Sphere Decoding v.s. Maximum Likelihood Detection

[gama]

sphere decoder is a ml maximum likelihood

Hi,

Do you think the performance of SD (sphere decoder) is just always
equals to ML (maximum likelihood) detection ?

regards,
gama

 

[gks83]

Sphere decoder is a ML decoder for specific MIMO applications...
therefore all sphere decoders are ML.

 

[changfa]

SD is just a way trying to implement ML (which is the optimal). Depending on the radius of the sphere, SD can approach ML, but no better.

 

[gama]

Somebody told me that:
SD can approach ML only if the radius is large enough to cover
every lattice points which skewed by channel and shifted by noise.

If we just make sure that we will find at least one point each searching time
in SD, that means we will not cover the whole lattice points but one point
at least. In such a case, the performance of SD do always equal to ML?
In other words, the lattice points which we got between SD and ML do always
be the same?


regards,
gama