This example demonstrates a simple hypothetical layered radio. The network
contains two wireless nodes, and one of them is pinging the other.

The simulation produces a diagram that can be open in the IDE using the anf file.
The diagram shows the packet loss as a function of the distance between the two
nodes in a number of different configurations. There are three parameter axes,
and the simulation is run for all parameter combinations resulting in 24
different curves.

The first axis is the simulated level of detail in the physical layer. The
example is run with the following different level of details:

 - packet domain (bits are not computed, error model computes packet error)
 - bit domain (bits are computed, error model computes erroneous bits)
 - symbol domain (symbols are computed, error model computes erroneous symbols)

The different level of details are expected to produce very similar curves. One
exception is that the symbol domain curves are different from the curves of the
other domains when forward error correction is used. For modulations that have
more than one bits per symbol, the demodulation process results in a different
(non-independent and non-uniform) distribution of bit errors, which in turn is
decoded differently.

The second axis is about the forward error correction that is used in the bit
domain. The simulation is run with and without using forward error correction.
The used convolutional code has a code rate R = 1 / 2, and memory m = 1 with
generator matrix G(D) = (1 1 + D). When forward error correction is used it is
expected to increase the communication range. Note that in this example the net
bit rate is constant, therefore the gross bit rate is increased when using
forward error correction.

The third axis is about the modulation that is used in the symbol domain. The
simulation is run with the following modulations:

 - BPSK
 - QPSK
 - QAM-16
 - QAM-64

In general, modulations with more bits per symbol tend to have smaller
communication range.