www.pudn.com > RAKE.rar > ljq_IEEEUWB.m

%/*-----------------IEEE802.15.3a UWB信道模型---------------*/ 
%------------------------by LinJiaqing----------------------- 
% Generates the channel impulse response for a multipath 
% channel according to the statistical model proposed by 
% 'fc' is the sampling frequency 
% 'TMG' is the total multipath gain 
% The function returns: 
% 1) the channel impulse response 'h0' 
% 2) the equivalent discrete-time impulse response 'hf' 
% 3) the value of the Observation Time 'OT' 
% 4) the value of the resolution time 'ts' 
% 5) the value of the total multipath gain 'X' 
function [h0,hf,OT,ts,X] = ljq_IEEEUWB(fc,TMG); 
% ---------------------------- 
% Step Zero - Input parameters 
% ---------------------------- 
%-------------IEEEUWB信道------------------------------------- 
OT = 200e-9;              % Observation Time [s] 
ts = 1e-9;                % time resolution [s] 
                          % i.e. the 'bin' duration   
LAMBDA = 0.0223*1e9;      % Cluster Arrival Rate (1/s) 
lambda = 2.5e9;           % Ray Arrival Rate (1/s) 
GAMMA = 7.1e-9;            % Cluster decay factor   
gamma = 4.3e-9;            % Ray decay factor   
sigma1 = 10^(3.3941/10);  % Stdev of the cluster fading 
sigma2 = 10^(3.3941/10);  % Stdev of the ray fading 
sigmax = 10^(3/10);       % Stdev of lognormal shadowing 
% ray decay threshold 
rdt = 0.001; 
% rays are neglected when exp(-t/gamma) 0 
% -------------------------------- 
% Step Two - Path characterization 
% -------------------------------- 
NC = length(CAT);   % Number of observed clusters 
logvar = (1/20)*((sigma1^2)+(sigma2^2))*log(10); 
omega = 1; 
pc = 0;             % path-counter 
for i = 1 : NC 
    pc = pc + 1; 
    CT = CAT(i);    % cluster time 
    HT(pc) = CT; 
    next = 0; 
    mx = 10*log(omega)-(10*CT/GAMMA); 
    mu = (mx/log(10))-logvar; 
    a = 10^((mu+(sigma1*randn)+(sigma2*randn))/20); 
    HA(pc) = ((rand>0.5)*2-1).*a; 
    ccoeff = sigma1*randn;  % fast fading on the cluster 
    while exp(-next/gamma)>rdt 
    pc = pc + 1; 
    next = next + expinv(rand,t); 
    HT(pc) = CT + next; 
    mx = 10*log(omega)-(10*CT/GAMMA)-(10*next/GAMMA); 
    mu = (mx/log(10))-logvar; 
    a = 10^((mu+ccoeff+(sigma2*randn))/20); 
    HA(pc) = ((rand>0.5)*2-1).*a; 
    end 
end % for i = 1 : NC 
% Weak peak filtering 
peak = abs(max(HA)); 
limit = peak/10^(PT/10); 
HA = HA .* (abs(HA)>(limit.*ones(1,length(HA)))); 
for i = 1 : pc 
    itk = floor(HT(i)/dt); 
    h(itk+1) = HA(i); 
end 
% ------------------------------------------- 
% Step Three - Discrete time impulse response 
% ------------------------------------------- 
N = floor(ts/dt); 
L = N*ceil(length(h)/N); 
h0 = zeros(1,L); 
hf = h0; 
h0(1:length(h)) = h; 
for i = 1 : (length(h0)/N) 
    tmp = 0; 
    for j = 1 : N 
        tmp = tmp + h0(j+(i-1)*N); 
    end 
    hf(1+(i-1)*N) = tmp; 
end 
% Energy normalization 
E_tot=sum(h.^2); 
h0 = h0 / sqrt(E_tot); 
E_tot=sum(hf.^2); 
hf = hf / sqrt(E_tot); 
% Log-normal shadowing 
mux = ((10*log(TMG))/log(10)) - (((sigmax^2)*log(10))/20); 
X = 10^((mux+(sigmax*randn))/20); 
h0 = X.*h0; 
hf = X.*hf;