580 California St., Suite 400
San Francisco, CA, 94104
Key research themes
1. How can decentralized algorithms enable self-healing and topology maintenance in distributed networks under node failures?
This research theme focuses on developing fully decentralized and scalable mechanisms by which distributed networks such as sensor, robotic, or cloud systems can autonomously restore or maintain their network topology after node failures without centralized control. It explores local information exchanges, adaptive data dissemination, and distributed reconstruction techniques that ensure the network quickly recovers its connectivity and functionality despite node removals or communication constraints. This matters as modern large-scale systems must sustain high availability and fault tolerance amid dynamic and unpredictable failures.
Key finding: Proposed and evaluated a two-stage fully decentralized self-healing system where local data collection about network topology is achieved via either Mobile Agents or the Trickle gossip protocol, followed by localized... Read more
Key finding: Analyzed robustness of a decentralized multi-robot connectivity maintenance control that estimates and preserves the algebraic connectivity (Fiedler eigenvalue) of the communication graph using local computations. The... Read more
Key finding: Developed DACYCLEM, a hybrid decentralized algorithm leveraging local interactions through connected dominating set construction combined with attractive-repulsive motion rules, to maintain network connectivity while... Read more
2. What are the trade-offs and control methods for connectivity maintenance in mobile robotic and multi-agent systems under communication range and delay constraints?
This theme investigates analytic and algorithmic approaches to maintain or restore communication connectivity in multi-agent and robotic networks where agents have limited transmission ranges, mobility-induced topology changes, and possibly communication delays. It studies local versus global connectivity preservation strategies, including k-hop alternative path checks and algebraic connectivity estimation, focusing on optimizing agents' freedom of movement while ensuring robust global connectivity. Insights into the dynamical system control design and delay effects are essential to guarantee sustained cooperation and information exchange in mobile robotic teams.
Key finding: Critically analyzed existing local and global connectivity maintenance methods, showing that local approaches restrict robot mobility by preserving all initial links, while global methods relying on power iteration for... Read more
Key finding: Compared local methods preserving all initial communication links with global methods preserving overall network connectivity by allowing selective link disconnections. Showed that global methods provide greater workspace and... Read more
Key finding: Designed a distributed controller for nonlinear multi-agent systems under fading wireless communication channels using a weighted Laplacian matrix derived from communication path loss factors, aiming to maximize the second... Read more
3. How can network restoration and availability be quantitatively modeled and optimized in communication networks considering partial link/node failures and component reliability?
This research domain addresses the mathematical and algorithmic modeling of network restoration mechanisms that restore failed links or nodes to maintain end-to-end connectivity and service availability in communication networks. It incorporates probabilistic component failure rates, partial link degradations, and simultaneous multi-failure scenarios into optimization frameworks to determine backup capacities, route adjustments, and maintenance schedules that minimize downtime and cost. Understanding availability at the component and system level supports service-level agreements and informs preventive maintenance, restoration routing, and flow adjustment strategies over complex, heterogeneous networks.
Key finding: Developed availability block diagrams incorporating component and fiber failure rates to model end-to-end connection availabilities in wavelength-division multiplexed optical networks considering transparent and opaque node... Read more
Key finding: Proposed the Flow-Adjustment Routing (FAR) strategy allowing nominal path-flows to be locally thinned or thickened based on a finite set of link availability states representing partial rather than just total link failures.... Read more
Key finding: Introduced a time-dependent restoration routing model scheduling crews to repair disrupted links within a constrained time horizon, explicitly incorporating crew routing times and spatial distribution of failures. Developed... Read more
Key finding: Formulated a new network availability measure capturing the fraction of users receiving full network services given component failure probabilities and network topology, bridging technical soundness with intuitive business... Read more