TITLE
Building Efficient Wireless Sensor Networks with Low-Level Naming

AUTHOR
Heidemann

DATE
2001

LOCATION
SOSP

The authors replace topologically-assigned IP addresses with names
which are either based on capabilities or on geographic location.
They are examining the tradeoffs in radio-based sensor networks where,
even though the nodes are fairly weak computationally, the cost of
computation outweighs the costs in sending data over the network.
Thus, if nodes can perform aggregation or other computation which will
reduce the amount of data transfer, this is seen as a win.  One highly
motivating example taken from another context was: "[other
researchers] observe that 3000 instructions could be executed for the
same energy cost of sending a bit 100m by radio."  On difficulty with
their motivation was that they appeared to be saying that current
sensor network designs did not scale to millions of nodes but their
example of tracking animals in the forest did not really evoke a
millions-of-nodes usage.  In their design, nodes are not individually
addressible; they are communicate hop-by-hop.  Their choice of MAC
protocols dominate energy measurements due to the fact that they are
always listening for MAC information; they would like to introduce
more energy-efficient protocols like PAMAS and TDMA in the future.
Strangely, their relatively small network of about 12 nodes
experienced pockets of congestion and had difficulty communicating
quite often.

Their mechanism is very similar to Declarative Routing from MIT's
Lincoln Labs because the two groups collaborated.