Image Processing Reference
In-Depth Information
Idle listening represents the major source of energy waste in WSNs. A possible solution to reduce it is
the low-power listening (LPL) [Hil]. Such an approach operates at the PHY layer introducing a duty
cycle. he basic idea is to shift the cost from the receiver to the transmitter by increasing the length
of the preamble. This allows the receiver to periodically turn on the radio to sample for incoming
data. If the receiver detects a preamble, it will continue listening until the start symbol arrives and
the message can be properly received, otherwise it will turn off the radio and go to sleep until the next
sample. Energy saving with LPL entails a slight degradation in both latency and throughput. LPL can
be applied to any MAC protocol provided that the radio-switching time is short.
8.1.2 Performance Metrics for MAC Protocols Used in WSNs
The objective of MAC protocols for WSNs is to efficiently regulate the medium access, in terms of
both performance and energy efficiency, so a relevant metric is the channel access success probability
provided to the upper layers. Other metrics that are widely used to assess the effectiveness of a MAC
protocol are throughput and delivery ratio. However, in some WSN applications not all the traffic has
to be necessarily delivered in order to achieve reliable event detection, and redundant frames may be
droppedforthesakeofthenetworkperformanceandenergyeiciency.Asaresult,forsuchcases
neither the throughput nor the delivery ratio is suitable performance metrics. In [Vur] the goodput
is used, which is defined as the ratio of the packets received by the sink and the number of packets
that are actually generated by the nodes, thus taking into account early discarding of redundant data
as well. Another parameter of interest in these contexts is the event detection reliability expressed,
for example, in terms of distortion.
Among the metrics which are representative of the energy consumption or energy saving, the
most widely adopted are the average energy consumption per node, used, for example, in [Dam]
[Hal] [Vur], and the average power consumption of nodes [Enz]. As the energy consumption
is directly related with the duty cycle of nodes or with the average sleep time, these parameters are
also suitable metrics for an energy-efficient MAC protocol. In [Raj], the percentage of sleep time
is used together with the average length of the sleep interval. As the sleep and wake-up transitions
of the radio are both time- and energy-consuming, the average length of the sleep interval is useful
to estimate the amount of radio-mode switching and so the actual energy saving. In MAC protocols
which provide for dynamic or adaptive nodes' duty cycles (e.g., T-MAC), a metric that reflects the
energy consumption of every node is the distribution of active times, used e.g., in [Hal].
There are WSNs in which timeliness is an important requirement. In these networks, the channel
access delay is also an important metric. However, as this delay depends on the nodes' duty cycles,
it usually clashes with energy consumption. As a result, it is usually more interesting to evaluate the
trade-off between energy consumption and latency as, for example, in [Pol], rather than the delay
value alone.
8.2 Overview on Energy-Efficient MAC Protocols for WSNs
MAC protocols for WSNs can be classified into two basic classes: contention-based or schedule-
based. In contention-based protocols, the sender has to continuously sense the channel to assess
when it is idle, collisions may occur and a back-off procedure has to be run before retransmitting after
a collision. As said before, all these activities are energy-consuming. On the contrary, in schedule-
basedprotocols,datatransfersarescheduledinadvanceandtherecipientsknowexactlywhento
switch on the radio and when they can safely go to sleep.
Both these protocol classes have their pros and cons. On the one hand, contention-based pro-
tocols are energy-consuming due to collisions, idle listening, and overhearing. However, they
have some valuable benefits, such as, low implementation complexity, scalability, and flexibility
