Image Processing Reference
In-Depth Information
must also be computed. It can be expressed as
d
(
S , D
)
E
(
S , D ,
(
N , p
))=
E
(
p
)⋅
R
(
S ,
(
N , p
))⋅
(.)
d
(
S , D
)−
d
(
N , D
)
where
E
(
p
)
is the energy spent to send the packet at power level p
R
represents the expected number of retransmissions per hop before S success-
fully delivers a packet to N when transmitting at power level p (as computed by the delay
estimator)
the term d
(
S ,( N , p
))
(
S , D
)/(
d
(
S , D
)−
d
(
N , D
))
estimates the expected number of hops to the
destination
If the forwarding policy does not find a choice that meets the required velocity, the neighborhood
manager comes into action, searching for new forwarding choices that could meet the requirement.
Two techniques are implemented to search for new forwarding choices: power adaptation and
neighbor discovery. Power adaptation finds the most energy-efficient forwarding choice that still
meets the velocity requirements. When velocity requirements cannot be met for a packet, the power
adaptation scheme increases the transmission power to improve the velocity provided by neighbors
already in the neighbor table. Conversely, when the velocity requirements are satisfied, the power
adaptation scheme decreases the transmission power to improve the energy efficiency and network
capacity.
Neighbor discovery is performed when no eligible forwarding choice is found with power adap-
tation. he aim is to identify nodes that meet the velocity requirement. Neighbor discovery starts by
broadcasting a request to route (RTR) packets at power level p .Anode N hears the RTR and replies.
Upon receiving the reply, RPAR inserts the new forwarding choice ( N , p
in its neighbor table.
As the number of all combinations ( N , p ) could be very high, the neighborhood manager only
maintains the most frequently used entries to save space in memory-constrained devices.
This protocol addresses both real-time performance in terms of reduced deadline miss ratio
and energy saving through efficient transmission power adaptation. Compared with SPEED-like
protocols, both energy consumption and deadline miss ratio are reduced. However, in the presence
of holes, Euclidean distance is a poor approximation of the path length, so QoS may be affected.
As explained in [Chi], the RPAR power adaptation policy suffers from pathological behavior
when a node is congested, as explained below. Due to high contention, there is a high collision
probability, so a node needs a large number of retries before successfully transmitting a packet. As
a consequence, RPAR increases the transmission power, thus worsening the situation. In [Chi]
some solutions to tackle this problem are outlined. One solution is to adopt MAC layer approaches
enabling a node to distinguish between a packet being lost due to collisions or to poor link quality.
Such feedback from the MAC layer would prevent RPAR from performing any useless and harmful
increase in the transmission power. he second solution is the use of a congestion control protocol
for WSNs able to detect a node congestion, thus enabling RPAR to stop increasing the transmission
power. his solution would not only prevent the power control from worsening the congestion, but
wouldalsoenablethecongestionprotocoltomitigatetheproblem.
Finally, this protocol does not handle sleep schedules, so energy consumption may still be high as
compared to cluster-based approaches such as LEACH.
)
7.8 Topology Control Protocols for Energy-Efficient Routing
Topology control protocols are a slightly different approach to saving energy than standard routing
protocols, as they do not directly operate data forwarding. hese protocols run at a lower level of the
 
Search WWH ::




Custom Search