Module 3: Understanding Scalability of Distributed Estimation Algorithms

Faculty Contact: Joao Hespanha

Concepts: Distributed estimation

Research Areas:

Datasets: Noisy measurements of relative position obtained from radio signal strength and acoustic data.

Abstract: Distributed estimation and control arises in numerous multi-agent large-scale systems: the localization of a network of sensors in a common coordinate system based on noisy measurements of the relative position between sensors, the time synchronization of a group of computing/sensing nodes that are part of a communication network, the motion coordination of mobile agents based on feedback from the agents’ relative positions, etc. In all these applications the performance that can be obtained depends crucially on the topology of the network that expresses the interactions between the agents. Algebraic graph topology can provides quantitative measures of convergence speed for (linear) distributed control and estimation algorithms and circuit theory can provide fundamental bounds on the accuracy that can be obtained from such algorithms. This projects aims at familiarizing students with these concepts and introducing them to open problems in the area.