Docstoc

Gaussian Distribution

Document Sample
Gaussian Distribution Powered By Docstoc
					Additive White
Gaussian Noise (AWGN)
Channel and Matched
Filter Detection
  ELE 745 – Digital Communications
          Xavier Fernando
ELE 745 – AWGN Channel

PART I – GAUSSIAN DISTRIBUTION
  Gaussian (Normal) Distribution

• The Normal or Gaussian distribution, is an important
  family of continuous probability distributions
• The mean ("average", μ) and variance (standard
  deviation squared, σ2) are the defining parameters
• The standard normal distribution is the normal
  distribution with zero mean (μ=0) and unity variance
  (σ2 =1)
• Many measurements, from psychological to thermal
  noise can be approximated by the Gaussian
  distribution.
Gaussian RV
General Gaussian RV
PDF of Gaussian Distribution




                   Standard Norma Distribution
CDF of Gaussian Distribution
       The Central Limit Theorem
• The sum of independent, identically distributed
  large number of random variables with finite
  variance is approximately normally distributed
  under certain conditions
• Ex: Binomial distribution B(n, p) approaches normal for large n
  and p
• The Poisson(λ) distribution is approximately normal N(λ, λ) for
  large values of λ.
• The chi-squared distribution approaches normal for large k .
• The Student’s t-distribution t(ν) approaches normal N(0, 1)
  when ν is large.
         Area under Gaussian PDF




The area within +/- σ is ≈ 68% (dark blue)
The area within +/- 2σ is ≈ 95% (medium and dark blue)
The area within +/- 2σ is ≈ 99.7% (light, medium, and dark blue)
           Bit Error Rate (BER)
• BER is the ratio of erroneous bits to correct bits
• BER is an important quality measure of digital
  communication link
• BER depends on the signal and noise power
  (Signal to Noise Ratio)
• BER requirement is different for different
  services and systems
  – Wireless link BER < 10-6 while Optical BER < 10-12
  – Voice  Low BER while Data  High BER
Logic 0 and 1 probability distributions
     Digital Receiver Performance
Probability of error assuming
Equal ones and zeros


       Where,




Depends on the noise variance at on/off levels and the
Threshold voltage Vth that is decided to minimize the Pe;
Often Vth = V+ + V-
          The Q Function




Fx(x) = 1 – Q(X)
Error Probability of On-Off Signaling
  BER (Pe) versus
   Q factor in a
  Typical Digital
Communication Link
PART-II
MATCHED FILTER DETECTION

				
DOCUMENT INFO