We know that the square of a Rayleigh random variable has exponential distribution, i.e.,

Let the random variable $X$ have Rayleigh distribution with PDF $$f_X(x)=\frac{2x}{\alpha}e^{-x^2/{\alpha}}.$$

Then the random variable $Y=X^2$ has the PDF given by $$f_Y(y)=\frac{1}{\alpha}e^{-y/{\alpha}}.$$

For an exponentially distributed r.v. $Y$ with mean $\mathbb{E}[Y]=1$



Now, if the random variable $X$ has Rician distribution (unit power in direct and scattered paths), whose PDF is given by

$$f_X(x)=\frac{2x}{\alpha}\text{exp}\left(\frac{-(x^2+v^2)}{\alpha}\right)I_0\left(\frac{2xv}{\alpha}\right)$$ with $\frac{v^2}{\alpha}=1$ and $I_0(z)$ is the modified Bessel function of the first kind with order zero.

what is the PDF of $Y=X^2$?

And what is $\mathbb{E}[Y^{\delta}]$ when $\delta<1.$

Note: when $v^2=0$, $X$ has Rayleigh distribution.

  • $\begingroup$ I find it rather amusing that you write down explicitly the pdf of Rayleigh and exponential distributions (which are super-well known and not needed for your question), but you don't give us the pdf of the Rician distribution ;-) $\endgroup$ – Pierpaolo Vivo Feb 29 '16 at 10:04
  • $\begingroup$ @PierpaoloVivo, thanks for your concern. I have edited my question. $\endgroup$ – Dimitrios Mar 1 '16 at 2:37

Using a Dirac-delta method $$ f_Y(y)=\int_0^\infty dx\ \frac{2x}{\alpha}\text{exp}\left(\frac{-(x^2+v^2)}{\alpha}\right)I_0\left(\frac{2xv}{\alpha}\right)\delta(y-x^2) $$ $$ \int_0^\infty dx\ \frac{2x}{\alpha}\text{exp}\left(\frac{-(x^2+v^2)}{\alpha}\right)I_0\left(\frac{2xv}{\alpha}\right)\frac{\delta(x-\sqrt{y})}{2\sqrt{y}}=\frac{1}{\alpha}\exp\left(\frac{-(y+\nu^2)}{\alpha}\right)I_0\left(\frac{2\nu\sqrt{y}}{\alpha}\right)\ , $$ which for $\nu=0$ correctly reproduces the exponential PDF. Note that $\int_0^\infty f_Y(y)dy=1$, so the PDF is correctly normalized.

$$ \mathbb{E}[Y^\delta]=\int_0^\infty dy\ y^\delta \frac{1}{\nu^2}\exp\left(\frac{-(y+\nu^2)}{\nu^2}\right)I_0\left(\frac{2\sqrt{y}}{\nu}\right) =\nu ^{2 \delta } \Gamma (\delta +1) L_\delta(-1)\ , $$ as computed by Mathematica (for $\nu^2/\alpha=1$), where $L_\delta(x)$ is a Laguerre function [I checked a few numerical values and it seems it works just fine].

  • $\begingroup$ would you please provide the mathematica code. $\endgroup$ – Dimitrios Mar 1 '16 at 9:58
  • $\begingroup$ @Dimitrios Integrate[ y^[Delta] /[Nu]^2 Exp[-(y + [Nu]^2)/[Nu]^2] BesselI[0, ( 2 Sqrt[y])/[Nu]], {y, 0, Infinity}] $\endgroup$ – Pierpaolo Vivo Mar 1 '16 at 12:49

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.