Kaleh-Vallet Algorithm Comparison
Last Modified:
The simulation run above used 10,000 trials, for SNR 0 to 20 in steps of 5,
with the deterministic and Gaussian EM algorithms initialized using
the true symbol values. This cause those algorithms to
perform better than the known channel case, so I don't think
that we should do this since we'd like to use the known
channel case as a lower bound.
This could replace Figure 1 in the current EM/ISI paper draft, and
includes the results from the Kaleh-Vallet algorithm for SNR 0 to 10.