DocumentCode :
116289
Title :
Robotic surveillance and Markov chains with minimal first passage time
Author :
Agharkar, Pushkarini ; Patel, Rushabh ; Bullo, Francesco
Author_Institution :
Center for Control, Dynamical Syst. & Comput, Univ. of California at Santa Barbara, Santa Barbara, CA, USA
fYear :
2014
fDate :
15-17 Dec. 2014
Firstpage :
6603
Lastpage :
6608
Abstract :
We propose stochastic surveillance strategies for quickest detection of anomalies in discrete network environments. Our surveillance strategy is determined by optimizing the mean first passage time also known as the Kemeny constant of a Markov chain. We generalize the notion of the Kemeny constant to environments with heterogeneous travel and service times, denote this generalization as the weighted Kemeny constant, and characterize its properties. For reversible Markov chains, we show that both the Kemeny constant and its heterogeneous counterpart can be formulated as convex optimization problems and, moreover, can be expressed as semidefinite programs (SDPs). We numerically illustrate the proposed design: compared with other well-known Markov chains, the performance of our Kemeny-based strategies are always better and in many cases substantially so.
Keywords :
Markov processes; convex programming; graph theory; mobile robots; surveillance; SDP; anomaly detection; convex optimization problems; discrete network environments; heterogeneous service time; heterogeneous travel time; mean-first-passage time optimization; minimal-first-passage time; reversible Markov chains; robotic surveillance; semidefinite programs; stochastic surveillance strategies; weighted Kemeny constant; Algorithm design and analysis; Markov processes; Minimization; Robots; Surveillance; Symmetric matrices; Vectors;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Decision and Control (CDC), 2014 IEEE 53rd Annual Conference on
Conference_Location :
Los Angeles, CA
Print_ISBN :
978-1-4799-7746-8
Type :
conf
DOI :
10.1109/CDC.2014.7040425
Filename :
7040425
Link To Document :
بازگشت