The average power of this signal is it's variance which is its' variance which is var(x)=1.3333 in our assumed case.
Once you normalize x(n) by the square root of the power (i.e. standard deviation) you get the same signal of unit power, that is
x1 = sqrt(1/var(x)) * x
Now that you've got your signal with power 1 Watt (or mW depending on the units of the vector x(n) ), you can generate your signal with a desired power P (which in your case is tuned to be 150mW) by
x2= sqrt(P) * x1
Finally x2 is the same signal as x but with power P.
Well since x1 is now unit less, if you'd leave P as 150 then the total power is measured in mW but if you convert P to 150*10^-3 then the total power is measured in Watt.
You can generate a Gaussian signal via randn(1,10000) in MATLAB. This signal show you variance equal to 1 (1W power). To find 150W, you have to scale your generated data by sqrt(150).