getoutagelength.m

fading 在无线中的应用

function [bleng, nbursts] = getOutageLength(x, minvalue)
%  bleng = getOutageLength(x, minvalue)
%
%  This function gets a series of values, x, and compares it with a
% lower threshold, minvalue. If several consecutive points are
% below the threshold, that is considered an "outage burst". The
% function returns a vector with the burst length, i.e. the length
% of the vector is equal with the number of burst lengths in the
% series and the i-th position is the length of the i-th burst.
%
%   [bleng, nb] = getOutageLength(X, minvalue)
% gets several series, each one being a column in x. In this case
% bleng is a vector with all the burst lengths in X. nb is a vector
% with as many elements as columns in X; nb(i) how many bursts are
% there in X(:,i). Thus bleng(1:nb(1)) are the lengths of the
% bursts in the first series.
%

% Created 2006.03.09 by Bogdan Timus
% $Id: getOutageLength.m,v 1.1 2006/03/10 17:48:08 bogdant Exp $

% find out how many series are there and how long. N is the number
% of series, M is the length of each series.
[M, N] = size(x);

% mark the seconds in outage. Put an extra row of zeros, to be sure
% that all the outage burst end.
inoutage = [x < minvalue; logical(zeros(1,N))];

% mark the moments when the outage burst ends
isend = inoutage(1:M,:) &~inoutage(2:M+1,:);

% count how many bursts are there for each observation
nbursts = sum(isend)';

% count the burst lengths. Use the following tricks. Accumulate the
% number of observations in outage. At the end of each burst
% (marked by isend) the accumulated value shows how many points
% have been in outage up to then, i.e. the sum of burst length up
% to the end of that particular burst.
tmp = cumsum(inoutage(1:M,:));
tmp(~isend) = 0;
[dummy, realis, cumleng] = find(tmp);

% cumleng contains the partial sum of burst length, concatenated
% for all the series. Thus cumleng(1) is the length of the first
% burst in the first series, cumleng(i) is the sum of all the burst
% lengths up to the i-th, provided that the i-th burst is still in
% the first series. Use diff to get the individual burst lengths
% from the sum of lengths.
bleng = [cumleng(1);diff(cumleng)];

% Since the series have been concatenated, diff will give a wrong
% value for the first burst in the series, if it is not in the
% first series. nbursts could have been used here too, to find
% which burst is the first in the series.
indstart = bleng<0;
bleng(indstart) = cumleng(indstart);