ODiN: Enhancing Resilience of Disaster Networks through Regression Inspired Optimized Routing

Reliable communication in disaster-hit areas is crucial for effective rescue and relief operations. This has encouraged the emergence of innovative, infrastructure-less, and ad hoc communication frameworks with the aim to keep Disaster Networks (DiNets) operational. However, emerging DiNet frameworks are still in their nascent stages. A vital challenge in ensuring seamless communication in harsh post-disaster scenarios is the design of robust routing algorithms. In this work, we elucidate the various constraints placed by post-disaster scenarios upon the design of routing mechanisms. We then implement an AIIJoyn based DiNet prototype and gather real-time network data in an experimental site that resembles a disaster-hit zone. We subject gathered empirical data to Regression Analysis, create network models, and derive relationships between network parameters. The real-time regression equations of network parameter relationships serve as constraints in a high-level Mixed Integer Nonlinear Programming model named ODiN. The objective of ODiN is to offer optimal solutions to routing and next-hop relay selection in post-disaster scenarios. We demonstrate a significant reduction in convergence time while maintaining high accuracy through the use of ODiN.