Cedric Roux
56c66c09aa
Pre-ci test passed git-svn-id: http://svn.eurecom.fr/openair4G/trunk@4000 818b1a75-f10b-46b9-bf7c-635c3b92a50f
57 lines
1.2 KiB
Matlab
57 lines
1.2 KiB
Matlab
function [K, V2, s2] = estimate_rice(G)
|
|
% Ricean factor estimation
|
|
% [K, V2, s2] = estimate_rice(h) estimates the Ricean K-Factor from the set of
|
|
% impulse responses h using Moment Method estimation [1].
|
|
%
|
|
% [1] Greenstein, L., Michelson, D., and Erceg, V.,
|
|
% "Moment-method estimation of the Ricean K-factor",
|
|
% #IEEE_J_COML#, 1999, 3, 175-176
|
|
%
|
|
% Parameters:
|
|
% G - size [NTau x Nsamp] Sequence of impulse responses (absolute square)
|
|
%
|
|
% Author: Florian Kaltenberger
|
|
% Copyright: Eurecom Sophia Antipolis
|
|
%
|
|
% Version History
|
|
% Date Version Comment
|
|
% 20080620 0.1 Created
|
|
% 20080730 0.2 We caclculate the rice factor for every delay bin
|
|
|
|
[NTau, Nsamp] = size(G);
|
|
|
|
if ~isreal(G)
|
|
error('Input G must be real');
|
|
end
|
|
|
|
Ga = mean(G,2);
|
|
|
|
Gv = zeros(NTau,1);
|
|
V2 = zeros(NTau,1);
|
|
s2 = zeros(NTau,1);
|
|
K = zeros(NTau,1);
|
|
|
|
for t=1:NTau
|
|
Gv(t) = sqrt(mean((G(t,:)-Ga(t)).^2));
|
|
|
|
if Gv(t)>Ga(t)
|
|
V2(t)=0;
|
|
else
|
|
V2(t) = sqrt(Ga(t)^2 - Gv(t)^2);
|
|
end
|
|
|
|
s2(t) = Ga(t) - V2(t);
|
|
|
|
K(t) = V2(t)/s2(t);
|
|
end
|
|
|
|
% %% plot
|
|
% figure(69)
|
|
% cdfplot(G);
|
|
% hold on
|
|
% x = linspace(0,2.5,100);
|
|
% p = ricepdf(x,Ga,sqrt(s2));
|
|
% plot(x,cumsum(p)/sum(p),'r')
|
|
% legend('empirical cdf','reconstructed cdf')
|
|
% hold off
|