Adaptive Clock Skew Estimation with Interactive Multi-Model Kalman Filters for Sensor Networks
Authors:
- Z. YangZYZ. Yang
- J. PanJPJ. Pan
- L. CaiLCL. Cai
- 1 May 2010
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- No. 15503607,p. 1-5
Abstract
Clock synchronization is a fundamental issue in communication networks and distributed systems, and clock skew is the inherent cause for clock desynchronization. Clock skew estimation is essential to improve the efficiency and reduce the overhead of clock synchronization schemes, and it is especially beneficial for resource-constrained devices such as sensor nodes in dynamic environments. According to the measurement, clock skew is environment sensitive, and no existing clock skew estimation schemes can accurately capture such dynamic behaviors. In this paper, we investigate a general clock synchronization problem with variable clock skews and propose a new skew estimation model based on a hybrid approach to characterizing the dynamic of clock skews. To estimate the time-varying clock state vector, we employ the Interactive Multi-Model (IMM) Kalman filter, which can make soft decisions by combining results from different models. Extensive simulations have been conducted to demonstrate the effectiveness of the proposed adaptive clock skew estimation algorithm, which achieves a better performance with moderate computational complexity.Keywords
This publication has 10 references indexed in Scilit:
- ACES2008
- Vector Kalman Filter Using Multiple Parents for Time Synchronization in Multi-Hop Sensor Networks2008
- Adaptive Kalman Filter for Time Synchronization over Packet-Switched Networks: An Heuristic Approach2007
- Hot or not: Revealing hidden services by their clock skew2006
- Evaluation of Kalman filtering for network time keepingIEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control, 2005
- The flooding time synchronization protocol2004
- Robust synchronization of software clocks across the internet2004
- Clock synchronization algorithms for network measurements2003
- Internet time synchronization: the network time protocolIEEE Transactions on Communications, 1991
- The interacting multiple model algorithm for systems with Markovian switching coefficientsIEEE Transactions on Automatic Control, 1988