Risk-Adjusted Fuzzy Dijkstra Algorithm for Shortest-Path Optimization Under Uncertain Network Conditions: A Computational Study
DOI: https://doi.org/10.5281/zenodo.21236054
Abstract
Real network parameters such as travel time, latency, cost, and capacity are rarely known with complete precision. Classical shortest-path algorithms require crisp edge weights and can therefore select routes that are nominally attractive but vulnerable to congestion or measurement uncertainty. This paper develops a risk-adjusted fuzzy Dijkstra algorithm in which every edge cost is represented by a triangular fuzzy number and converted to an additive ranking score that combines possibilistic central tendency with uncertainty spread. The additive structure preserves the computational advantages of Dijkstra’s algorithm while permitting a decision maker to control risk aversion through a single parameter, λ. A reproducible computational experiment was conducted on connected random geometric networks containing 50, 100, 250, and 500 nodes. The calibration stage selected λ = 0.5, after which the proposed method was evaluated against modal-weight and centroid-weight Dijkstra baselines on 432 independent source-destination cases and 500 uncertainty scenarios per selected route. Across all cases, the proposed algorithm reduced the mean 95th-percentile travel time by 1.26% and conditional value-at-risk at 95% by 1.50% relative to modal Dijkstra. The probability of exceeding a route’s nominal time by more than 20% decreased from 4.20% to 1.92%, representing a 54.36% relative reduction. Compared with centroid Dijkstra, the proposed method produced a small 0.19% increase in average travel time but reduced CVaR95 by 0.32% and exceedance probability by 29.43%. The average computation time remained below 0.5 ms per query in the tested networks. The findings show that a simple risk-adjusted fuzzy ranking can improve route robustness without sacrificing scalability, providing a mathematically transparent foundation for later applications in transportation, communication, energy, and supply-chain networks.
How to Cite
Pardeep Malhan, Manisha, Rachan Khandelwal, "Risk-Adjusted Fuzzy Dijkstra Algorithm for Shortest-Path Optimization Under Uncertain Network Conditions: A Computational Study", Vol. 4, Issue 2, 29-05-2026, pp. 35-49. DOI: https://doi.org/10.5281/zenodo.21236054
