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: EE 478 Handout #14 Multiple User Information Theory October 20, 2008 Homework Set #3 Solutions 1. Solution: Let Y = X 1 + X 2 . Then P( Y = 0) = (1 p 1 )(1 p 2 ), P( Y = 1) = (1 p 1 ) p 2 + p 1 (1 p 2 ), and P( Y = 2) = p 1 p 2 , and H ( Y ) = summationdisplay y { , 1 , 2 } P( Y = y )log P( Y = y ) , = f ( p 1 ,p 2 ) . By differentiating with respect to p 1 and p 2 , and solving f p 1 = f p 2 = 0 , we get p 1 = p 2 = 0 . 5 which corresponds to H ( Y ) = 1 . 5. To verify that the entropy of Y is maximized at this point, we also need to investigate the boundary points. But if p i { , 1 } for i = 1 or i = 2, then H ( Y ) 1. 2. Solution: (a) The capacity region of this channel is the convex closure of the set of all ( R 1 ,R 2 ) such that R 1 H ( Y  X 1 ,Q ) , R 2 H ( Y  X 2 ,Q ) , R 1 + R 2 H ( Y  Q ) for some p ( q ) p ( x 1  q ) p ( x 2  q ). But since the RHS of each inequality is bounded by 1, the capacity region is the triangle region specified by R 1 + R 2 1. If we let X 1 Bern( p 1 ) and X 2 Bern( p 2 ), we see that the points on the boundary are achievable when p 1 p 2 = 1 p 1 p 2 , i.e., p 1 p 2 = 1 / 2. (b) The capacity region of this channel is again the convex closure of the set of all ( R 1 ,R 2 ) such that R 1 H ( X 1  Q ) , R 2 H ( X 2  Q ) , R 1 + R 2 H ( Y  Q ) for some p ( q ) p ( x 1  q ) p ( x 2  q ). But since the RHS of each inequality is bounded by 1, the capacity region is the triangle region specified by R 1 + R 2...
View Full
Document
 Spring '11
 Kelly

Click to edit the document details