INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XII, December 2025  
Predictive Maintenance and Reliability Modeling of Rural  
Broadband Networks: A Condition-Based Study of BharatNet  
Infrastructure  
Vasikaran P, Dr. S. Madhivanan  
NSB Academy-Bangalore  
DOI : https://doi.org/10.51583/IJLTEMAS.2025.1412000103  
Received: 28 December 2025; Accepted: 02 January 2026; Published: 09 January 2026  
ABSTRACT  
This study explores the operational status and reliability of rural broadband infrastructure in 113 Gram  
Panchayats (GPs) in the Puducherry district, focusing on critical assets such as Optical Network Terminals  
(ONT), Central Control Units (CCU), battery systems, and solar panels, using a condition-based performance  
evaluation. Results from data analysis in SPSS show that 87.6% of GPs are operational (\"UP\"), but technical  
issues, including high fault rates in earthing and solar systems, are still prevalent, highlighting the need for  
predictive maintenance and real-time monitoring to ensure uninterrupted rural connectivity in support of India's  
digital empowerment objectives.  
INTRODUCTION  
The digital revolution has changed the socio-economic fabric of modern nations, but a key challenge remains in  
the disparity between urban and rural connectivity (Larsson, 2023). The BharatNet project is at the core of the  
Digital India vision and aims to transform 250,000 Gram Panchayats (GPs) into digitally enabled Gram  
Panchayats. In the Union Territory of Puducherry, this infrastructure is not just a technical installation but a  
public utility, with its backbone (ONT, CCU, solar panels and battery backups) providing the main access point  
for rural citizens to access e-governance, tele-education, and digital commerce. The research presented here  
explores a predictive and condition-based analysis of these assets to ensure that the promise of connectivity  
becomes realized in the delivery of high-quality service (Hagen & Andersen, 2024).  
Theoretical Framework  
To evaluate the efficacy and sustainability of rural broadband infrastructure, this research draws upon four  
primary academic lenses:  
Technology Acceptance Model (TAM): While installation is a technical task, the success of BharatNet will  
depend on user adoption; TAM posits that the key factors in whether or not rural communities will integrate  
these digital tools into their daily lives are Perceived Usefulness (reliable, high-speed access) and Perceived Ease  
of Use (seamless integration via solar backups) (Davis, 1989)  
Diffusion of Innovation (DOI) Theory: DOI theory focuses on how new technological standards, such as solar-  
powered broadband nodes, diffuse across administrative blocks (Rogers, 2003); it notes that the Rate of Adoption  
in Puducherry is often mediated by local technical literacy and availability of immediate maintenance support  
(Pfeffer & Salancik, 1978; Larsson, 2023).  
Resource Dependence Theory (RDT): This framework focuses on how local Gram Panchayats are dependent  
on external resources (stable grid power and centralized data networks); the study employs RDT to advocate for  
decentralized energy (solar) as a means to mitigate dependence. (Pfeffer & Salancik, 1978; Larsson, 2023)  
www.ijltemas.in  
Page 1173  
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XII, December 2025  
Statement of the Problem  
Although Puducherry has aggressively rolled out high-speed fiber, there is a big difference between the presence  
of infrastructure and the availability of service. Rural networks suffer from downtime due to equipment  
degradation, environmental stressors, and delayed maintenance response; current operational models are often  
reactive and begin repairs only after a total service outage is reported, which results in long periods of blackout  
and loss of public confidence in digital governance (Larsson, 2023). The urgent question becomes: How can we  
analyze the current condition of the hardware to create a predictive maintenance strategy that prevents failures?  
Industry and Regional Context  
Evolution of Rural Digital Infrastructure in India  
The Indian telecommunications sector has transitioned to "last-mile connectivity." The BharatNet project under  
Bharat Broadband Network Limited (BBNL) is the largest rural broadband project in the world. In Puducherry,  
this has been implemented through a complex mix of public sector undertakings such as BSNL and private  
partners such as Polycab and Xentric Integrated Solutions. This ecosystem is not a silo and feeds into other  
"related industries"( Larsson, 2023):  
1. Telecommunications: Fiber backbone for 4G/5G rollout  
2. Renewable Energy: Driving demand for localized solar-storage solutions  
3. Data Analytics: Utilizing performance data to optimize network traffic and maintenance schedules.  
Competitive Landscape and Stakeholder Analysis  
What started as a mandate to connect rural India has become a competitive space where public mandates clash  
with private efficiency as private giants like Reliance Jio, Bharti Airtel, and Vodafone Idea compete to deliver  
last-mile services through government contracts(Larsson, 2023).  
The supply chain for hardware is backed by industrial leaders(Hagen & Andersen, 2024):  
•Sterlite Technologies (STL): a global optical fiber and integrated digital network solution provider•  
KEI Industries: offers heavy-duty cabling to suit varying terrains.  
system integrators: firms like Xentric, play a critical role in connecting the hardware procurement to field-  
level operationality.  
This competition requires the infrastructure to be at a high level of performance, making condition-based analysis  
a necessity rather than an option for those who want to keep their contractual and operational edge.  
REVIEW OF LITERATURE  
A review of the existing literature reveals that a shift from reactive to proactive maintenance strategies has been  
observed in the large-scale rural infrastructure projects. Some key thematic areas from the literature are:  
Need for Data-Driven Models: Traditional maintenance in many photovoltaic (PV) and telecom systems is still  
corrective (reactive) or schedule-driven, which often leads to significant downtime and lower-than-expected  
system performance; the authors (Larsson, 2023) &(Hagen & Andersen, 2024) stress the need for a transition to  
predictive and prognostic maintenance using data-driven methods for identifying potential failures before service  
disruption occurs; and research indicates that economic models can be used to balance maintenance costs with  
component reliability and system performance.  
www.ijltemas.in  
Page 1174  
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XII, December 2025  
Sustainable Energy and Battery Management  
Degradation due to high state-of-charge (SOC) corrosion is a dominant aging mechanism in lead-acid batteries  
in solar home systems, and researchers (Wei, 2023,Perriment et al., 2023) have proposed intelligent battery  
management systems (BMS) to extend battery life by up to 25% through adaptive charging protocols based on  
real-time field data.•A recent review of 186 studies in the PV sector demonstrates the need for designing  
maintenance strategies around certain performance and reliability indicators.  
Optical Network Optimization  
There is a trade-off between attenuation thresholds and operational costs: Research on PON suggests that too  
low a threshold causes unnecessary technician dispatches, and too high a threshold puts service quality at risk,  
so data-driven policies are recommended to balance service-level agreements (SLAs) with maintenance  
expenditures . In short-reach optical systems and photovoltaic (PV) systems more and more ML models are  
being employed for anomaly detection to increase the accuracy and interpretability of fault diagnosis (Shao et  
al., 2024)  
Emerging Technologies in Infrastructure Monitoring  
Machine Learning (ML) Applications: ML techniques are being widely adopted for anomaly detection in  
photovoltaic (PV) systems and short-reach optical networks, enhancing the precision, reliability, and  
interpretability of fault diagnosis (Shao et al., 2024).  
• Federated Learning: For large, geographically distributed deployments such as BharatNet, federated learning  
enables collaborative, data-driven monitoring while avoiding centralized storage of raw data, thus safeguarding  
data privacy and minimizing communication and bandwidth overhead( Larsson, 2023).  
Research Gaps Identified  
Despite extensive research in individual domains, several gaps remain:  
Lack of Integrated Site Diagnosis: Most studies focused and considered solar systems and telecom hardware  
as separate entities, without taking them as combined frameworks that assess solar, battery, and networking  
equipment (ONT/CCU) collectively(Hagen & Andersen, 2024)..  
Limited Rural Field Data: While hypothetical models exist, there is a insufficiency of empirical studies  
focusing on the real-time reliability of telecom field infrastructure specifically in rural or semi-urban  
geographies(Larsson, 2023).  
Visualization for Field Support: There is a significant lack of user-friendly, concurrent decision-support  
dashboards specifically designed for field technicians to quickly identify multi-component faults at remote sites.  
(Hagen & Andersen, 2024).  
RESEARCH METHODOLOGY  
The research methodology for this study is designed to assess the operational consistency of rural broadband  
infrastructure using a data-driven, analytical approach. It utilizes a blend of descriptive, diagnostic, and  
predictive methods to assess the performance of critical hardware assets.  
Research Design  
The study employs a descriptive and analytical research design.  
Descriptive Analysis: Used to analyze the current availability and health status of equipment across 113 Gram  
Panchayats (GPs).  
www.ijltemas.in  
Page 1175  
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XII, December 2025  
Diagnostic Analysis: Focused on identifying the origin reason of network downtime, such as equipment failure  
clusters or power supply gaps.  
Predictive Analysis: Leveraged to institute a framework for anticipating future maintenance needs based on  
existing equipment conditions.  
Data Collection and Sources  
The study exclusively depends on secondary data collected from the internal field operations and maintenance  
records of Xentric Integrated Solutions Pvt. Ltd..  
Primary Source: Regular site inspection reports, infrastructure logs, and maintenance records from the  
BharatNet rural broadband network.  
Parameters Tracked: Data includes technical conditions of Optical Network Terminals (ONT), Central Control  
Units (CCU), battery backups, solar panels, earthing systems, building stability, and Electricity Board (EB)  
power availability.  
Verification: Data was initially gathered by field engineers and later verified against the company’s Network  
Operations Center (NOC) support logs.  
Sampling Design  
The research uses a structured sampling plan to ensure representation of both geographic and technical factors  
across the Puducherry district.  
Sampling Method: A non-probability purposive sampling method was adopted. This method was  
chosen because the study required existing technical inspection data rather than subjective opinions from  
public users.  
Sampling Unit: Each individual Gram Panchayat (GP) network site served as the primary unit of  
observation.  
Sample Size: The total sample consists of 113 Gram Panchayat network sites across three  
administrative blocks: Ariyankuppam, Villianur, and Karaikal.  
Sampling Frame: The frame comprises 113 commissioned and active sites documented in the  
company’s asset maintenance database  
Data Preprocessing and Analysis Plan  
To make sure data accuracy, the dataset underwent thorough cleaning and preparation before analysis:  
1. Cleaning: elimination of duplicate entries, correction of location names, and standardization of  
categorical statuses (e.g., "Healthy/Faulty").  
2. Tools: Data cleaning and initial analysis were performed using Microsoft Excel and Google Sheets,  
while complex statistical modeling was conducted in SPSS.  
3. Statistical Techniques: The study utilized frequency distribution, cross-tabulation, correlation analysis  
(to identify interdependencies between ONT and CCU systems), and logistic regression to predict GP  
operational status based on equipment health.  
www.ijltemas.in  
Page 1176  
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XII, December 2025  
Data Analysis and Interpretations  
This section presents the data analysis and interpretations derived from the study of 113 Gram Panchayats (GPs)  
in Puducherry. The analysis focuses on the operational status and health of core hardware assets, including  
Optical Network Terminals (ONT), Central Control Units (CCU), battery systems, and solar panels  
General Operational Status  
The study categorized the operational status of GPs as "UP" (functional) or "DOWN" (non-functional) based on  
their connectivity and equipment health.  
Table 1. Categorization of operational status  
GP Name ONT Status CCU Status Battery Status Solar Panel Status Building Condition  
GP 101  
GP 102  
Healthy  
Healthy  
Healthy  
Faulty  
Missing  
Healthy  
Healthy  
Missing  
Good  
Needs Repair  
Table 2. Descriptive statistics  
Category Component Available (Frequency) Available (%) Faulty (Frequency) Faulty (%)  
Networking ONT  
Networking CCU  
98  
91  
78  
67  
86.73%  
80.53%  
69.03%  
59.29%  
2
1
4
6
2.04%  
1.10%  
5.13%  
8.96%  
Power  
Power  
Battery  
Solar Panel  
A majority (87.61%) of the Gram Panchayats are operational. However, the 12.39% designated as "DOWN"  
represent significant service gaps, primarily due to equipment failure or a total lack of critical hardware  
Equipment Health and Availability  
Based on table 2 the Networking Hardware such as ONTs and CCUs show the highest availability and healthiest  
operational status, representing a healthy primary network infrastructure. In case of Power Backup Gaps: There  
is significant drop-off in the availability of batteries (69%) and solar panels (59%). This absence of alternative  
energy makes nearly 40% of the sites susceptible to grid power fluctuations. Regards to Fault Distribution: Solar  
panels have the highest fault rate (8.96%), followed by batteries (5.13%), indicative of renewable energy systems  
as the primary maintenance blockage.  
Regional Comparative Analysis (Block-wise)  
Table 3. Comparative analysis –Block wise  
Block Name  
Karaikal  
Total GPs GPs "UP" GPs "DOWN" Efficiency % Avg. Health Score (1-5)  
28  
42  
27  
38  
1
4
96.40%  
90.50%  
3.29  
3.07  
Ariyankuppam  
www.ijltemas.in  
Page 1177  
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XII, December 2025  
Villianur  
43  
34  
9
79.10%  
2.21  
From the above given table it infers The Karaikal block is the most efficient, likely due to superior maintenance  
practices and energy stability. on the other hand, Villianur faces the most significant challenges, requiring  
targeted infrastructure audits and more frequent maintenance checks.  
Correlation and Predictive Analysis  
Table 4. Correlation and Predictive Analysis  
Variable A  
Variable B  
Correlation Coefficient (r)  
ONT Availability  
EB Power  
CCU Availability  
Battery Health  
Site Uptime  
0.796  
0.65  
Solar Panel  
0.734  
The correlation performed between variables ONT Availability & CCU Availability, EB Power& Battery Health  
and Solar Panel with Site Uptime shows positive correlation with each other. This successfully recognizes 100%  
of operational sites based on current equipment health data. This proves that real-time monitoring of hardware  
health is a highly accurate predictor of service availability, supporting the move toward predictive maintenance  
models over reactive ones  
DISCUSSION:  
Challenges and Contemporary Issues  
Infrastructure Reliability and Service Uptime  
The major findings specify that while 87.6% of Gram Panchayats are operational ("UP"), a critical 12.4%  
are non-functional ("DOWN") due to specific hardware failures or a complete lack of essential equipment.  
The data recommends that network availability is not exclusively about having fiber optics; it is heavily reliant  
on the "healthy" status of onsite components like ONTs and CCUs.  
Networking Consistency: Optical Network Terminals (ONT) and Central Control Units (CCU) were the most  
constantly available and functional units across the surveyed sites.  
System Co-dependence: Correlation analysis revealed a strong relationship between ONT and CCU (r =  
0.796) and CCU and Battery (r = 0.734). This implies that the failure of one power or networking component  
often leads to a "cascade failure" of the entire broadband node.  
The Energy Resilience Gap  
A significant subject in the discussion is the weakness of rural sites to power fluctuations. While 81.42% of GPs  
have regular grid (EB) power, approximately 18.58% lack this basic utility, mandating a complete reliance  
on alternative energy.  
Missing Backups: In 30.97% of locations, batteries are not available, and 40.71% lack solar panels.  
Maintenance Bottlenecks: Solar panels and earthing systems were found to have the highest fault rates among  
all equipment, identifying them as the primary technical challenges to stable connectivity.  
www.ijltemas.in  
Page 1178  
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XII, December 2025  
Regional Disparities in Maintenance  
The research highlights a "maintenance segregate" across different administrative blocks in Puducherry:  
Karaikal: confirmed 100% operational efficiency with a high equipment health mean score of 3.29, suggesting  
superior local maintenance practices and stable energy.  
Villianur: Recorded the lowest mean health score (2.21) and the highest non-operational rate (23.26%), pointing  
toward environmental challenges or a lack of localized technical support.  
Theoretical Implications  
Our study supports Technology Acceptance Model (TAM) and Resource Dependence Theory representing the  
aspect of Perceived Usefulness For rural users, the usefulness of the network is directly tied to uninterrupted  
access. If batteries fail and the grid goes down, the perceived utility of the BharatNet project diminishes. Study  
also proves the need for Moving from reactive to Condition-Based Maintenance—using real-time field data to  
predict failures—is essential to prolonging equipment life and reducing the service disruptions that frustrate rural  
users.  
Management Lessons Learned  
From a strategic perspective, the study emphasizes that infrastructure management requires empowered field  
technicians and standardized reporting rather than just installation; The success of the Karaikal block  
witnessed that localized patrolling and timely device replacement are the most effective drivers of reliability  
Limitations  
The study has several limitations:  
Regulatory Disparities: inconsistency in policy implementation and vendor payments across local  
administrations often delays project milestones.  
Maintenance Gaps: A scarcity of trained technical personnel and lack of replacement cycles lead to prolonged  
equipment downtime.  
Digital Divide: Connectivity gaps in isolated or seaside areas are exacerbated by low digital literacy and unstable  
power.  
Key Findings  
Most technical failures are emerge from earthing systems and solar panel faults rather than primary  
networking equipment.  
There is a strong association between the health of power backup systems and overall network uptime.  
Regional maintenance receptiveness varies significantly across administrative blocks.  
Strategic Recommendations  
Transition to Predictive Maintenance: Integrate AI-driven models and IoT-enabled monitoring to detect  
abnormalities before service disruption.  
Hybrid Energy Management: adopt renewable energy integration to reduce dependence on unstable power  
grids.  
www.ijltemas.in  
Page 1179  
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XII, December 2025  
Digital Asset Tracking: Adopt QR-based tracking and digital inventory systems to improve accountability and  
field harmonization.  
Capacity Building: Establish standard operating procedures (SOPs) and continuous training programs for field  
technicians to improve response times.  
CONCLUSION  
The research underscores that digital inclusion in rural Puducherry is not merely dependent on installation but  
on the reliability and sustainability of physical infrastructure. While the BharatNet initiative has established a  
strong foundation, achieving a truly digitally-empowered India requires a strategic shift toward data-driven  
maintenance and energy-resilient ecosystems  
REFERENCES  
1. Hagen, D., & Andersen, S. (2024). Asset Management, Condition Monitoring and Digital Twin: Damage  
Detection and Virtual Inspection.  
2. Larsson, J. (2023). Data Use and Data Needs in Critical Infrastructure Risk Analysis.  
3. Pfeffer, J., & Salancik, G. R. (1978). The External Control of Organizations: A Resource  
Perspective. Harper & Row.  
Dependence  
4. Perriment, R., et al. (2023). Lead-acid battery lifetime extension in solar home systems.  
5. Rogers, E. M. (2003). Diffusion of Innovations (5th ed.). Free Press  
6. Shao, C., et al. (2024). Machine learning in short-reach optical systems: a comprehensive survey.  
7. Wei, Y. (2023). Prediction of State of Health of Lithium-Ion Battery Using Health Index Informed  
Attention Model.  
www.ijltemas.in  
Page 1180