Efficient Time Synchronization in a Wireless Sensor Network by Adaptive Value Tracking

Gradient Descent Algorithm Inspired Adaptive Time Synchronization in Wireless Sensor Networks

“…In order to get closer to the optimal value, it is necessary to decelerate the adjustment. It is worth to mention that this algorithm is inspired from [4] in which the multipliers 2 and 1/3 are shown to be good values in terms of convergence performance. Figure (2) presents the evolution of the logical clock frequency of node u with constant and adaptive step sizes during a numerical simulation.…”

“…In order to evaluate their multi-hop synchronization performances, we prepared identical testbed setup as presented in [2], [3], [4] and preferred a line topology of 20 sensor nodes to evaluate scalability and adaptivity. We used 7.37 MHz quartz oscillator on the MICAz board as the clock source for the timer used for timing measurements.…”

“…Least-squares regression is a common technique employed by these protocols to adjust the offsets and the frequencies of the logical clocks simultaneously [1], [2]. However, previous studies [3], [4] revealed that this technique is heavyweight in terms of processing and memory overhead, and simultaneous offset and frequency estimation gives rise to poor performance scalability. Other proposed techniques, such Kasım Sinan Yıldırım is with Embedded Software Group, TU Delft, Delft, The Netherlands (k.s.yildirim@tudelft.nl).…”

“…To overcome aforementioned problems in practice, two iterative methods are proposed [4], [8]. AVTS protocol [4] employs an efficient and adaptive search technique that adjusts the relative frequency value of the clocks by observing the sign of the clock skew.…”

“…AVTS protocol [4] employs an efficient and adaptive search technique that adjusts the relative frequency value of the clocks by observing the sign of the clock skew. The logical clock offset is adjusted independently from its frequency by adding the observed skew to the logical clock directly.…”