Therefore,
it is important to find smart and e?±cient ways to allocate MAC, routing
and location identifiers to sensor nodes, without relying on strict per-node
addressing, except when there is a need to communicate with an individual
sensor node.
As an example, consider Geographic Adaptive Fidelity (GAF) with Manhattan
routing [55]. Here, a network of sensor nodes is divided into cells in
a grid-like fashion, with each cell being assigned a location identifier. Thus,
every node belonging to a particular cell shares the same location identifier.
These location identifiers are used in conjunction with a geographic routing
protocol (Manhattan routing), where packets are routed from one cell to another
regardless of the actual identity of the nodes in both the cells. Thus,
all nodes in a particular cell also share the same routing address. Another
example can be found in AIMRP [26], where nodes use randomly chosen addresses
on a per-packet transmission attempt basis, rather than using their
hardware MAC identifiers, to distinguish between one another at the MAC
layer. The length of these random addresses can be much smaller than the
length of MAC identifiers, and these random addresses can be shared over
time by all the nodes.
Routing Overheads
There are generally two types of routing mechanisms that a network can use:
proactive and reactive. In proactive routing, the status of routing paths is
periodically updated through control message exchanges between neighboring
14
nodes.
Pages:
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58