Optimal consensus‐based clock synchronisation algorithm in wireless sensor network by selective averaging

Abstract

Wireless sensor networks (WSNs) have received much attention in recent years because of its broad area of applications. In the same breadth, it also faces many challenges. Time synchronisation is one of those fundamental and important challenges faced by WSN. Several approaches have been proposed in recent years for time synchronisation. Most of the approaches are based on synchronising to a reference (root) node's time by considering a hierarchical backbone for the network. However, this approach seems to be not purely distributed, higher accumulated synchronisation error for farthest node from the root and subjected to the root node failure problem. In this study, a fully distributed time synchronisation algorithm is proposed which synchronises all nodes' time to a consensus value, that is, an average of all nodes' initial clock values. The algorithm exploits a selective pairwise averaging approach with a linear message complexity of O(n), to achieve faster convergence and better synchronisation accuracy than the existing random pairwise average time synchronisation protocol. Simulation results show that almost 50% improvement is achieved in convergence speed, measured in terms of number of iterations, and nearly 30% improvement in total synchronisation error.

Energy Consumption Analysis of Consensus Time Synchronization Algorithms for Wireless Sensor Networks

“…So, to have a close resemblance with the realistic environment, the skew is generated in the specified range using random uniform distribution. To have a fair comparison between SATS [14], ATSP [7] and CCS [8], the clock offsets are generated using random uniform distribution between 0 and 1 which is same as specified in the above algorithms [7,8,14]. The interval for one iteration, which denotes an upper bound for maximum oscillation period, is set to 10 seconds.…”

“…Using the above discussed models, the following section gives a detail energy consumption analysis of three recent CTS algorithms, namely, SATS [14], ATSP [7], and CCS [8]. Though, the authors in their papers have used different name for type of synchronization messages, for the sake of simplicity and uniformity, generic names for type of messages are used for all the algorithms in this paper.…”

“…Mathematical analysis in [14] shows that for a network of 'n ′ nodes, SATS algorithm exchanges (4n − 2) number of messages per iteration which is comparatively higher than ATSP algorithm which exchanges 3n number of messages per iteration and CCS algorithm which exchanges (3n − 2) messages.…”

“…Recently, to develop fully distributed and internal time synchronization mechanism, Consensus Time Synchronization (CTS) method has gained much attention [7,8,9,10,11,12,13,14]. CTS method is essentially based on distributed average consensus principle which states that all the nodes in a network can converge to a consensus or synchronized state after a finite number of iterations, by communicating and performing averaging only with the neighbors.…”

“…The work in [14] proposes a selective averaging scheme for faster consensus convergence and synchronization accuracy. In [13], the authors have proposed a cluster consensus based time synchronization method.…”