arXiv subjects

Sohini Roy

Publications and source records attributed to Sohini Roy.

12 recordsLinked to original sources

arXiv2026

A Machine Learning Surrogate for Component Criticality Ranking in Interdependent Power-Communication Networks

Cyber-physical power systems are vulnerable to cascading failures caused by interdependencies between power and communication infrastructures. Because evaluating large N-k contingency sets with a high-fidelity simulator is computationally expensive, this paper develops a machine-learning surrogate using the previously published Modified Implicative Interdependency Model (MIIM) as the ground-truth cascade simulator. The surrogate predicts contingency severity from leakage-free structural features and derives an association-based component-criticality ranking for resilience screening. On the IEEE 118-bus system, Gradient Boosting achieves a held-out Spearman correlation of 0.849 for contingency-severity ranking. Using five-fold out-of-fold predictions, the resulting component ranking achieves a Spearman correlation of 0.838 with the MIIM-derived ranking and closely approaches the observed cross-sample reproducibility level. Feature-ablation results show that inter-layer dependency features drive most of the surrogate's advantage, while end-to-end screening is approximately 158x faster than direct MIIM evaluation. The results support a two-stage workflow in which the surrogate screens candidate contingencies and components, and MIIM provides selective verification rather than directly identifying optimal hardening actions.

cs.LG↗

arXiv2026

Persistent Spatio-Temporal Outage Hotspot Detection for Infrastructure Resilience Planning

Extreme weather events are producing persistent geographic patterns of power-grid disruption across the United States, yet outage hotspot detection and infrastructure cascade modeling are often studied separately. This paper presents a data-driven geospatial framework that links persistent outage vulnerability with downstream cascade impact in interdependent power-communication networks. Using a national outage dataset from 2015-2023, we introduce the Hotspot Persistence Index (HPI), a severity-aware metric for identifying counties that repeatedly emerge as outage hotspots over time. We then apply a multi-scale DBSCAN refinement procedure to convert persistent county-level hotspots into geographically interpretable regional failure scenarios characterized by recurrence, severity, and spatial extent. To evaluate their system-level relevance, these empirically derived scenarios are injected into the Modified Implicative Interdependency Model (MIIM), which captures cascading behavior across coupled power and communication layers. Results show that three persistent regional clusters account for 54.4% of total HPI-weighted cascade impact, while communication-layer entities fail at 2.5X the rate of power buses under high-persistence scenarios. HPI-guided hardening reprioritizes protection candidates relative to a degree- and betweenness-centrality baseline, identifying high-value buses that topology-only rankings overlook. These results demonstrate how persistent geospatial outage patterns can support targeted and empirically grounded infrastructure resilience planning.

cs.NI↗

arXiv2026

Data-Calibrated Climate Outage Stress Testing for Joint Power-Communication Networks

Climate-driven outages pose a growing threat to cyber-physical power system (CPPS) resilience, particularly as modern grids increasingly rely on communication, sensing, and control infrastructure for situational awareness and coordinated response. Empirical outage studies and interdependency-aware resilience analyses are often studied separately, limiting our understanding of how observed climate-risk patterns translate into cascading degradation across coupled cyber and physical layers. This paper presents a data-calibrated stress-testing framework for joint power-communication networks. Using EAGLE-I outage records from 2015-2023, we characterize national climate-related outage patterns and identify severe-risk contexts defined by outage duration and customer impact and associated with event type, season, and geography. EAGLE-I customer-impact percentiles are log-normalized and scaled into representative stress intensities for spatially clustered benchmark scenarios on an IEEE 118-bus power system with an overlaid communication network. The Modified Implicative Interdependency Model (MIIM) is used to simulate cross-layer cascade propagation and quantify post-event operability, resilience gaps, and affected cyber-physical entities. Results show that even the highest-severity benchmark scenario degrades post-cascade operability to approximately 60%, producing a resilience gap roughly twice that of the inland baseline. The findings suggest that resilience assessment based only on outage statistics may miss bounded but meaningful amplification effects in interdependent power-communication infrastructure. The proposed framework provides an initial step toward data-calibrated, interdependency-aware resilience assessment for extreme-weather-affected CPPS.

stat.AP↗

arXiv2026

Critical Transit Infrastructure in Smart Cities and Urban Air Quality: A Multi-City Seasonal Comparison of Ridership and PM2.5

Public transit is a critical component of urban mobility and equity, yet mobility and air-quality linkages are rarely operationalized in reproducible smart-city analytics workflows. This study develops a transparent, multi-source monitoring dataset that integrates agency-reported transit ridership with ambient fine particulate matter PM2.5 from the U.S. EPA Air Quality System (AQS) for four U.S. metropolitan areas - New York City, Chicago, Las Vegas, and Phoenix, using two seasonal snapshots (March and October 2024). We harmonize heterogeneous ridership feeds (daily and stop-level) to monthly system totals and pair them with monthly mean PM2.5 , reporting both absolute and per-capita metrics to enable cross-city comparability. Results show pronounced structural differences in transit scale and intensity, with consistent seasonal shifts in both ridership and PM2.5 that vary by urban context. A set of lightweight regression specifications is used as a descriptive sensitivity analysis, indicating that apparent mobility-PM2.5 relationships are not uniform across cities or seasons and are strongly shaped by baseline city effects. Overall, the paper positions integrated mobility and environment monitoring as a practical smart-city capability, offering a scalable framework for tracking infrastructure utilization alongside exposure-relevant air-quality indicators to support sustainable communities and public-health-aware urban resilience.

physics.soc-ph↗

arXiv2021

A Leader-Follower Game Theoretic Approach to Arrest Cascading Failure in Smart Grid

The Smart Grid System (SGS) is a joint network comprising the power and the communication network. In this paper, the underlying intra-and-interdependencies between entities for a given SGS is captured using a dependency model called Modified Implicative Interdependency Model (MIIM) [1]. Given an integer K, the K-contingency list problem gives the list of K-most critical entities, failure of which maximizes the network damage at the current time. The problem being NP complete [2] and owing to the higher running time of the given Integer Linear Programming (ILP) based solution [3], a much faster heuristic solution to generate an event driven self-updating K-contingency list [4] is also given in this paper. Based on the contingency lists obtained from both the solutions, this paper proposes an adaptive entity hardening technique based on a leader-follower game theoretic approach that arrests the cascading failure of entities in the SGS after an initial failure of entities. The validation of the work is done by comparing the contingency lists using both types of solutions, obtained for different K values using the MIIM model on a smart grid of IEEE 14-Bus system with that obtained by simulating the smart grid using a co-simulation system formed by MATPOWER and Java Network Simulator (JNS). The K-contingency list obtained for a smart grid of IEEE 14-Bus system also indicate that the network damage predicted by both the ILP based solution and heuristic solution using MIIM are more realistic compared to that obtained using another dependency model called Implicative Interdependency Model (IIM) [2]. Advantage of using the MIIM based heuristic solution is also shown in this paper when larger SGS of IEEE 118-Bus is considered. Finally, it is shown how the adaptive hardening helps in improving the network performance.

cs.GT↗

arXiv2021

Identification and Mitigation of False Data Injection using Multi State Implicative Interdependency Model (MSIIM) for Smart Grid

Smart grid monitoring, automation and control will completely rely on PMU based sensor data soon. Accordingly, a high throughput, low latency Information and Communication Technology (ICT) infrastructure should be opted in this regard. Due to the low cost, low power profile, dynamic nature, improved accuracy and scalability, wireless sensor networks (WSNs) can be a good choice. Yet, the efficiency of a WSN depends a lot on the network design and the routing technique. In this paper a new design of the ICT network for smart grid using WSN is proposed. In order to understand the interactions between different entities, detect their operational levels, design the routing scheme and identify false data injection by particular ICT entities, a new model of interdependency called the Multi State Implicative Interdependency Model (MSIIM) is proposed in this paper, which is an updated version of the Modified Implicative Interdependency Model (MIIM) [1]. MSIIM considers the data dependency and operational accuracy of entities together with structural and functional dependencies between them. A multi-path secure routing technique is also proposed in this paper which relies on the MSIIM model for its functioning. Simulation results prove that MSIIM based False Data Injection (FDI) detection and mitigation works better and faster than existing methods.

cs.CR↗

arXiv2021

A Self-Updating K-Contingency List for Smart Grid System

A reliable decision making by the operator in a smart grid is contingent upon correct analysis of intra-and-interdependencies between its entities and also on accurate identification of the most critical entities at a given point of time. A measurement based self-updating contingency list can provide real-time information to the operator about current system condition which can help the operator to take the required action. In this paper, the underlying intra-and-interdependencies between entities for a given power-communication network is captured using a dependency model called Modified Implicative Interdependency Model (MIIM) [1]. Given an integer K, the event-driven self-updating contingency list problem gives the list of K-most critical entities, failure of which maximizes the network damage at the current time. Owing to the problem being NP complete, a fast heuristic method to generate a real-time contingency list using system measurements is provided here. The validation of the work is done by comparing the contingency list obtained for different K values using the MIIM model on a smart grid of IEEE 14-Bus system with that obtained by simulating the smart grid using a co-simulation system formed by MATPOWER and Java Network Simulator (JNS). The results also indicate that the network damage predicted by both the ILP based solution [2] and the proposed heuristic solution using MIIM are more realistic compared to that obtained using another dependency model called Implicative Interdependency Model (IIM) [3].

eess.SY↗

arXiv2020

Identification of the K-most Vulnerable Entities in a Smart Grid System

A smart grid system can be considered as a multi-layered network with power network in one layer and communication network in the other. The entities in both the layers exhibit complex intra-and-interdependencies between them. A reliable decision making by the smart grid operator is contingent upon correct analysis of such dependencies between its entities and also on accurate identification of the most critical entities in the system. The Modified Implicative Interdependency Model (MIIM) [1] successfully captures such dependencies using multi-valued Boolean Logic based equations called Interdependency Relations (IDRs) after most of the existing models made failed attempts in doing that. In this paper, for any given integer K, this model is used to identify the K-most vulnerable entities in a smart grid, failure of which can maximize the network damage. Owing to the problem being NP complete, an Integer Linear Programming (ILP) based solution is given here. Validation of the model [1] and the results of the ILP based solution is done by simulating a smart grid system of IEEE 14-Bus using MATPOWER and Java Network Simulator (JNS). Simulation results prove that not only the model MIIM [1] is correct but also it can predict the network damage for failure of K-most vulnerable entities more accurately than its predecessor Implicative Interdependency Model (IIM) [2].

cs.NI↗

arXiv2020

SSGMT: A Secure Smart Grid Monitoring Technique

Critical infrastructure systems like power grid require an improved critical in-formation infrastructure (CII) that can not only help in monitoring of the crit-ical entities but also take part in failure analysis and self-healing. Efficient designing of a CII is challenging as each kind of communication technology has its own advantages and disadvantages. Wired networks are highly scala-ble and secure, but they are neither cost effective nor dynamic in nature. Wireless communication technologies on the other hand are easy to deploy, low cost etc. but they are vulnerable to cyber-attacks. In order to optimize cost, power consumption, dynamic nature, accuracy and scalability a hybrid communication network is designed in this paper where a portion of the communication network is built using wireless sensor networks (WSN) and the rest is a wired network of fiber optic channels. To offer seamless opera-tion of the hybrid communication network and provide security a Secure Smart Grid Monitoring Technique (SSGMT) is also proposed. The perfor-mance of the proposed hybrid CII for the generation and transmission sys-tem of power grid coupled with the SSGMT during different cyber-attacks is tested using NS2 simulator. The simulation results show that the SSGMT for a joint power communication network of IEEE 118-Bus system performs better than the prevailing wireless CIIs like Lo-ADI and Modified AODV.

cs.NI↗

arXiv2020

Secure and Energy Efficient Remote Monitoring Technique (SERMT) for Smart Grid

Monitoring and automation of the critical infrastructures like the power grid is improvised by the support of an efficient and secure communication net-work. Due to the low cost, low power profile, dynamic nature, improved ac-curacy and scalability, wireless sensor networks (WSN) became an attractive choice for the Information and Communication Technology (ICT) system of the smart grid. However, the energy efficiency and security of WSN depends highly on the network design and routing scheme. In this paper, a WSN based Secure and Energy Efficient Remote Monitoring Technique (SERMT) is proposed by demonstrating a WSN based ICT network model for perva-sive monitoring of the generation and transmission part of the power network in a smart grid system. The performance of the proposed network model de-signed for a smart grid of IEEE 118-Bus system coupled with the secure routing technique is tested during cyber-attacks by means of NS2 and the simulation results indicate that it performs better than existing smart grid monitoring methods like Lo-ADI [1] with respect to packet drop count and throughput.

cs.NI↗

arXiv2020

A New Model to Analyze Power and Communication System Intra-and-Inter Dependencies

The reliable and resilient operation of the smart grid necessitates a clear understanding of the intra-and-inter dependencies of its power and communication systems. This understanding can only be achieved by accurately depicting the interactions between the different components of these two systems. This paper presents a model, called modified implicative interdependency model (MIIM), for capturing these interactions. Data obtained from a power utility in the U.S. Southwest is used to ensure the validity of the model. The performance of the model for a specific power system application namely, state estimation, is demonstrated using the IEEE 118-bus system. The results indicate that the proposed model is more accurate than its predecessor, the implicative interdependency model (IIM) [1], in predicting the system state in case of failures in the power and/or communication systems.

cs.NI↗

arXiv2018

Health Monitoring of Critical Power System Equipments using Identifying Codes

High voltage power transformers are one of the most critical equipments in the electric power grid. A sudden failure of a power transformer can significantly disrupt bulk power delivery. Before a transformer reaches its critical failure state, there are indicators which, if monitored periodically, can alert an operator that the transformer is heading towards a failure. One of the indicators is the signal to noise ratio (SNR) of the voltage and current signals in substations located in the vicinity of the transformer. During normal operations, the width of the SNR band is small. However, when the transformer heads towards a failure, the widths of the bands increase, reaching their maximum just before the failure actually occurs. This change in width of the SNR can be observed by sensors, such as phasor measurement units (PMUs) located nearby. Identifying Code is a mathematical tool that enables one to uniquely identify one or more {\em objects of interest}, by generating a unique signature corresponding to those objects, which can then be detected by a sensor. In this paper, we first describe how Identifying Code can be utilized for detecting failure of power transformers. Then, we apply this technique to determine the fewest number of sensors needed to uniquely identify failing transformers in different test systems.

eess.SP↗