you notice in the akabori reaction a 2 fold excess of aldehyde is used?
Well it has a carbonyl group that forms a shiff's base with the amino acid, this shiff's base may abstract a proton from the carboxylic acid causing it to form a leaving group.
This mechanism applies to alpha amino acids (those seperated by one carbon)
In this case it's a beta amino acid so I don't know if the same holds true.
Looking at DMSO you see the carbonyl group in the middle?
with it's high B.P., polarity and carbonyl group capable of forming the shiff's base I would react the "serine" in the free acid form with the proton dangling from that carboxyl group, as opposed to it's sodium or potassium salt as this effect can not take place in this instance.
who knows it might just work!
for instance you could just use the classical akabori reaction by heating benzaldehyde and alanine in DMSO.
this would serve both as a solvent and a decarboxylation promoter, we'll see.'
allow me to correct myself, It would seem wiser to first form the alanine bezaldehyde serine derivative and then decarboxylate using a high boiling polar solvent with a sulfonyl (not carbonyl) function (like dmso)
I fear using dmso as a sovent in a straight akabori condensation/decarboxylation reaction might interfere.
The advantage here is that you can form this shiff's base intermediate which will facilitate the decarboxylation of the COOH+ group by proton abstraction to the shiff's base.
but the interatomic distance in a alpha VS beta amino acid (like the one you are referring too) may cause this not to work.