By Topic

Compressive topology identification of interconnected dynamic systems via Clustered Orthogonal Matching Pursuit

Sign In

Cookies must be enabled to login.After enabling cookies , please use refresh or reload or ctrl+f5 on the browser for the login options.

Formats Non-Member Member
$31 $13
Learn how you can qualify for the best price for this item!
Become an IEEE Member or Subscribe to
IEEE Xplore for exclusive pricing!
close button

puzzle piece

IEEE membership options for an individual and IEEE Xplore subscriptions for an organization offer the most affordable access to essential journal articles, conference papers, standards, eBooks, and eLearning courses.

Learn more about:

IEEE membership

IEEE Xplore subscriptions

3 Author(s)
Sanandaji, B.M. ; Dept. of Electr. Eng. & Comput. Sci., Colorado Sch. of Mines, Golden, CO, USA ; Vincent, T.L. ; Wakin, M.B.

In this paper, we consider topology identification of large-scale interconnected dynamical systems. The system topology under study has the structure of a directed graph. Each edge of the directed network graph represents a Finite Impulse Response (FIR) filter with a possible transport delay. Each node is a summer, whose inputs are the signals from the incoming edges, while the output of the summer is sent to outgoing edges. Edges of the graph can be of different unknown orders and delays. Both the graph topology and the FIR filters and delays that make up the edges are unknown. We aim to do the topology identification from the smallest possible number of node observations when there is limited data available and for this reason, we call this problem Compressive Topology Identification (CTI). Inspired by Compressive Sensing (CS) which is a recent paradigm in signal processing for sparse signal recovery, we show that in cases where network interconnections are suitably sparse (i.e., the network contains sufficiently few links), it is possible to perfectly identify the network topology along with the filter orders and delays from small numbers of node observations, even though this leaves an apparently ill-conditioned identification problem. The main technical novelty of our approach is in casting the identification problem as the recovery of a clustered-sparse signal z ∈ RN from the measurements b = Az ∈ RM with M <; N, where the measurement matrix A is a block-concatenation of Toeplitz matrices. To this end, we introduce a greedy algorithm called Clustered Orthogonal Matching Pursuit (COMP) that tackles the problem of recovering clustered-sparse signals from few measurements. In a clustered-sparse model, in contrast to block-sparse models, there is no prior knowledge of the locations or the sizes of the clusters. We discuss the COMP algorithm and support our discussions with simulations.

Published in:

Decision and Control and European Control Conference (CDC-ECC), 2011 50th IEEE Conference on

Date of Conference:

12-15 Dec. 2011