Integrated Fault Localization and Field Service Logistics

This dissertation addresses the need for more efficient field service operations in the telecommunications industry, focusing on reducing the kilometers driven by technician workforces while maintaining high service quality. The work is conducted as part of the project, GREENFORCE, an initiative aimed at leveraging advanced methodologies to enhance operational efficiency and reduce the environmental impact of field service operations, with this dissertation’s primary
focus being achieving these goals through optimized routing and scheduling using Operations Research.

The telecommunications sector faces unique challenges in managing its field service operations, particularly in last-mile networks, which connect customers to the main grid. These networks are characterized by their dynamic nature,
with fault and maintenance tasks arising unpredictably every day, often with uncertain task durations and locations. Incomplete or outdated information about network topologies further complicates routing and task planning, leading
to inefficiencies and unnecessary driving. These issues contribute to what the dissertation refers to as micro-routing—the additional driving and time spent during task completion that is often underestimated during the planning phase.

This dissertation addresses both micro-routing challenges and optimization problems related to improving the underlying data foundation for routing and scheduling decisions. By enhancing data accuracy and availability—particularly with
respect to network topology and fault locations—this work aims to support more efficient planning and resource allocation in telecommunications field service operations.

The research outcomes are structured around five papers that collectively demonstrate the potential for both theoretical and practical advancements. The first paper introduces a novel method for reconstructing incomplete network topolo-gies using phylogenetic principles. This method produces accurate reconstructions of the cabling topology of last-mile networks, reducing the uncertainty technicians face when responding to fault and maintenance tasks. The second paper builds on this by proposing a framework for extracting valuable insights from time series data collected by customer modems, enabling more informed approaches to fault detection and network maintenance

The third paper introduces a scoring system for identifying the most likely locations of network faults, providing technicians with a decision-making tool that guides them to high-probability fault locations. This not only reduces driving time but also enhances the accuracy of fault resolution. The fourth paper integrates the results of the first three and presents a fast, recursive algorithm that computes the optimal search strategy for technicians in the context of fault resolution with stochastic resolution locations, demonstrating how improved data and micro-routing considerations can lead to an 80% reduction in combined service and driving time. Finally, the fifth paper explores the topic of strategic workforce investments, offering a fast and practical framework for deciding how to hire an efficient field service workforce based on detailed operational scenarios, further demonstrating how an improved data foundation can inform long-term decision-making.