This preview shows pages 1–2. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: ECE155A PROBLEMS Vlad Dorfman Class #10 Due: 2/19/07 1. Noise Enhancement. Let s(t) be the Lorentzian step response of the channel ( s(t) = 1/(1+(2*t/PW50)^2) ), and suppose we design a zeroforcing linear equalizer with target z(t)=sinc(t/T). If sampled, z(kT) = (k) and therefore will have no ISI. The formula for the frequency response of the continuous zeroforcing equalizer is = otherwise S T G T , ), ( ) ( The frequency response of the Lorentzian is 2 50 50 2 ) ( PW e PW S = a. Plot G ( e j ), the frequency response of the equalizer for PW50=1 and D=PW50/T=1.0, 1.6, 2.0 ,and 2.5. Organize your Mfiles as a script, calling a function eq_fun(D,PW50). In your script file set PW50=1, then call eq_fun(D,PW50) with the requested values of D. (Comment: the problem shows what happens if different numbers of bits are placed into the constant area; this is represented by the PW50 value which is the same for all cases with only the density D changing). b. Calculate the noise power at the output of the equalizer for the above values of normalized density D . If you cannot find an analytical solution, you may use a computer to make an approximation, setting 2 = 1 in your eq_fun. Hint: For a linear system that has a FT magnitude H( ), and input random process X(t) with power spectral density S X ( ), the output power spectral density will be S Y ( ) = H( ) 2 S X ( ). c. Make your conclusions as to dependency of noise enhancement on parameter D assuming the target z(t)=sinc(t/T). 2. ZeroForcing Equalizer Implementation in Frequency Domain Consider a discretetime channel impulse response h [ k ] defined as h (2)=2, h (1)=0.5, h (0)=5, h (1)=2, h (2)=1, h (k)=0 for k > 2. Use Matlab commands fft and ifft to determine (which is approximation, since fft...
View Full
Document
 Winter '07
 WOlf

Click to edit the document details