Image Processing Reference
In-Depth Information
itself to the sink. As this approach does not need to maintain explicit path information for each packet
or flow, it scales well with the number of data flows. However, before using this algorithm the cost
field needs to be established.
Establishment of the cost field is started by the sink node, which broadcasts an ADV packet with
a cost. Each node
N
that receives an ADV from a neighbor
M
computesthecostfromitselftothe
sink using this message, i.e.,
L
M
C
N
,
M
,where
L
M
is the cost for the node
M
and
C
N
,
M
is the cost
from
M
to
N
. If it is lower than the current
L
N
value, the
L
N
valueisupdatedandaback-oftimeris
scheduled (or reset). When the timer expires, an ADV message containing the
L
N
value is broadcast.
This protocol is efficient as long as the network is static. However, if there is a change in the
topology, i.e., a node runs out of energy, the cost field has to be reestablished. Fault tolerance is low,
because when a node fails many sources will not be able to forward data to the sink until the cost
field has been reestablished. Although this protocol could be used with a generic cost function, the
authors propose the use of energy as the path cost. So this protocol guarantees that the selected hop
is always the optimum in terms of depleted energy. But while this approach achieves optimal results
in terms of energy consumption, it does not optimize the network lifetime, as packets from the same
node always use the same path. Load and energy consumption are thus not balanced so the most
used node can quickly drain its battery.
+
7.4.2 Sequential Assigned Routing Protocol
The sequential assigned routing (SAR) protocol [Soh] is a table-driven, multi-path routing
protocol. As compared to classical ad-hoc protocols for MANET such as the AODV or the
temporally-ordered routing algorithm (TORA)
∗
, SAR tries to improve energy efficiency and fault
tolerance in low-mobility networks such as WSNs while maintaining a desired QoS. he main idea of
the SAR protocol is to route packets along multiple paths depending on the energy and QoS require-
ments. In addition to energy and required QoS, the SAR protocol also takes packet priority into
consideration.
In order to set up multiple paths from each node to the sink, multiple trees are built. Each tree is
rooted from a one-hop neighbor of the sink. Once the trees are built, most of the nodes will belong
to more than one tree, so they will have multiple disjoint paths from which the sink node can be
reached.
Each path is assigned two parameters by each node: a QoS metric and an energy resource, that is,
the number of packets to be sent through that path before the node energy runs out. Path selection
is performed by the source node, through computation of a weighted QoS metric, obtained from the
additive QoS metric multiplied by a weight coefficient depending on the packet priority. Notice that
by choosing the path, the one-hop neighbor of the sink is also selected. Furthermore, having differ-
ent paths to the sink, in the event of failure a node can perform failure recovery, thus also improving
fault tolerance. However, as the available energy of the nodes changes with time, it has to be peri-
odically updated. In general this protocol is efficient, but it needs a high overhead to build the trees,
maintain the tables, update the state, etc. he simulation results in [Soh] show that the SAR proto-
col features better performance than algorithms that always select the minimum-energy path regard-
less of network conditions. he approach used to achieve energy efficiency is to optimize a weighted
metric which takes the residual energy of nodes into account. his should achieve balanced energy
consumption and an improved network lifetime. To further increase energy saving, the authors
∗
AODV is a source-initiated routing protocol that performs on-demand route discovery and maintenance, and relies on
sequence numbers assigned by the destination to avoid loops. TORA is a link reversal algorithm that builds a directed
acyclic graph (DAG) rooted at the destination in which data flows from higher to lower nodes.
