Distributed Kalman filtering based on consensus strategies
read more
Citations
Gossip Algorithms for Distributed Signal Processing
Diffusion Strategies for Distributed Kalman Filtering and Smoothing
Distributed model predictive control: A tutorial review and future research directions
Adaptation, Learning, and Optimization Over Networks
Kalman-Consensus Filter : Optimality, stability, and performance
References
Consensus problems in networks of agents with switching topology and time-delays
The Theory of Matrices
Coordination of groups of mobile autonomous agents using nearest neighbor rules
Related Papers (5)
Frequently Asked Questions (7)
Q2. What are some subclasses of normal matrices?
Relevant subclasses of normal matrices are, for instance, Abelian Cayley matrices [2], circulant matrices and symmetric matrices.
Q3. What is the consequence of Lemma 3.1?
An immediate consequence of Lemma 3.1 is that, when the communication graph is undirected, the minimum of the cost function J is reached by symmetric matrices.
Q4. how many times are the eigenvalues counted?
In order to calculate the optimal gain the authors have to solvearg min `∈(0,1)qN1− (1− `)2 + r` 2N−1∑h=0|λh|2 1− (1− `)2|λh|2 (15)4Multiple eigenvalues are counted as many times as their algebraic multiplicity.
Q5. what is the simplest way to find a nonnormal Q?
May 17, 2007 DRAFT14Remark 3: It is important to note that normality plays a fundamental role in the previous lemma which cannot be generalized to stochastic matrices Q. In fact, it is easy to find a nonnormal Q for which the symmetric part Qsym gives a larger cost index.
Q6. What is the simplest example of the theorem?
The previous theorem shows also that for q << r the optimizing Q, while being consistentwith the communication graph, has to minimize the number of unitary eigenvalues.
Q7. What is the simplest way to prove that limi (qopt(?
Consider now a sequence {`opt(m)}∞m=0 and let ρ̃ be such that ρ̄ < ρ̃ < 1. Let us introduce a matrix Q̃ such that σ(Q̃) = {1, ρ̃, . . . , ρ̃}, that is Q̃ has N − 1 eigenvalues equal to ρ̃.