Combinatorial bounds and scaling laws for noise amplification in networks

Motivated by the problem of designing and analyzing clock synchronization protocols for large-scale networks, we provide combinatorial bounds on the mean square disagreement of synchronization dynamics subject to additive noise. While previous work has used eigenvalues to analyze the mean square disagreement when the synchronization dynamics is governed by a normal matrix, in the non-normal case eigenvalues cease to be an adequate measure. We first show that the so-called 2-norm coefficient of ergodicity used in the study of inhomogeneous Markov chains can be used to bound mean square disagreement even when the underlying dynamics is governed by a non-normal matrix. Our main result then demonstrates that the 2-norm coefficient of ergodicity has a natural combinatorial interpretation as a combined measure of matrix non-normality and graph connectivity. We apply this result to yield new performance guarantees for distributed clock synchronization: we show that in a simple (possibly irregular) random graph model, the natural clock synchronization scheme wherein neighboring nodes average their clocks is order-optimal: it drives the expected clock offsets between any two nodes to at most a constant, independently of the total number of nodes.

[1]  P.R. Kumar,et al.  Distributed Clock Synchronization over Wireless Networks: Algorithms and Analysis , 2006, Proceedings of the 45th IEEE Conference on Decision and Control.

[2]  Yang Liu,et al.  Consensus problem of high‐order multi‐agent systems with external disturbances: An H∞ analysis approach , 2010 .

[3]  P. Diaconis,et al.  Geometric Bounds for Eigenvalues of Markov Chains , 1991 .

[4]  Y. Bar-Ness,et al.  Distributed synchronization in wireless networks , 2008, IEEE Signal Processing Magazine.

[5]  E. Seneta.  Non-negative Matrices and Markov Chains , 2008 .

[6]  Ruggero Carli,et al.  Networked clock synchronization based on second order linear consensus algorithms , 2010, 49th IEEE Conference on Decision and Control (CDC).

[7]  Panganamala Ramana Kumar,et al.  Fundamental Limits on Synchronizing Clocks Over Networks , 2011, IEEE Transactions on Automatic Control.

[8]  A. Rhodius,et al.  The maximal value for coefficients of ergodicity , 1988 .

[9]  Luca Schenato,et al.  Average TimeSynch: A consensus-based protocol for clock synchronization in wireless sensor networks , 2011, Autom..

[10]  Ilse C. F. Ipsen,et al.  Ergodicity Coefficients Defined by Vector Norms , 2011, SIAM J. Matrix Anal. Appl..

[11]  Stephen P. Boyd,et al.  Distributed average consensus with least-mean-square deviation , 2007, J. Parallel Distributed Comput..

[12]  Béla Bollobás,et al.  The Isoperimetric Number of Random Regular Graphs , 1988, Eur. J. Comb..

[13]  Naomi Ehrich Leonard,et al.  Rearranging trees for robust consensus , 2011, IEEE Conference on Decision and Control and European Control Conference.

[14]  Ruggero Carli,et al.  Optimal Synchronization for Networks of Noisy Double Integrators , 2011, IEEE Transactions on Automatic Control.

[15]  Vivek S. Borkar,et al.  A New Distributed Time Synchronization Protocol for Multihop Wireless Networks , 2006, Proceedings of the 45th IEEE Conference on Decision and Control.

[16]  Qun Li,et al.  Global Clock Synchronization in Sensor Networks , 2006, IEEE Trans. Computers.

[17]  Fikret Sivrikaya,et al.  Time synchronization in sensor networks: a survey , 2004, IEEE Network.

[18]  Stephen P. Boyd,et al.  Fast linear iterations for distributed averaging , 2003, 42nd IEEE International Conference on Decision and Control (IEEE Cat. No.03CH37475).

[19]  I Poulakakis,et al.  Coupled stochastic differential equations and collective decision making in the Two-Alternative Forced-Choice task , 2010, Proceedings of the 2010 American Control Conference.

[20]  Luca Schenato,et al.  A distributed consensus protocol for clock synchronization in wireless sensor network , 2007, 2007 46th IEEE Conference on Decision and Control.

[21]  Naomi Ehrich Leonard,et al.  Robustness of aggregation in networked dynamical systems , 2009, 2009 Second International Conference on Robot Communication and Coordination.

[22]  Alessandro Chiuso,et al.  A PI Consensus Controller for Networked Clocks Synchronization , 2008 .

[23]  Naomi Ehrich Leonard,et al.  Robustness of noisy consensus dynamics with directed communication , 2010, Proceedings of the 2010 American Control Conference.

[24]  João Pedro Hespanha,et al.  Distributed Optimal Estimation from Relative Measurements for Localization and Time Synchronization , 2006, DCOSS.

[25]  Bruce W. Suter,et al.  Novel Clock Phase Offset and Skew Estimation Using Two-Way Timing Message Exchanges for Wireless Sensor Networks , 2007, IEEE Transactions on Communications.

[26]  E. Seneta.  Coefficients of ergodicity: structure and applications , 1979, Advances in Applied Probability.