Protocols for Secure Communication in Wireless Sensor Networks

170 Chapter 6. Integrity-Preserving Communications
3,4,6,7,8,9,10,11
7
4
8
2,4,5,6
2
1
3
3,5,6,7,8,
9,10,11 2,6,8,9,10,11
2,5,8,9,
10,11
5
6
9
(a)
3,5,6
10
11
7
4,7,8
4
8
2,4,5
2
9
5
1
5,6,9,10,11
Figure 6.4: Canvas message overhead on the link level. Left: Authentication codes for
broadcast-based flooding. Right: Authentication codes for a tree structure. Using a tree is
much more economical
in a subtree will receive the message. Nodes that become a member of multiple
subtrees, which is likely to happen for many nodes, will receive the message
multiple times. As root, such a node will construct its own subtree as the union
of all subtrees it is part of. This ensures that all nodes will eventually receive
the message.
Compared to simple flooding, the number of authentication codes that are
transmitted is significantly reduced. This can be seen by looking at a simple
example. Figure 6.4 shows a network where nodes are labelled with numbers.
Node number 1 is the source and floods a message to its neighbours. It adds
authentication codes for its 1-hop and 2-hop neighbours (the links are labelled
with the target nodes of these codes; the message flow is from top to bottom). In
Figure 6.4(a), simple flooding is shown. In addition to the authentication code
for its immediate neighbour, a node adds authentication codes for all of its 2hop
neighbours to each outgoing link. This is necessary since simple flooding
is not aware of the topology beyond single-hop neighbourhood. Figure 6.4(b)
shows a similar graph with some edges missing. Since a node constructs a
tree before it forwards a message, not all links will be used to transmit the
message. Nodes are now aware of their 2-hop neighbourhood and thus they
can limit the set of authentication codes that have to be sent over a certain
link. Thereby, a significant amount of data can be saved. This is confirmed
by a large-scale simulation (500 nodes on a 1000 × 1000 plane with varying
transmission range and therefore varying number of neighbours per node). As
Figure 6.5 shows, the difference between simple flooding and the structured
approach is approximately 10-fold.
(b)
3,6
3
10
5,6,11
6
11