Overview of Time Synchronization Issues in Sensor Networks - Networked Embedded Systems - page 92

Receivers

Transmitters

B

C

A

Translation nodes

FIGURE . Illustration of the RBS.

a two-way message handsake as shown in Figure . to synchronize and correct any clock ofset. he
objective of TSHL is to address long propagation delay in acoustic sensor networks.
As for type () of the timing techniques, the RBS provides an instantaneous time synchronization
amongasetofreceiversthatarewithinthereferencebroadcastofthetransmitter.hetransmitter
broadcasts m reference packets. Each of the receivers that are within the broadcast range records
the time-of-arrival of the reference packets. Afterwards, the receivers communicate with each other
to determine the offsets. To provide multihop synchronization, it is proposed to use nodes that are
receiving two or more reference broadcasts from different transmitters as translation nodes. These
translation nodes are used to translate the time between different broadcast domains.
As shown in Figure ., nodes A, B, and C are the transmitter, receiver, and translation nodes,
respectively. he transmitter nodes broadcast their timing messages, and the receiver nodes receive
these messages. Afterwards, the receiver nodes synchronize with each other. he sensor nodes that
are within the broadcast regions of both transmitter nodes A are the translation nodes. When an
event occurs, a message describing the event with a time stamp is translated by the translation nodes
when the message is routed back to the sink. Although this time synchronization service is tunable
and lightweight, there may not be translation nodes on the route path that the message is relayed.
As a result, services may not be available on some routes. In addition, this protocol is not suitable
for medium access scheme such as TDMA since the clocks of all the nodes in the network are not
adjusted to a common time.
PalChaudhuri et al. [] extends the work on RBS and provides a probabilistic bound on the accu-
racy of clock synchronization. he synchronization error is assumed to be Gaussian distributed, and
it is possible to calculate the probability of synchronization error given the maximum allowed error.
For instance, given the maximum error between two synchronizing nodes is є max ,theprobabilityof
synchronization with an error є ≤ є max is calculated as follows:

√ n є max
σ

P (∣ є ∣≤ є max )= e rf ( ) (.)

where
e rf is the error function
n is the number of messages needed to achieve the probability P
σ is the standard deviation of the Gaussian distribution