentioned in section II-B3 was chosen
tuning stage, single-ended outputs from the three geophone for comprehension of the intricate subsurface changes
terminals were seamlessly directed toward analog-to-digital beneath the earth’s surface and the dynamics of landslides.
conversion (ADC) units. The dig of the three outputs were The proposed data acquisition system for geophones was
formatted to facilitate storage within Secure Digital (SD) engineered to record signals at a rate of 1000 samples
cards and subsequent transmission stages. It is imperative per second at each channels, and empowered to locally
to comprehend that this data acquisition hardware is the store microseismic data as a means of data back-ups.
key for the accurate capture of microseismic vibrations from This sensing concept operated within a holistic framework,
geophones, ensuing in a reliable and robust measurement constituted of data acquisition, signal amplification, signal
system. preprocessing for high-frequency noise reduction, and data
Programming modules in Figure 3 are energy-efficient compression modules, all seamlessly integrated into an IoT
systems-on-a-chip solutions equipped with built-in Wi-Fi environment. This multifaceted functionality empowered the
capabilities and ADC input channels of 12-bit resolution. The sensing system as a highly efficient and versatile edge node.

VOLUME 12, 2024 97791

 P. K. Indukala et al.: IoT- Driven Microseismic Sensing System and Monitoring Platform

3) DATA GENERATION At each location of the SM-s nodes, there are strategically
The proposed sensing system generated geophone data positioned Wi-Fi routers, operating as Access Points in
that represented ground vibrations evolving from diverse station mode, termed as Long-Range Wireless Wi-Fi stations
geological processes- rock fracturing, tectonic movements, (LRW-S), along with heterogenous wireless networks [41].
or slope instabilities, along with the rainfall rates measured Data from SM-s nodes are seamlessly routed through LRW-S
by the supportive rain gauges. These parameters collectively to the FMC via LRW-BS, as shown in Figure 1. This process
offer insights into environmental conditions and seismic enables the collection and aggregation of data from SM-s
activities in landslide-prone areas. In addition, metadata nodes into a local storage/database, where raw sensor data is
is generated to provide supplementary information for also stored. FMC facilitates local processing, data analytics,
each parameter. The metadata is comprised of timestamps, and initial decision-making, bringing intelligence to the edge
location coordinates, sensor information, and other pertinent of the network.
parameters, that enrich the utility of the collected data. The intermediate processors at the FMC aggregate data
Every sensing node produces significant amounts of data. from SM-s nodes, enabling regional-scale data analysis and
To store the geophone data every second, from the three a first level of alert dissemination. FMC also facilitates
channels on an SD card, a minimum storage capacity determination of site-specific thresholds, and event detection
of approximately 14.4 gigabytes per hour with a data- to boost efficiency in signal analysis. This strategic step
rate of 4 Mbps is required. Such sizable storage demand enables the regional-scale analysis of data and the dissem-
compel real-time data transmission. A minimum data-rate ination of initial alerts. The cloud-based edge intelligence
of 36 Kbps, is attained by sampling each channel at 1 kHz algorithms are seamlessly deployed to the processors within
with a minimum resolution of 12 bits. the FMC. These algorithms are continuously refined and
updated, driven by valuable feedback collected from the field.
C. DATA COMMUNICATION
Given the study region’s rugged terrain, discussed in III-D1, D. SMART MICROSEISMIC ANALYTICS CLOUD
an effective communication system is imperative, con- 1) EDGE-CLOUD NETWORK, EDGE DEVICE CONTROL AND
sidering factors such as bandwidth, data rate, frequency, DATA MANAGEMENT
and propagation factors, specific to the deployment site. The Smart microseismic Analytics Cloud is comprised
Traditional cellular network standards, such as Global of a centralized hub or cloud infrastructure (server) via
System for Mobile Communications (GSM), 3G (Third Internet Gateways for long-term storage (data repository),
Generation), and 4G (Fourth Generation) stipulates limited archiving, and systematic data management. Routers and
bandwidth and lower data rates, rendering them incapable gateways, operating at edge, and cloud, facilitate end-to-end
for real-time transmission of extensive geophone-generated communication among SM-s nodes, FMCs, and cloud data
data. As 5G (Fifth Generation) technology may not be centers, ensuring dependable data transfer. The edge-cloud
universally accessible, alternative Low Power Wide Area network and edge device control are managed at the cloud
Network (LPWAN) technologies like Long Range (LoRa) level by monitoring the status of connected SM-s nodes and
and Narrowband Internet of Things (NB-IoT) were developed the connectivity to the FMC and cloud server. Over-the-
for lower data rates that are ill-suited for this purpose. air (OTA) updates keep the SM-s nodes secure and up to
Given the specific conditions of the deployment area, date with firmware changes (bug fixes and enhancements).
including dense vegetation and dynamic climatic extremes, Cloud-based device management facilitates the distribution
Wi-Fi stands out as the optimal choice for real-time of OTA updates, ensuring that SM-s nodes receive patches
transmission of geophone data. Long-range Wi-Fi systems and improvements obviating manual intervention.
extend connectivity over vast distances, enabling point-to- A data synchronization system in [42] combines data from
point or point-to-multipoint connections across ranges of various landslide deployments into a unified central cloud
hundred meters to several kilometers [41]. server. Communication between the sites and the server is
Accordingly, Wi-Fi and long-range Wi-Fi technologies established through sockets, allowing the server to receive
were adopted for intermediate data transmission along streaming data from the edge devices and store it in a
with connectivity protocols such as Transmission Control centralized database, Aggregated Landslide Data [42]. The
Protocol/Internet Protocol (TCP/IP), and Hypertext Trans- FMC-Site client system excels at managing reconnections
fer Protocol (HTTP)/ Hypertext Transfer Protocol Secure without compromising performance and functionality. The
(HTTPS) to facilitate data transmission with efficiency centralized repository enables cross-site analysis, enhancing
and scalability. These technologies perfectly match the predictive capabilities for landslide monitoring across multi-
requirements by extending network coverage to challenging ple locations [42].
environments, furnishing internet access, and establishing
connectivity between remote outlying locations. 2) CLOUD DATA ANALYTICS: GETTING INSIGHTS FROM
Functioning as a Base Station, the Field Management MICROSEISMIC SIGNALS
Center (FMC), leverages long-range Wi-Fi technology, The Smart microseismic Monitoring Cloud ensures data
is denoted as Long Range Wi-Fi Base Station (LRW-BS). durability and seamless accessibility, facilitating its utilization

97792 VOLUME 12, 2024

P. K. Indukala et al.: IoT- Driven Microseismic Sensing System and Monitoring Platform

by the adoption of advanced machine learning /artificial

intelligence algorithms and data mining techniques [43],
[44], [45], [46], [47], [48]. The core objective is to glean
valuable insights from the microseismic data originating from
diverse locations, for systematic analysis of the landslide
signals. Multi-tier signal processing framework, discussed
in III-E allows for selective noise removal from various
frequency bands, while retaining signal features, fine-tuned
to site-specific thresholds for event detection and signal
analysis.
The framework is seamlessly integrated into the FMC
processors at the edge, empowers the FMC to issue timely
alerts and warnings, strengthening its proactive response
capabilities. Such an approach significantly reduces the
burden on central data centers and cloud resources. This FIGURE 4. Set up of landslide testbeds.

translates to cost-effective and energy-efficient operations,

rendering the framework sustainable for long-term monitor-
ing efforts. conditions, variability in rainfall rates, and dynamic pore
pressures [12].
3) CLOUD SERVICES The laboratory houses a medium-scale test bed of (1400
Centralized monitoring at the cloud is facilitated by a mm x 500 mm x 400 mm) dimensions capable of holding
dedicated monitoring center which serves as the nexus for 2 tons of soil and a large-scale test bed of (1400 mm x
real-time streaming of microseismic data originated from 500 mm x 908765 mm) accommodating up to 8 tons of
diverse locations under surveillance. Its primary mission soil, facilitating realistic landslide simulations. Both the test
is to ensure uninterrupted monitoring, swift responses to beds incorporate a seepage simulator designed to replicate
microseismic activities detected at the deployment sites, underground water movement with precise control over
enabled by the utilization of advanced visualization tools. seepage rates and water pressures. The simulator can handle
The monitoring center further reinforces the decision multiple seepages, at varying levels to simulate diverse
support systems by promptly conveying alerts, automated flows across soil layers, alongside flow meters and pressure
report generation, and dispatch of notifications to researchers. gauges integrated into the rainfall and seepage simulators for
In cases where genuine alerts are identified, immediate accurate measurement of flow rates.
communication with disaster management authorities ensues, The test beds are equipped with adjustable slope angles
facilitating prompt necessary actions and interventions. This ranging from (zero to 45) degrees, enabling the simulation of
coordinated approach, substantially enhances the system’s varied geomorphic scenarios, supported by angle gauges for
efficacy in vigilant monitoring of microseismic events in precise measurements. Additionally, the setups include rain-
areas susceptible to landslides. fall simulators with a controlling valve to replicate varying
The data centers continuously monitor seismic activities intensities of rainfall. Specialized sprinklers attached (1)-(2)
in real-time, over extended periods, identifying any unusual meters above the soil surface ensure even water distribution,
or anomalous patterns that call for immediate attention. accurately mimicking natural rainfall patterns.
Leveraging the insights derived from regional-level analysis, Furthermore, the laboratory setup features a soil com-
can lead to improved assessment of future seismic events, pactor for achieving different degrees of soil compaction,
based on historical data and ongoing monitoring efforts. complemented by a compaction meter for quantification of
compaction levels. The setups allow to simulate multiple
III. METHODS depths of soil layers, facilitating homogeneous and stratified
This section discusses the experiments conducted in landslide soil configurations. Testbeds can also incorporate vegetation
laboratory test setups, outdoor deployments to validate the effects for a more realistic simulation. The laboratory setup
functionalities of the proposed microseismic sensing system serves as a platform for calibration of geological sensors prior
in landslide-prone areas, and the data processing methods to their field-deployment in real-world scenarios.
used for the characterization of the acquired microseismic
signals. B. EXPERIMENTATION IN LANDSLIDE LABORATORY
The experiments represent a critical validation phase of the
A. LANDSLIDE LABORATORY TEST SETUPS proposed data acquisition system, apropos landslide early
A laboratory setup, depicted in Figure 4, integrates advanced warning systems. The test plan was drawn to provide insights
features for the scientific simulation of landslides. These into responses from geophone sensors under laboratory-
features encompass variable slope angles, the ability to controlled conditions. This assessment showcases the prof-
simulate diverse soil layer compositions reflecting field fered system’s versatility in capturing simulated landslide

VOLUME 12, 2024 97793

 P. K. Indukala et al.: IoT- Driven Microseismic Sensing System and Monitoring Platform

FIGURE 6. Geophone sensor placement at the deployment site at

chandmari, sikkim.

with different activities (jumping, walking, running, etc.) and

capturing the movement of lightweight vehicles at a constant
speed of around (20-25) km/hr.

D. REAL WORLD DEPLOYMENT

The real-time monitoring of microseismic signals, in landslide-
prone mountainous regions, was achieved by deploying a
network of the proposed SM-s nodes. This approach provided
a comprehensive understanding of regional variability and
facilitated better decision-making, during various triggering
factors in landslide-prone regions.
FIGURE 5. Experimentation for data collection.

1) STUDY REGION
events, attesting its readiness for general field deployment for The state of Sikkim, located in the northeastern part of India,
early warning for landslides. encounters heightened risks within this geologically active
For conducting experiments, the medium-scale test bed and fragile Himalayan mountain range, rendering it fertile
was utilized, filled with a four-layer soil structure. The ground for the study of landslide dynamics and development
latter included clay, a mixture of gravel and boulders, of effective monitoring systems [9], [10], [11]. The region’s
a combination of sand and clay, and a mix of sand and geological complexity, exemplified by major thrust faults-
rocks. The slope angle was precisely set to 35 degrees Main Boundary Thrust (MBT) and the Main Central Thrust
and maintained consistently throughout the experiment, (MCT) and highly jointed rocks undergoing continuous
mirroring a natural hill slope within the laboratory setting. deformation, presents arduous drilling activities, with exac-
Multiple geophones were installed onto the testbed, and erbated vulnerability to landslides and mass movements [23].
the proffered sensing system was used to acquire geophone Chandmari, a locality in Gangtok, East Sikkim District, was
signals at 1 kHz, for each channel. The rainfall was simulated identified as the study area due to its tracked history of
for a continuous duration of eight hours at a precipitation rate numerous landslides during the monsoon season [49]. It is
of 45 mm/hour. The rainfall contributed to saturation of the important to note that the region is classified as a high-risk
soil layers in the testbed. Incessant water infiltration increases seismic zone IV on the Indian seismic zoning map. The region
the pore pressure, eventually leading to slope instabilities, encompassing the deployment site experienced moderate
as shown in Figure 5). earthquakes, involving thrust motions and strike-slip motions,
at the fault lines [10], [11], [14].
C. OUTDOOR FIELD TRIALS
Outdoor experiments were conducted in a road serving vehi- 2) SENSOR PLACEMENT
cle and pedestrian traffic, to simulate real-world conditions, This study utilized Deep Earth Probes (DEPs), an integrated
as shown in Figure 5b). The signals associated with traffic monitoring module constituted of several heterogeneous
movements were acquired by the proffered sensing system, sensors as part of the real-time Landslide Early Warning
to capture and understand the associate noise patterns. This system [1]. Drilling for the placement of sensors is arduous
included ten trials of single and multiple pedestrians engaged due to the loose soil and fragile rocks in the site area. For

97794 VOLUME 12, 2024

P. K. Indukala et al.: IoT- Driven Microseismic Sensing System and Monitoring Platform

FIGURE 7. Chandmari Geophone DEP Locations ( black colored dots), and

geophone alignment at each location. [49].

the placement of sensors in the ground, small square wells

were dug manually, to minimize inadvertent disturbance to
the earth. Figure 6 shows the placement of geophones at
one of the test site locations. The geophone spikes were FIGURE 8. A DEP location where one of the SM-s Node is deployed at
fixed firmly, carefully oriented parallel to the ground, using Chandmari, Sikkim.
built-in level indicators. Furthermore, the entire setup was
made waterproof to protect geophones from rainfall or
other precipitation. The sensor configuration was seamlessly
area, detailed in section III-D1, unveiled potential sources
integrated with the smart microseismic sensing system,
of signal interference- nearby water streams, shrubs and
transforming into a fully functional IoT edge node.
trees, roads, pedestrian pathways, wildlife, and construction
activities. High-voltage transmission power lines nearby can
3) FIELD DEPLOYMENT OF SMART MICROSEISMIC SENSING introduce additional noise into the geophone signals. Given
SYSTEMS the diverse noise sources, comprehensive characterization
Prototypes of the microseismic edge-based geophone sensing of the microseismic signals was called for, to differentiate
systems were strategically deployed at different positions, them from the surrounding noise sources at the deployment
termed as DEP locations, in the hilly terrain, such as the toe, site. Signal processing techniques enabled the identification
middle, and crown regions. The fully operational SM-s nodes, of noise patterns, their elimination, and the inference of
shown as black colored dots in Figure 7, provide crucial relevant noise-minimized microseismic information from the
microseismic data for early warning systems and advancing geophone data [50].
research on landslides and seismic activity in the region. Its A multi-tier signal processing frame work was utilized for
uninterrupted functionality ensures a reliable and valuable characterizing the geophone signals. Firstly, the streaming
microseismic sensing system for large-scale spatiotemporal values of voltage from geophones were extracted from the
monitoring in landslide-prone regions. three channels, X, Y, and Z axes. The three-channel data
A sample DEP location is shown in Figure 8. Since the time formed the basis for subsequent processing. To eliminate any
of its pilot deployment in 2018, the state-of-the-art geophone- DC offset or baseline drift, mean normalization was applied.
based microseismic sensing systems have been effectively This step ensured that the data was centered around zero,
capturing and transmitting data to the FMC located in the which is essential for accurate signal analysis. A digital But-
landslide-prone region of Chandmari. terworth LPF was utilized to improve signal quality. The filter
parameters, and the filter order, were optimized for removal
E. DATA PROCESSING METHODS of high-frequency noise, to preserve the essential components
The SM-s nodes play a crucial role in capturing ground move- of the acquired signal. The preprocessed geophone signals
ments stimulated by various sources, encompassing surface were also subjected to a series of bandpass filters. Each
and subsurface activities. However, these vibrational signals filter had a specific cutoff frequency range, such as (0-5) Hz,
from geophones are susceptible to significant influences from (5-10) Hz, (10-20) Hz, and so on. This step was undertaken to
environmental factors, fluctuations in temperature, power line comprehend the frequency characteristics of the data within
interferences, and sensor drift. Examination of the study a particular range, which helps identification of dominant

VOLUME 12, 2024 97795

 P. K. Indukala et al.: IoT- Driven Microseismic Sensing System and Monitoring Platform

frequencies, frequency-spreads, and noise patterns that may

be site-specific.
The geophone signals were put through a Short-Time
Fourier Transform (STFT) analysis, to discern the frequency
range of the acquired signal, providing valuable insights into
the characteristics and dynamics of the microseismic activ-
ities [50], [51], [52]. Withal, the changes in ground motion
parameters such as Peak Ground Acceleration (PGA), Peak
Ground Velocity (PGV), and Peak Ground Displacement
(PGD) were estimated using OpenSeismoMatlab [53], aiding
in understanding the precursory activities associated with the
slope instabilities under study.

IV. RESULTS AND DISCUSSIONS

This section presents findings derived from executing val-
idation tests of the proposed microseismic sensing system
in controlled laboratory settings and outdoor field trials.
It further delves into the preliminary analysis of microseismic
geophone signals, captured from real-world environments
intricately linked with the causative factors of these natural
disasters, all aimed at establishing effective early warning
systems.

A. LABORATORY SIMULATIONS: SLIP SURFACE

MOVEMENTS
The geophone signals captured during the landslide labo-
ratory simulation represented slip surface movements, with
signals from the crown (G1) and middle (G2) geophones
shown in Figure 9. They depict the downslope movement
of soil material along the slip surface of the testbed.
The geophone signals from three channels are portrayed
using a spectrogram to discern their attributes and track
FIGURE 9. Recorded Signals of slip movements by the two geophone
changes in frequency over time. Owing to the distinct sensor sensor systems a) G1 at crown b) G2 at middle region and their respective
positions, there is a noticeable delay in data capture. The spectrogram for each channel.
signal’s evolution, as the soil descends the slope, is distinctly
observable, with G1 at the crown registering the onset of the
sliding movement, followed by a slight delay in data capture spectrogram effectively captures the occurrence of footsteps
by G2 as it progresses downwards, as shown in Figure 9a) at regular intervals, revealing a fundamental frequency range
and 9b). of (0.5-1) Hz. This periodicity indicates that the footstep
This behavior mirrors that of an actual landslide event, signal repeats itself approximately every 1 to 2 seconds.
as evidenced by the spectrogram displaying high intensity As the individual approaches while walking, the strength
in the middle, signifying the transmission of energy from of the footstep signal increases over a frequency range
the crown to the middle. The signals were acquired at the of (20-50) Hz. This behavior is expected, as the footstep
rate of 1 kHz, each spanning 1 minute 30 seconds and signals are distinctly pronounced when the person is closer
1 minute 10 seconds. The signals showed maximum strength to the geophone. Conversely, as the individual moves away,
in the frequency range of (0-5) Hz at both positions. Besides, the strength of the footstep signals decreases, resulting
consistent frequency ranges between (10-20) Hz were in a gradual decline in signal amplitude. However, the
observed, which could be attributed to interferences from spectrogram also reveals the presence of consistent frequency
the ambient environment. The analysis revealed frequencies interference around 10 Hz. These interfering frequencies are
below 5 Hz, consistent with the anticipated frequency range likely due to noises from the ambient environment.
for landslide signals as reported in [36]. One of the trials for vehicle movement is shown in Fig-
ure 10b) lasting 28 seconds, captured by the geophone sensor
B. OUTDOOR FIELD TRIALS: PEDESTRIAN AND VEHICLE system. The accompanying spectrogram analysis effectively
MOVEMENTS captures the vehicle movements, reaching frequencies of up
Figure 10a) shows one of the trials for footstep movement, to 60 Hz. Notably, the strength of the frequencies persists
which lasted for a duration of 40 seconds. The accompanying for (2.0-4.8) seconds, particularly as the vehicle approaches

97796 VOLUME 12, 2024

P. K. Indukala et al.: IoT- Driven Microseismic Sensing System and Monitoring Platform

TABLE 2. Observations from lab and outdoor simulations.

1) GEOPHONE SIGNAL RESPONSE FOR SLOPE

INSTABILITIES DURING EARTHQUAKE TRIGGERS
As noted in Table 3, E1 refers to an earthquake, on September
12, 2018, measuring 5.5 on the Richter scale that struck the
Kokrajhar District in Assam, India [54]. The quake occurred
at approximately 10:20:49 AM Indian Standard Time (IST),
with the epicenter at 26.4◦ N and 90.1◦ E. Despite being
around 187 km away from the epicenter, the proposed system
for geophones was able to record the microseismic triggers
of this tremor event at three different DEP locations: DEP 8,
DEP 10, and DEP 5.
Figure 11 shows the time domain representation of
the geophone data captured by the monitoring system
from the three DEP locations. For further analysis, the
geophone signals were cropped to cover a total duration of
approximately 1 hour, including half an hour both before,
and after the time of the earthquake incident. Figure 11a),
FIGURE 10. Recorded signals captured by a geophone sensor system for shows significant ground movements that persisted for about
a) footstep and b) vehicular movements and their respective spectrogram.
2.4 minutes. Notably, the signal exhibited substantial strength
within the frequency range of (0 -10) Hz, with dominant
frequency peaks concentrated around (1-2) Hz.
closer to the geophones. This behavior is indicative of the In Figure 11b), notable changes in signal strength lasting
vehicle’s proximity to the sensor. Moreover, in the frequency for about 1.33 minutes can be observed. The dominant
spectrum, a prominent strength is observed around (10-15) spectral range, (below 5 Hz) which was consistent throughout
Hz, which gradually increases up to the range of (50-60) this event can be due to electrical interferences of that
Hz before decreasing again to around (10-15) Hz. These channel. It is observed that there is a dominant range
variations in frequency content can be attributed to specific of (0.5-1) Hz across all channels, around 1.33 minutes.
vehicle engine sounds and their harmonics, which become In Figure 11c), the ground shakings due to the earthquake
more pronounced as the vehicle gets nearer and gradually triggers caused a sudden drop in frequency ranges, with
recedes. The findings are consistent with the expected the effect lasting for approximately 15 minutes. However,
footstep monitoring and vehicular movement frequencies higher frequencies up to 30 Hz were observed in the
reported in [20]. spectrum throughout this event, which can be due to electrical
Table 2 lists the observations from laboratory and out- interferences of the channels itself.
door experimentation. The results indicate that human and Ground motion parameters, PGA, PGV, PGD, alongside
vehicular movements can generate vibrations and signals that the occurrence of earthquake triggers, were estimated for
overlap the frequency ranges of microseismic events, despite the 24 hour-geophone data from 9 to 25 September 2018.
their distinct characteristics. Figure 12) displays the fluctuations in ground motion
parameters, where a significant increase followed by a
C. PRELIMINARY ANALYSIS OF MICROSEISMIC SIGNALS: gradual decrease in the values was observed during the event.
DEPLOYMENT SITE, CHANDMARI Earthquake event, E2, a 5.4 magnitude earthquake occured
The slope instabilities, captured by the deployed SM-s on March 20, 2020, approximately 153 km NW of Yuksom,
nodes in Chandmari, correlated with earthquakes and rain- Sikkim, India, at 07:03:17 IST. This event was followed by
fall precipitation events in the vicinity of the test site. two more quakes on March 25, 2020, measuring 3.8 and
Table 3 displays a few events recorded by the SM-s 3.3 on the Richter scale; the first earthquake occurred at
nodes, corresponding to the triggers for slope instability, 05:09:32 IST, 167 km NNW of Yuksom, Sikkim and the
type of movement, observed range of frequencies, and the second tremors were recorded at 04:43:25 IST, 197 km NW
estimates of ground motion. This highlights the proposed of Yuksom, Sikkim. [54]. The proposed microseismic system,
sensing system’s effectiveness in capturing the microseismic situated approximately 40 km away from the epicenter at
precursor signals of the factors triggering landslides. Yuksom, Sikkim captured the slope instabilities, as shown

VOLUME 12, 2024 97797

 P. K. Indukala et al.: IoT- Driven Microseismic Sensing System and Monitoring Platform

FIGURE 11. Geophone Signals captured by the developed microseismic system at three DEP locations at the deployment site, Chandmari during the
slope instability caused due to earthquake trigger recorded on 12 September 2018.

FIGURE 12. Variation of the Ground Motion Parameters: Peak Ground

Acceleration (PGA), Peak Ground Velocity (PGV), Peak Ground
Displacement (PGD), and occurrence of slope instability due to
earthquake trigger during September 2018, from DEP 9.

in Figure 13. The spectrogram revealed a prominent range

of (0.1-1.0) Hz, lasting less than 3 minutes. Other consistent
frequencies observed in the spectrum can be due to the
electrical interference in the channel.
Figure 13b) reveals a gradual increase in ground motion
parameters after the earthquake event on March 20, 2020,
indicative of the aftershocks in the region. However, for the FIGURE 13. a) Geophone Signals captured by the developed microseismic
system during slope stability due to earthquake trigger recorded
events on March 25, feasible variations in ground motion on 20 March 2020 b) Variation of the Ground Motion Parameters: PGA,
parameters were evident. The results shown in Figure 12 and PGV, PGD, and occurrence of slope instability due to earthquake triggers
during March 2020, from DEP 9.
Figure 13b) provide valuable insights into ground motion
variations associated with ground shakings in the region
subsequent to the earthquake events. The significant decrease
in ground motion parameters observed during and after the 2) GEOPHONE SIGNAL RESPONSE FOR SLOPE
tremor events suggests that these variations could serve as INSTABILITIES DURING RAINFALL TRIGGERS
precursors for the detection of slope instabilities set in by Figure 14 shows the available rainfall data collected from
ground shaking. However, arriving at conclusive findings the Indian Meteorological Department for the district of East
mandate a thorough analysis, which is beyond the scope of Sikkim for the years 2018-2021 [55]. It reveals that heavy
this paper’s investigation. rainfall was experiences during the period from April to

97798 VOLUME 12, 2024

P. K. Indukala et al.: IoT- Driven Microseismic Sensing System and Monitoring Platform

FIGURE 14. east sikkim rainfall data for the years 2018-2021.

FIGURE 16. a) Time domain representation of the slope instabilities

observed in the geophone data during the period 17-21 June 2020 for the
FIGURE 15. Slope instabilities near the site, triggered by rainfall. corresponding event observed on 19 June 2020, b) Variation of the ground
motion parameters during the period June 12-23, 2020, from DEP 9.

September. A comparison of the cumulative rainfall for June

and July shows an increase in the downpour from 432.45 mm
in 2018 to 579.85 mm in 2020 and 542.7 mm in 2021,
highlighting the increased probability of landslides at the
selected field site during the rainy season.
The slope instability caused by the sliding movement of
rock-debris, referred to as E3, was triggered following heavy
rainfall, on 19 June 2020, around 450 meters away from one
of our site locations at 27◦ 20.371‘‘N 88◦ 37.552’’E, as shown
in Figure 15. The geophone signals were recorded during the FIGURE 17. Variations observed in dominant frequency ranges at each
channels, during the period June 12-23, 2020, from DEP 9.
slope instabilities, due to rainfall triggers that happened in the
days as depicted in Figure 16a). The variation of the ground
motion parameters during the period (12-23) June 2020, severe, with sudden increase and decrease, which stabilized
from the location DEP 9 is shown in Figure 16b). It depicts in the following days, to the typical range of (15- 28) Hz,
that there was an increase in the amplitude of the ground which is considered as the frequency ranges observed during
motion parameters (3-4) days before the recorded event and a normal day scenarios.
significant decrease a day before the event. These variations Table 3 shows the observed dominant frequency ranges of
could serve as precursors for detecting slope instabilities of the geophone signals, correlated with triggers for the recorded
landslide initiation and seismic events. events at the Chandmari deployment site. The frequency
Figure 17 shows the variations in dominant frequency ranges differ depending on the specific site conditions, and
ranges, which may be due to rainfall and external noises the underlying geology. Nevertheless, this indication could
generated by the rain over time. This suggests that the impact serve as supplementary information, when combined with
of the heavy rainfall on the frequency range was initially more data from other sensors [13], [56], [57]. Doing so, enhances

VOLUME 12, 2024 97799

 P. K. Indukala et al.: IoT- Driven Microseismic Sensing System and Monitoring Platform

TABLE 3. Observations from geophone signals from chandmari. sensor malfunction and the possibility of power loss due
to extreme weather conditions in remote or rugged terrain,
such as the Himalayas. Although context-aware energy
management systems have been implemented [40], extreme
weather conditions may still result in power loss, preventing
the system from operating or leading to data loss during
transmission. Logistical and operational challenges may
turn up when installing the system in remote or rugged
terrain. These include maintainence and calibration of the
sensor network, to ensure data integrity and security, and
managing power sources in off-grid locations. Attention to
these potential hurdles is crucial for ensuring the long-term
the efficacy of decision-making support systems, with a clear viability of the proposed system.
link established between frequency ranges and signs of slope
instability during the rainy season. This necessitates further
V. COMPARISON WITH EXISTING SYSTEM
exploration of the geophone signals during specific weather
An assessment was conducted to evaluate whether the
conditions, which suggest ongoing creep and subsidence
presented research adequately addressed key functionalities
movements, indicative of impending landslides.
denoted as F1-F11, derived from the requirements discussed
Findings from laboratory experiments and field deploy-
in section I. Table 4 presents a curated selection of publicly
ments highlight the potential of the proposed microseismic
available research focused on microseismic monitoring using
sensing system to capture earth tremors, even the subtlest
in-situ geophysical methods, to detect triggers that ensue
microseismic triggers. This establishes the foundation for a
in landslides. The existing microseismic studies primarily
cost-effective, non-intrusive, and long-term spatiotemporal
address single triggers with limited data rates and lack
monitoring solution, laying the groundwork for an efficient
real-time and early warning capabilities, which renders
early warning system in landslide-prone areas. The results
them ill-suited for spatio-temporal monitoring. As a result,
of the analysis from the geophone-based sensing systems
achieving a fair comparison with previous approaches is
can be cross-referenced with other sensor data to enhance
not feasible. Nevertheless, the study aims to present some
early warning capabilities. However, achieving accurate event
comparative analysis. Table 4 outlines the applications and
detection mandates a precise distinction between relevant
sensing instruments utilized in the literature.
signals and background noise, which compel focused effort
on refinement of signal processing algorithms, and its • F1: High frequency data capture-Incessant acquisition of
applicability at the edge. Enhancement of context-aware data samples.
data and energy management techniques can significantly • F2: Microseismic data acquisition- Generation of
conserve power consumption in edge sensing systems, seismic waves from ground/ anthropogenic activities/
thereby extending system lifetime and effective manipulation industrial processes.
of resource constraints [40], [58]. • F3: Resource constraints- Limited processing, storage
potential, minimum power in IoT environment.
D. LIMITATIONS • F4: Continuous, and real time monitoring- Delivery of
Albeit the proposed microseismic sensing system affords continuous microseismic data
promising capabilities for landslide detection and early warn- • F5: Large scale spatio-temporal monitoring- Extensive
ing, careful consideration of potential risks and challenges surveillance and perusal of a vast geographical area over
is imperative to maximize its effectiveness, and minimize an extended period.
adverse impacts. One of the significant risks is the likelihood • F6: Precursor signal detection- Identify early pre-triggers
of false alarms or misinterpretation of micro-seismic signals, associated with F2.
leading to unnecessary panic. The proposed solution has • F7: Microseismic activity detection- Methods for detect-
taken into account scenarios such as misinterpretation ing microseismic data, related to F2.
of microseismic signals due to footsteps, and vehicular • F8: Early warning capabilities
movements. The solution includes methodologies to detect • F9: Mapping to microseismic triggers- due to heavy
and classify these signals, thereby avoiding false alarms. rainfall, seismic activity.
In addition, misinterpretation of signals during initiation of • F10: Communication Technology- Data transmission or
slope instability has been mitigated by integrating heteroge- local storage.
neous sensors, capable of monitoring multiple trigger types • F11: Monitoring Period- Short term (Monthly, or shorter
that are prominent in different stages of the slope instability durations), Long term (Yearly or longer durations)
initiation pathway [12], [49], [56]. A critical review of 20 selected articles in the field
It is also important to consider events such as the of microseismic monitoring was conducted to compare
gradual movement of slopes by creep landslides, causing them with the proposed system. Figure 18 presents the

97800 VOLUME 12, 2024

P. K. Indukala et al.: IoT- Driven Microseismic Sensing System and Monitoring Platform

TABLE 4. Comparison of existing systems and proposed system.

VOLUME 12, 2024 97801

 P. K. Indukala et al.: IoT- Driven Microseismic Sensing System and Monitoring Platform

FIGURE 18. Comparative analysis with the exisiting systems.

comparative analysis of these reviewed articles, highlighting tegration pattern of the landslide body, fast transmission, and
factors F1 through F11. Firstly, 85 % of the reviewed comprehensive analysis in creep slopes on a continuous basis.
studies emphasize the importance of capturing microseismic However, such systems have limitations, including lower data
data (F2), and 90 % discuss methods for microseismic transmission rates, lack of microseismic signal capturing,
activity detection (F7). Additionally, 80 % consider their and spatiotemporal monitoring capabilities. Studies such
applications to have early warning capabilities (F8), and as [61] and [78] presents insights into technologies for
75 % use standard communication technologies for data landslide monitoring, while [62], [68] discuss the design and
transmission (F10), the remaining 25 % either do not specify experimental systems for detecting early signs of landslide
their methods or rely on local storage. However, only 35 events using integrated sensors but lack a comprehensive
% focus on high-frequency data capture (F1), revealing an framework and end-to-end real-time system architecture.
underutilization of high-frequency microseismic signals that Another study by [71], investigates early warning system
could provide more detailed insights into landslide dynamics. frameworks for rapidly detecting earthquake ground motions,
Furthermore, 20 % of the reviewed studies consider does not involve sensing hardware deployment or mainte-
resource constraints in the development of IoT systems (F3), nance, but utilizes smartphones incurring high operational
highlighting the need for further research into cost-effective costs. Similar methodologies for landslide early warning
and efficient IoT solutions for resource-limited regions. Only systems with advanced capabilities remain limited. The
15 % of the investigations explore precursory triggers and existing systems in Table 4 cover only a portion of the
the detection of landslides (F6), indicating a pressing need essential functionalities required for effective microseismic
for more research into identifying and understanding early monitoring, suggesting the need for a more comprehensive
warning signs. Additionally, 35 % of the studies fail to approach.
adequately map microseismic triggers (F9), suggesting a The proposed system, in comparison, showcases robust
gap in integrating microseismic data with geological and capabilities of the sensing system with its cost-effective
environmental factors. 50 % of the reviewed studies focus passive sensing system, utilizing geophone sensors in detect-
on short-term monitoring, which can be monthly or even ing and monitoring microseismic events and their triggers.
shorter, while the other 50 % focus on long-term monitoring, It captures microseismic vibrations at exceptionally high data
considered as yearly or more. Moreover, 45 % of the studies rates (1 kHz, across each sensor channel), from multiple
lack systems capable of large-scale spatiotemporal and real- locations, fulfilling all functionalities, and provides numerous
time monitoring (F4, F5), underscoring a significant paucity advantages over traditional monitoring methods [16], [19],
in comprehensive monitoring capabilities. This comparative [20], [21], [22]. The proposed system has been deployed
analysis emphasizes the critical areas where advancements in real-world scenarios supporting long-term large-scale
are needed, particularly in high-frequency data capture, spatio-temporal monitoring, and enabling the mapping of
integration of microseismic data with environmental factors, microseismic triggers associated with slope instabilities due
comprehensive monitoring capabilities, and cost-effective to rainfall and earthquakes. The seamless connectivity of
IoT solutions. the proposed system significantly enhances its capability
A few studies satisfiying the minimum functionalities are to deliver actionable insights to stakeholders, emergency
detailed below. The study in [74] discusses an IoT based responders, local authorities, and residents, marking a
landslide monitoring and early warning system, offering significant advancement in microseismic monitoring and
accurate data acquisition of morphological change and disin- early warning systems.

97802 VOLUME 12, 2024

P. K. Indukala et al.: IoT- Driven Microseismic Sensing System and Monitoring Platform

VI. CONCLUSION AND FUTURE DIRECTIONS REFERENCES

This paper presented an IoT-driven platform for the capture [1] M. V. Ramesh, ‘‘Design, development, and deployment of a wireless sensor
of subterranean microseismic vibrations in regions prone network for detection of landslides,’’ Ad Hoc Netw., vol. 13, pp. 2–18,
Feb. 2014.
to landslides. The proposed microseismic edge sensing [2] A. Novellino, C. Pennington, K. Leeming, S. Taylor, I. G. Alvarez,
system, incorporating geophone sensors (SM-s nodes), was E. McAllister, C. Arnhardt, and A. Winson, ‘‘Mapping landslides from
deployed to continuously record and transmit real-time space: A review,’’ Landslides, vol. 21, no. 5, pp. 1041–1052, May 2024.
microseismic data from landslide-prone areas of Chandmari, [3] Public EM-DAT Platform. Accessed: May 5, 2024. [Online]. Available:
https://public.emdat.be
in the state of Sikkim, in North eastern Himalayas. The [4] M. J. Froude and D. N. Petley, ‘‘Global fatal landslide occurrence
proposed system successfully captured the slope instability from 2004 to 2016,’’ Natural Hazards Earth Syst. Sci., vol. 18, no. 8,
events in June 2020, and earthquake events in September pp. 2161–2181, Aug. 2018.
[5] C.-W. Lin, S.-H. Liu, S.-Y. Lee, and C.-C. Liu, ‘‘Impacts of the chi-chi
2018, and March 2020. Results from laboratory experiments earthquake on subsequent rainfall-induced landslides in central Taiwan,’’
and field deployments demonstrate the effectiveness of the Eng. Geol., vol. 86, nos. 2–3, pp. 87–101, Aug. 2006.
proposed sensing system in capturing subtle pre-triggers of [6] D. K. Keefer, J. Wartman, C. N. Ochoa, A. Rodriguez-Marek, and
G. F. Wieczorek, ‘‘Landslides caused by the 7.6 Tecomán, Mexico
slope instabilities, precipitated by rainfall and earthquakes.
earthquake of Jan. 21, 2003,’’ Eng. Geol., vol. 86, nos. 2–3, pp. 183–197,
This forms the basis for a cost-effective, non-intrusive, 2003.
and long-term spatiotemporal monitoring solution, setting a [7] Z.-Y. Feng, C.-M. Hsu, and S.-H. Chen, ‘‘Discussion on the characteristics
robust foundation for an efficient early warning system in of seismic signals due to riverbank landslides from laboratory tests,’’
Water, vol. 12, no. 1, p. 83, Dec. 2019.
landslide-prone regions. [8] S. Cuomo, ‘‘Modelling of flowslides and debris avalanches in natural and
The proposed microseismic geophone sensing system engineered slopes: A review,’’ Geoenvironmental Disasters, vol. 7, no. 1,
demonstrated promising capabilities in monitoring a diverse pp. 1–25, Dec. 2020.
[9] R. Bhasin, E. Grimstad, J. O. Larsen, A. K. Dhawan, R. Singh, S. K. Verma,
range of geological phenomena such as rock falls, rock- and K. Venkatachalam, ‘‘Landslide hazards and mitigation measures at
slides, debris flows, earthquakes, snow avalanches, and Gangtok, Sikkim Himalaya,’’ Eng. Geol., vol. 64, no. 4, pp. 351–368,
other seismic activities induced by anthropogenic actions. Jun. 2002.
The microseismic signal analysis will aid in understanding [10] N. Vasudevan and K. Ramanathan, ‘‘Geological factors contributing
to landslides: Case studies of a few landslides in different regions of
the principal mechanisms governing each stage of these India,’’ Proc. IOP Conf. Series, Earth Environ. Sci., vol. 30, pp. 011–012,
potential hazards. It is crucial for the refinement of robust Aug. 2016.
signal processing algorithms and comprehensive validation [11] N. Vasudevan, K. Ramanathan, and M. S. Ananthapathmanabhan, ‘‘Visual
representation of slope stability for landslide monitoring systems,’’ in Proc.
protocols, including real-time validation. Ongoing research 4th Int. Conf. Converg. Technol. (I2CT), Oct. 2018, pp. 1–5.
involves analysis of microseismic data from various locations [12] T. Hemalatha, M. V. Ramesh, and V. P. Rangan, ‘‘Effective and accelerated
and cross-referencing with environmental factors, intended to forewarning of landslides using wireless sensor networks and machine
learning,’’ IEEE Sensors J., vol. 19, no. 21, pp. 9964–9975, Nov. 2019.
be incorporated in future studies. The proposed system can be [13] M. V. Ramesh, D. Pullarkatt, T. H. Geethu, and P. V. Rangan, ‘‘Wireless
further enhanced to derive adaptive thresholds using detailed sensor networks for early warning of landslides: Experiences from a decade
data analysis and machine learning approaches. This will long deployment,’’ in Proc. Workshop World Landslide Forum, 2017,
pp. 41–50.
contribute to real-time identification and classification of the
[14] M. V. Ramesh and N. Vasudevan, ‘‘The deployment of deep-Earth sensor
characteristics and rate of landslide movements in different probes for landslide detection,’’ Landslides, vol. 9, no. 4, pp. 457–474,
materials, enabling automatic classification and mapping of Dec. 2012.
imminent landslides. [15] H. Thirugnanam, S. Uhlemann, R. Reghunadh, M. V. Ramesh, and
V. P. Rangan, ‘‘Review of landslide monitoring techniques with IoT
integration opportunities,’’ IEEE J. Sel. Topics Appl. Earth Observ. Remote
Sens., vol. 15, pp. 5317–5338, 2022.
ACKNOWLEDGMENT [16] S. Uhlemann, A. Smith, J. Chambers, N. Dixon, T. Dijkstra, E. Haslam,
P. Meldrum, A. Merritt, D. Gunn, and J. Mackay, ‘‘Assessment of
The authors express our deep gratitude to Sri Mata Amri- ground-based monitoring techniques applied to landslide investigations,’’
tanandamayi Devi (AMMA), Chancellor, Amrita Vishwa Geomorphology, vol. 253, pp. 438–451, Jan. 2016.
Vidyapeetham, who bestowed the motivation and inspiration [17] T. Carla, T. Nolesini, L. Solari, C. Rivolta, L. Dei Cas, and N. Casagli,
to pursue this research work. They undertook this work fol- ‘‘Rockfall forecasting and risk management along a major transportation
corridor in the Alps through ground-based radar interferometry,’’ Land-
lowing their recognition as the ‘‘World Center of Excellence slides, vol. 16, no. 8, pp. 1425–1435, Aug. 2019.
in landslide disaster risk reduction,’’ conferred on them by [18] J. S. Whiteley, J. E. Chambers, S. Uhlemann, P. B. Wilkinson, and
the ‘‘International Programme on Landslides,’’ in August J. M. Kendall, ‘‘Geophysical monitoring of moisture-induced landslides:
A review,’’ Rev. Geophys., vol. 57, no. 1, pp. 106–145, Mar. 2019.
2017, November 2021, and November 2023. They gratefully [19] S. Pirasteh and J. Li, ‘‘Landslides investigations from geoinformatics
appreciate the contributions of the Entire Landslide Team perspective: Quality, challenges, and recommendations,’’ Geomatics,
in their research center for their support in various aspects. Natural Hazards Risk, vol. 8, no. 2, pp. 448–465, Dec. 2017.
[20] V. N. Deekshit, M. V. Ramesh, P. K. Indukala, and G. J. Nair, ‘‘Smart
They acknowledge the valuable suggestions from Dr. Venkat
geophone sensor network for effective detection of landslide induced
Rangan, Prof. Balaji Hariharan, Dr. T. Hemalatha, Dr. M. geophone signals,’’ in Proc. Int. Conf. Commun. Signal Process. (ICCSP),
Ravisankar, and Nitin Kumar. They wish to acknowledge Apr. 2016, pp. 1565–1569.
the support provided by Sravan Thampan, Dawa Tshering [21] V. Del Gaudio, J. Wasowski, and S. Muscillo, ‘‘New developments
in ambient noise analysis to characterise the seismic response of
Lepcha, Ramesh Guntha, Balmukund Singh, Dr. Rekha landslide-prone slopes,’’ Natural Hazards Earth Syst. Sci., vol. 13, no. 8,
Prabha, Deepak Brahmanandan, and Arun Kumar Jijilal. pp. 2075–2087, Aug. 2013.

VOLUME 12, 2024 97803

 P. K. Indukala et al.: IoT- Driven Microseismic Sensing System and Monitoring Platform

[22] A. T. Kunnath and M. V. Ramesh, ‘‘Integrating geophone network to real- [44] Y. Pu, D. B. Apel, and R. Hall, ‘‘Using machine learning approach for
time wireless sensor network system for landslide detection,’’ in Proc. 1st microseismic events recognition in underground excavations: Comparison
Int. Conf. Sensor Device Technol. Appl., Jul. 2010, pp. 167–171. of ten frequently-used models,’’ Eng. Geol., vol. 268, Apr. 2020,
[23] D. M. Cruden and D. J. Varnes, ‘‘Landslides: Investigation and mitigation. Art. no. 105519.
chapter 3-landslide types and processes,’’ Transp. Res. board special Rep., [45] S. Kim, B. Yoon, J.-T. Lim, and M. Kim, ‘‘Data-driven signal–noise
vol. 247, no. 1, pp. 1–20, 1996. classification for microseismic data using machine learning,’’ Energies,
[24] F. G. Murillo-García, M. Rossi, F. Ardizzone, F. Fiorucci, and I. Alcántara- vol. 14, no. 5, p. 1499, Mar. 2021.
Ayala, ‘‘Hazard and population vulnerability analysis: A step towards [46] W. Jinqiang, P. Basnet, and S. Mahtab, ‘‘Review of machine learning
landslide risk assessment,’’ J. Mountain Sci., vol. 14, no. 7, pp. 1241–1261, and deep learning application in mine microseismic event classification,’’
Jul. 2017. Mining Mineral Deposits, vol. 15, no. 1, pp. 19–26, 2021.
[25] P. Lacroix, A. L. Handwerger, and G. Bièvre, ‘‘Life and death of slow- [47] M. Shafiq, Z. Tian, Y. Sun, X. Du, and M. Guizani, ‘‘Selection of effective
moving landslides,’’ Nature Rev. Earth Environ., vol. 1, no. 8, pp. 404–419, machine learning algorithm and bot-IoT attacks traffic identification for
Jul. 2020. Internet of Things in smart city,’’ Future Gener. Comput. Syst., vol. 107,
[26] T. R. Martha, N. Kerle, V. Jetten, C. J. van Westen, and K. V. Kumar, pp. 433–442, Jun. 2020.
‘‘Characterising spectral, spatial and morphometric properties of land- [48] M. Shafiq, Z. Tian, A. K. Bashir, A. Jolfaei, and X. Yu, ‘‘Data mining and
slides for semi-automatic detection using object-oriented methods,’’ machine learning methods for sustainable smart cities traffic classification:
Geomorphology, vol. 116, nos. 1–2, pp. 24–36, Mar. 2010. A survey,’’ Sustain. Cities Soc., vol. 60, Sep. 2020, Art. no. 102177.
[27] N. Hovius and P. Meunier, Earthquake Ground Motion and Patterns [49] G. T. Harilal, D. Madhu, M. V. Ramesh, and D. Pullarkatt, ‘‘Towards
of Seismically Induced Landsliding. Cambridge, U.K.: Cambridge Univ. establishing rainfall thresholds for a real-time landslide early warning
Press, 2012, pp. 24–36. system in Sikkim, India,’’ Landslides, vol. 16, no. 12, pp. 2395–2408,
[28] D. M. Cruden and R. Couture, ‘‘More comprehensive characterization Dec. 2019.
of landslides: Review and additions,’’ in Proc. 11th Int. IAEG Congr., [50] K. P. Soman, S. S. Kumar, N. Mohan, and P. Poornachandran, ‘‘Modern
Geologically Act., 2010, pp. 5–10. methods for signal analysis and its applications,’’ Recent Adv. Comput.
[29] J. J. Clague and D. Stead, Landslides: Types, Mechanisms Model. Intell., vol. 1, pp. 263–290, Jun. 2019.
Cambridge, U.K.: Cambridge Univ. Press, 2012. [51] U. V. Krishna, R. Priyamvada, G. Jyothish, V. Sowmya, and K. P. Soman,
‘‘A comparative evaluation of decomposition methods based on pitch
[30] B. P. Bhandari and S. Dhakal, ‘‘A multidisciplinary approach of landslide
estimation of piano notes,’’ in Proc. 4th Int. Conf. Smart Comput. Informat,
characterization: A case of the siwalik zone of Nepal Himalaya,’’ J. Asian
vol. 1, 2021, pp. 833–843.
Earth Sci., X, vol. 5, Jun. 2021, Art. no. 100061.
[52] P. Adarsh, P. K. Indukala, and B. Hariharan, ‘‘An adaptive thresholding
[31] N. Tisato and C. Madonna, ‘‘Attenuation at low seismic frequencies
framework for detecting the subsurface movements in landslide-prone
in partially saturated rocks: Measurements and description of a new
sites,’’ in Proc. 3rd Int. Conf. Pervasive Comput. Social Netw. (ICPCSN),
apparatus,’’ J. Appl. Geophys., vol. 86, pp. 44–53, Nov. 2012.
Jun. 2023, pp. 1502–1507.
[32] G. Koç and K. Yegin, ‘‘Footstep and vehicle detection using slow and quick
[53] G. Papazafeiropoulos and V. Plevris, ‘‘OpenSeismoMATLAB: A new
adaptive thresholds algorithm,’’ Int. J. Distrib. Sensor Netw., vol. 9, no. 10,
open-source software for strong ground motion data processing,’’ Heliyon,
Oct. 2013, Art. no. 783604.
vol. 4, no. 9, Sep. 2018, Art. no. e00784.
[33] A. E. Faghfouri and M. B. Frish, ‘‘Robust discrimination of human
[54] U.S. Geological Surv. (2023). M5.3—6km Wnw of Toms Place.
footsteps using seismic signals,’’ Proc. SPIE, vol. 8046, May 2011,
[Online]. Available: https://earthquake.usgs.gov/earthquakes/eventpage/
Art. no. 80460D.
us2000hd8v/map
[34] P. Anghelescu, G. V. Iana, and I. Tramandan, ‘‘Human footstep detection
[55] Rainfall Data of Sikkim. Accessed: May 5, 2024. [Online]. Available:
using seismic sensors,’’ in Proc. 7th Int. Conf. Electron., Comput. Artif.
http://sikenvis.nic.in/Database/Rainfall_4227.aspx#
Intell. (ECAI), Jun. 2015, pp. 1–15.
[56] H. Thirugnanam, M. V. Ramesh, and V. P. Rangan, ‘‘Enhancing the
[35] L. Yang, Y. Xu, L. Wang, and Q. Jiang, ‘‘Seismic signal characteristics reliability of landslide early warning systems by machine learning,’’
and numerical modeling analysis of the Xinmo landslide,’’ Sustainability, Landslides, vol. 17, no. 9, pp. 2231–2246, Sep. 2020.
vol. 15, no. 7, p. 5851, Mar. 2023.
[57] M. Sudheer, ‘‘Wireless sensor network for disaster monitoring,’’ in
[36] Y. Yan, Y. Cui, X. Tian, S. Hu, J. Guo, Z. Wang, S. Yin, and Wireless Sensor Networks: Application—Centric Design. Rijeka, Croatia:
L. Liao, ‘‘Seismic signal recognition and interpretation of the 2019 ‘7.2’ InTech, 2010, doi: 10.5772/13119.
shuicheng landslide by seismogram stations,’’ Landslides, vol. 17, no. 5, [58] P. Pradeep, S. Krishnamoorthy, R. K. Pathinarupothi, and A. V. Vasilakos,
pp. 1191–1206, May 2020. ‘‘Leveraging context-awareness for Internet of Things ecosystem: Repre-
[37] E. Por, M. van Kooten, and V. Sarkovic, ‘‘Nyquist–Shannon sampling sentation, organization, and management of context,’’ Comput. Commun.,
theorem,’’ Leiden Univ., vol. 1, no. 1, p. 5, 2019. vol. 177, pp. 33–50, Sep. 2021.
[38] S. M. Karunarathne, M. Dray, L. Popov, M. Butler, C. Pennington, [59] C. Abancó, M. Hurlimann, B. Fritschi, C. Graf, and J. Moya, ‘‘Transfor-
and C. M. Angelopoulos, ‘‘A technological framework for data-driven mation of ground vibration signal for debris-flow monitoring and detection
IoT systems: Application on landslide monitoring,’’ Comput. Commun., in alarm systems,’’ Sensors, vol. 12, no. 4, pp. 4870–4891, Apr. 2012.
vol. 154, pp. 298–312, Mar. 2020. [60] M. Hurlimann, C. Abancó, and J. Moya, ‘‘Rockfalls detached from a
[39] P. Maxwell and M. Lansley, ‘‘What receivers will we use for lateral moraine during spring season. 2010 and 2011 events observed at
low frequencies?’’ in SEG Technical Program Expanded Abstracts. the rebaixader debris-flow monitoring site (Central pyrenees, Spain),’’
Houston, TX, USA: Society of Exploration Geophysicists, 2011, Landslides, vol. 9, no. 3, pp. 385–393, Sep. 2012.
pp. 72–76. [61] A. M. Zambrano, I. Perez, C. Palau, and M. Esteve, ‘‘Technologies of
[40] R. Prabha, M. V. Ramesh, V. P. Rangan, P. V. Ushakumari, and Internet of Things applied to an earthquake early warning system,’’ Future
T. Hemalatha, ‘‘Energy efficient data acquisition techniques using context Gener. Comput. Syst., vol. 75, pp. 206–215, Oct. 2017.
aware sensing for landslide monitoring systems,’’ IEEE Sensors J., vol. 17, [62] Q. A. Gian, D.-T. Tran, D. C. Nguyen, V. H. Nhu, and D. Tien Bui,
no. 18, pp. 6006–6018, Sep. 2017. ‘‘Design and implementation of site-specific rainfall-induced landslide
[41] S. Kumar and M. V. Ramesh, ‘‘Lightweight management framework (IMF) early warning and monitoring system: A case study at nam dan
for a heterogeneous wireless network for landslide detection,’’ in Proc. Int. landslide (Vietnam),’’ Geomatics, Natural Hazards Risk, vol. 8, no. 2,
Conf. Web Semantic Technol., 2010, pp. 457–471. pp. 1978–1996, Dec. 2017.
[42] R. Guntha, S. Kumar, and B. Hariharan, ‘‘Scalable, secure, fail [63] M. Klapez, C. A. Grazia, S. Zennaro, M. Cozzani, and M. Casoni, ‘‘First
safe, and high performance architecture for storage, analysis, and experiences with earthcloud, a low-cost, cloud-based IoT seismic alert
alerts in a multi-site landslide monitoring system,’’ in Advancing system,’’ in Proc. 14th Int. Conf. Wireless Mobile Comput., Netw. Commun.
Culture of Living With Landslides. Cham, Switzerland: Springer, 2017, (WiMob), Oct. 2018, pp. 262–269.
pp. 61–69. [64] H. Tariq, F. Touati, M. A. E. Al-Hitmi, D. Crescini, and A. Ben Mnaouer,
[43] D. Anikiev, C. Birnie, U. B. Waheed, T. Alkhalifah, C. Gu, D. J. Verschuur, ‘‘A real-time early warning seismic event detection algorithm using smart
and L. Eisner, ‘‘Machine learning in microseismic monitoring,’’ Earth- geo-spatial bi-axial inclinometer nodes for Industry 4.0 applications,’’
Science Rev., vol. 239, Apr. 2023, Art. no. 104371. Appl. Sci., vol. 9, no. 18, p. 3650, Sep. 2019.

97804 VOLUME 12, 2024

P. K. Indukala et al.: IoT- Driven Microseismic Sensing System and Monitoring Platform

[65] J. Fu, Z. Li, H. Meng, J. Wang, and X. Shan, ‘‘Performance evaluation U. G. GOSH received the bachelor’s degree in
of low-cost seismic sensors for dense earthquake early warning: 2018– electronics and communication engineering from
2019 field testing in Southwest China,’’ Sensors, vol. 19, no. 9, p. 1999, Amrita Vishwa Vidyapeetham. He is currently
Apr. 2019. a Project Associate with the Center for Wire-
[66] C. Peng, P. Jiang, Q. Chen, Q. Ma, and J. Yang, ‘‘Performance evaluation of less Networks and Applications, Amrita Vishwa
a dense MEMS-based seismic sensor array deployed in the sichuan-yunnan Vidyapeetham. He has more than 12 years of
border region for earthquake early warning,’’ Micromachines, vol. 10, experience working on different research projects
no. 11, p. 735, Oct. 2019.
funded by organizations, such as the Ministry
[67] L. D’Errico, F. Franchi, F. Graziosi, A. Marotta, C. Rinaldi, M. Boschi, and
of Earth Sciences (MoES), and the Government
A. Colarieti, ‘‘Structural health monitoring and earthquake early warning
of India. His interests include embedded system
on 5G uRLLC network,’’ in Proc. IEEE 5th World Forum Internet Things
(WF-IoT), Apr. 2019, pp. 783–786. development and the Internet of Things.
[68] J. Faillettaz, M. Funk, J. Beutel, and A. Vieli, ‘‘Towards early warning of
gravitational slope failure with co-detection of microseismic activity: The
case of an active rock glacier,’’ Natural Hazards Earth Syst. Sci., vol. 19,
no. 7, pp. 1399–1413, Jul. 2019.
[69] K. Fauvel, D. Balouek-Thomert, D. Melgar, P. Silva, A. Simonet,
G. Antoniu, A. Costan, V. Masson, M. Parashar, I. Rodero, and A. Termier,
‘‘A distributed multi-sensor machine learning approach to earthquake
early warning,’’ in Proc. AAAI Conf. Artif. Intell., vol. 34, no. 1, 2020,
pp. 403–411.
[70] K. K. Khedo, Y. Bissessur, and D. S. Goolaub, ‘‘An inland wireless sensor
network system for monitoring seismic activity,’’ Future Gener. Comput.
Syst., vol. 105, pp. 520–532, Apr. 2020.
[71] R. M. Allen, Q. Kong, and R. Martin-Short, ‘‘The MyShake platform:
A global vision for earthquake early warning,’’ Pure Appl. Geophys.,
vol. 177, no. 4, pp. 1699–1712, Apr. 2020.
[72] M. Elmoulat, O. Debauche, S. Mahmoudi, S. A. Mahmoudi, P. Manneback,
and F. Lebeau, ‘‘Edge computing and artificial intelligence for landslides
monitoring,’’ Proc. Comput. Sci., vol. 177, no. 1, pp. 480–487, Jun. 2020.
[73] M. Gamperl, J. Singer, and K. Thuro, ‘‘Internet of Things geosensor
network for cost-effective landslide early warning systems,’’ Sensors,
vol. 21, no. 8, p. 2609, Apr. 2021.
[74] Z. Li, L. Fang, X. Sun, and W. Peng, ‘‘5G IoT-based geohazard monitoring
and early warning system and its application,’’ EURASIP J. Wireless
Commun. Netw., vol. 2021, no. 1, pp. 1–16, Dec. 2021.
[75] A. Makama, K. Kuladinithi, and A. Timm-Giel, ‘‘Wireless geophone
networks for land seismic data acquisition: A survey, tutorial and
performance evaluation,’’ Sensors, vol. 21, no. 15, p. 5171, Jul. 2021.
[76] Q.-M. Xie, K. Long, Z.-M. Li, L.-C. Chen, H. Liu, H. Xu, F. He, D. Liang,
X.-K. Xiang, and Q. Ma, ‘‘Stability evaluation for steep bank slope with
microseismic monitoring in three Gorges reservoir area,’’ J. Mountain Sci., MANEESHA VINODINI RAMESH (Senior
vol. 19, no. 6, pp. 1588–1601, Jun. 2022. Member, IEEE) received the Ph.D. degree in com-
[77] C.-M. Yang, J.-M. Chang, C.-Y. Hung, C.-H. Lu, W.-A. Chao, and puter science and engineering from Amrita Vishwa
K.-H. Kang, ‘‘Life span of a landslide dam on mountain valley caught on Vidyapeetham, in 2009, where she is currently
seismic signals and its possible early warnings,’’ Landslides, vol. 19, no. 3, the Provost, Strategic Initiatives, Research and
pp. 637–646, Mar. 2022. Innovation, Interdisciplinary, International, and
[78] R. Dhanagopal and B. Muthukumar, ‘‘A model for low power, high speed AI+X. She also holds the positions of the Director
and energy efficient early landslide detection system using IoT,’’ Wireless of Amrita Center for Wireless Networks, the Dean
Pers. Commun., vol. 117, no. 4, pp. 2713–2728, Apr. 2021. of International Programs, the Dean of the School
for Sustainable Development, and the UNESCO
Chair of experiential learning for Sustainable Innovation and Development,
P. K. INDUKALA received the bachelor’s degree School for Sustainable Development. She is listed among the top two
in information technology from Cochin University scientists in the world by an index made by Stanford University, in 2021.
of Science and Technology and the master’s degree She has more than 200 articles, including several journals and best paper
in computational engineering and networking awards. She has given invited talks at several eminent universities all over
from Amrita Vishwa Vidyapeetham, Coimbatore. the world. She is also the Co-Principal Investigator of European Commission
She is currently a Research Associate with the funded Wireless Sensor Networks with Self Organization Capabilities for
Center for Wireless Networks and Applications, Critical and Emergency Applications (WINSOC) Project and the Principal
Amrita Vishwa Vidyapeetham. She has more than Investigator of eight internationally recognized projects funded by different
nine years of experience working on different organizations from all over the world. In 2012, she received the NABARD
research projects funded by organizations, such as Award for Rural Innovation 2nd prize from the Honorable Finance Minister,
the Ministry of Earth Sciences (MoES) and the Department of Science and Government of India, for her research activities that benefited the rural
Technology (DST), Government of India. Her research interests include community. She served as the Program Chair for ACWR 2011. She is an
the Internet of Things, signal processing, machine learning, computational Editor of the Ad Hoc Networks Elsevier.
optimization, and deep learning.

VOLUME 12, 2024 97805