This study has shown through simulation that SE-CMA performs better than CMA 2-2
for a communications system that has a BPSK source. This hypothesis could not
be proven analytically due to the complicated cost expression for SE-CMA.
However, I have included the expression for the cost, gradient, and hessian
for SE-CMA. I also presented two methods of choosing (the Godard
radius) and it appears that neither method is always superior. Future
research could focus on: 1) Proving analytically the claim that SE-CMA is
superior to CMA 2-2 when the source is BPSK; and 2) Deriving an expression
for
that always provides the lowest error.