Structural Health Monitoring using Wireless Sensor Networks - Mobile and Pervasive Computing in Construction - page 220

sampling, it is highly desirable if the sensor nodes can be awakened in a fast and
synchronous way. Generally speaking, the existing approaches to wake up the
wireless sensor nodes in a network can be largely divided into two categories:
schedule based and event triggered.
Schedule-based wakeup
For a WSN-based SHM system for a long-term monitoring purposes, the most
widely adopted power saving strategy is that sensor nodes are working and sleeping
in a pre-defined schedule (Caffrey et al ., 2005; Krishnamurthy et al ., 2005; Olund
et al ., 2007). When the internal timers fire, sensor nodes are awakened from sleep
mode. Sensor nodes then perform their routing work, such as sampling, data
processing, and transmission, possibly after a time synchronization procedure, and
then go to sleep until the next round starts. The duration in the sleep mode can be
fixed or adjusted. For a systemoperating in a very low duty cycle, lifetimes can reach
the order of a few months or even years.
Although simple to implement, the schedule-based approach has significant
drawbacks. This approach limits the ability of the application users to initiate
network operations at random or if an event of interest occurs. To address this
problem, researchers at UIUC have developed a scheduling scheme called
“ SnoozeAlarm ” (Jang et al ., 2010). SnnozeAlarm provides extra flexibility for
the fixed scheduling. When wireless sensor nodes are in the SnoozeAlarm mode,
they wake up periodically, during which they can listen and receive instructions
from gateway or application users. A similar approach can be found in the CC1100/
CC2500 RF transceiver which offers wake-on-radio (WOR) functionality. This
functionality of the CC1100/CC2500 enables it to stay in a power-saving sleep state
and periodically wake up and listen for incoming packets without microcontroller
interaction (Syvertsen and Namtvedt, 2009).
The schemes mentioned above significantly increase the flexibility of schedule-
based wakeup. However, the dilemma of energy consumption and wakeup delay
always exists. A shorter period of sleep time decreases the wakeup delay but
inevitably causes greater energy consumption, while a larger period of sleep time is
energy efficient but has larger wakeup delay. A SHM system using the above scheme
generally does not fit in a time critical application where extreme events with short
durations need to be captured, such as earthquakes or bridge overloads. For
example, in the Jindo Bridge (Jang et al ., 2010), it takes 1-5 minutes to wake up all
the sensor nodes in the SnoozeAlarm mode.
Event-based wakeup
To address the limitation of schedule-based wakeup, another strategy, event-
triggered wakeup is proposed. Compared with the schedule-based wakeup,
event-triggered wakeup can put the sensor nodes into sleep state as long as possible
and wake them up immediately if required. Trigger events can be, for example,
changes in vibration or external radio transmission. To use this strategy, each
sensor node should be equipped with a special hardware component designed to
sense the event of interest. The component itself should be low powered or even