Derivative superposition LNA

Iam designing my LNA using derivative superposition .but it is not giving improved results.wat may be the reason.any design procedure to find out optimum biasing for the transistors.

Will you please elaborate your method of "Derivative superposition".

two transistors are connectedin parellel one is operated in strong inversion (vgs far from vth) other is operated in weak inversion (vgs<vth) so that effective gm'' cancel each other

I guess this method focus on gm cancellation only, what about the individual noise contributions?
Check they contribute minimum NF by both transistors at their bias condition. If you have the model then put it in simulator and check the NF.

But how the noise is related to linearity?

Transistor NF is based on device physics and Linear amplifier is complex subject.
So am help less other than giving you the idea to solve the problem.