Importance of Unconditional Stability in LNA design.

I'm trying to design a LNA in S band for 2.4 GHz Wlan applications , Input VSWR<1.5(probably antenna will be matched to 75 ohms) , NF<0.5 , S21=50dB.
I will cascade 3 or 4 transistors for S21. I was able to get 0.45 NF for the first transistor on AWR but the transistor is unstable after 4 GHz. K-factor goes below 1 after 4 GHz and ripples alot after.I can use another transistor and get stability till 7GHz aswell but still not Unconditionally Stable. What I was trying to ask is how important is the unconditional stability and till which frequency it should be Unconditionally Stable= K>1 ? 3rd harmonic or 15-20GHz or Ft?

As the stability factor depends on gain and reflection and reverse gain, it should be > 1 over the whole frequency spectrum. It's an indicator that your LNA is potentially able to oscillate at any frequency where k<1. In reality, your LNA will have no gain at 3rd harmonic or ft because of filters and so on, so k would be always greater than unity at these frequencies.

If 0<K<1, it means that the LNA "may" oscillate or not.Unconditionally stability is desired one but conditional stability may be accepted in such cases.Unconditional stability must be valid up to Fmax of the transistor so no combination of load and/or source load will drift the LNA into oscillation regime.
But just a remark, 50 dB Gp is too much for 2.4GHz, even-tough you don't see any problem in the simulation world, you will obviously live serious troubles when you put this LNA on a board or silicon.