This preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: ECE 5670 : Digital Communications (Spring 2011) Homework 4 Due February 24 in class Instructor: Salman Avestimehr Oﬃce 325 Rhodes Hall 1. Recall that in class it was shown that for a binaryPAM constellation, reliable com munication was possible at all rates less than R ∗ = 1 − log 2 1 + e − E 2 σ 2 . In this problem we generalize this result for an MPAM constellation. Consider the channel ~y = ~x + ~w where ~w is a T × 1 noise vector whose entries are i.i.d. Gaussian random variables with zero mean and variance σ 2 . The vector ~x is a T × 1 vector representing the codeword that is transmitted. There are 2 RT possible codewords to transmit which we label ~v 1 , ~v 2 , . . . , ~v 2 RT . These codewords are generated as follows. Let C be a random generator matrix with T rows and RT log 2 M columns. It has RT 2 log 2 M entries and these are i.i.d. random variables taking on values in the set { 1 , 2 , . . . , M } with equal probability 1 /M . The data vector is denoted by B . It has dimension RT log 2 M × 1. There are M RT / log 2 M = 2 RT possible data vectors B 1 , B 2 , . . . , B 2 RT , representing all possible combinations of elements from the set { , 1 , . . . , M − 1 } . Thus R ≤ log 2 M . For example the first data vector is B 1 = [0 , , . . . , 0] t and the last one is B 2 RT = [ M − 1 , M − 1 , . . . , M − 1] t . The codeword ~v i is given by ~v i = √ E 2 M − 1 ( CB i mod M ) − 1 In other words, we map the Mary valued entries of the data vector into an MPAM constellation with maximum amplitude √ E . For example if M = 4, and CB i = [0 , 1 , 2 , 3] t , then we transmit ~v i = − √ E − √ E/ 3 + √ E/ 3 + √ E (a) Using the union bound, show that the probability that the ML decoder makes an error satisfies Pr( E ) ≤ 2 RT X i =1 2 RT X j =1 ,j 6 = i Pr( ~v i → ~v j  ~x = ~v i ) Pr( ~x = ~v i ) where Pr( ~v i → ~v j  ~x = ~v i ) denotes the probability that the received vector lies...
View
Full Document
 Spring '11
 SCAGLIONE
 Probability theory, Linear code, information bits, iterative decoding algorithm

Click to edit the document details