Well just went through the paper.. you are right.
What he has done is that he has compared results for various modulation schemes ( NRZ, RZ , CSRZ , MDRZ ) for various data rates.. he has used different amplifiers as he has mentioned in each case and calculated for the results.. he has done comparison on bases of amplifiers but as for as selection of bandwidth for multiplexer is concerned , he has not mentioned any reason why he has done that.. i don’t know what possibly is the logic behind this. I hope someone helps on this.
Thanks & regards