03/24/2026
By Danielle Fretwell

The Francis College of Engineering, Department of Electrical and Computer Engineering, invites you to attend a Master's Thesis defense by Vina Dang on: "Improving Resilience and Efficiency in Drone Network Applications using Greedy Algorithms."

Candidate Name: Vina Dang
Degree: Master’s
Defense Date: Monday, March 30, 2026
Time: 1 - 2:30 p.m.
Location: Ball Hall 302

Committee:

  • Advisor: Lewis Tseng, Associate Professor, Electrical & Computer Engineering, University of Massachusetts Lowell
  • Chunxiao (Tricia) Chigan, Professor, Electrical & Computer Engineering, University of Massachusetts Lowell
  • Paul Robinette, Associate Chair/Associate Professor, Electrical & Computer Engineering, University of Massachusetts Lowell

Brief Abstract:
This thesis shows how greedy algorithms can improve the resilience and efficiency of drone network applications operating under uncertainty, limited communication, and constrained deployment budgets. It develops a common probabilistic graph-based planning framework for two representative settings: distributed multi-drone search and drone-assisted vehicular networking. In the multi-drone search setting, the environment is modeled as a Terrain Search Probability Graph, where candidate regions are connected to targets through detection probabilities that reflect terrain conditions and sensing spillover. In the vehicular networking setting, a Vehicular Coverage Probability Graph captures the probability that a UAV positioned at a feasible aerial region can successfully serve road segments, intersections, or traffic hotspots. In both cases, the core objective is to assign a limited set of regions to K drones so as to maximize the expected number of uniquely detected or covered targets.

The thesis shows that these planning problems are NP-hard, motivating approximation methods that remain practical for edge-based deployment. It therefore focuses on greedy algorithms that exploit monotone sub-modular structure, yielding simple and interpretable planners with a provable (1−1/e) approximation guarantee. To support more realistic probabilistic evaluation, the work also adopts a live-edge Monte Carlo formulation that estimates marginal gains through repeated sampling and enables uncertainty reporting through confidence intervals.

Experiments are conducted on synthetic benchmark search environments and on a road-aware urban V2X scenario with blockage and no-fly constraints. Across the search benchmarks, greedy planning consistently outperforms random selection, closely tracks exact optimal solutions on tractable instances, and remains highly competitive on larger clustered terrains, where diminishing returns naturally emerge as more drones are added. In the V2X setting, greedy placement likewise remains near-optimal while substantially outperforming random feasible placement across multiple traffic patterns and urban channel conditions. The thesis also clarifies the role of synchronization in
edge-enabled drone coordination: under a fixed probability graph, increasing synchronization frequency alone does not change planning outcomes unless new observations update the underlying belief map.

Overall, this work demonstrates that greedy planning over probabilistic target maps provides a scalable, efficient, and resilient foundation for real-world drone network applications, including disaster response, environmental monitoring, and on-demand vehicular connectivity support.