CVX says it is infeasible however I don't know the reason

Hello everyone, could you please tell me how to solve my problem? Thanks a lot !

The code is as follows

N=5;
para_R =1.0e-06 *[0.1965, 0.2548,  0.1956,0.2055,0.1666]' ;
para_H =1.0e-04 *[0.6667, 0.6667, 0.6667, 0.6667, 0.6667]' ;
vepsilon =[500000, 500000, 200000, 200000, 900000]' ;
lambda =[2000, 2000, 2000, 2500, 2000]' ;
G_R =1.0e+09 *[0.7071, 0.2700, 0.7184, 0.5992, 1.3018]' ;
G_H =1.0e+08 *[2.8295, 2.8293, 2.8295, 2.8293, 2.8293]' ;
T_off =[0.2504, 0.8707, 5.1737, 0.1872, 3.7307]' ;
P_max =[0.2000, 0.2000, 0.2000, 0.2000,  0.2000]' ;
P_R=[0.0276294318263112;0.0724771549524832;0.0194614728358243;0.0219179215981384;0.0150227196428354];
P_H=[0.0953780282663401;0.0913987565551282;0.0720023792360408;0.0758810980458612;0.175543976944540];
b_R=[0.0554769851565014;0.0575317054196634;0.0153969156224736;0.0122249993451960;0.0751708727405072];
b_H=[0.161958849810163;0.161472466304325;0.0287743359255664;0.0464143606059894;0.381045453479579];
x_L=[0.239570375042641;0.239701592259968;0.589374827073154;0.466875918713161;0.134982324613437];
x_R=[0.175189318833783;0.181283287349420;0.113358159121712;0.0595430120162252;0.127509421676755];
x_H=[0.585240306123576;0.579015120390613;0.297267013805134;0.473581069270614;0.737508253709808];
f_L=1e9;
F_R=10e9;
F_H=50e9;
B_max=1e6;
para=log(2);
cvx_begin 
    variables   tao_R(N,1)  tao_H(N,1)  f_R(N,1)  f_H(N,1) ;
    T_L=x_L.*vepsilon.*lambda./f_L;
    T_R=para_R+tao_R+x_R.*vepsilon.*lambda.*inv_pos(f_R)./(F_R);
    T_H=para_H+tao_H+x_H.*vepsilon.*lambda.*inv_pos(f_H)./(F_H);
    T_matrix=max([T_L,T_R,T_H],[],2)

minimize sum(T_matrix); 
tao_R>0;
tao_H>0;
f_R>0;
f_H>0;


for n=1:N 
     para_R(n)+tao_R(n)+x_R(n)*vepsilon(n)*lambda(n)*inv_pos(f_R(n))<=T_off(n);
     -tao_R(n)*b_R(n)*B_max*log(b_R(n)*B_max/(b_R(n)*B_max+P_R(n)*G_R(n)))>=x_R(n)*vepsilon(n)*para;
     -tao_H(n)*b_H(n)*B_max*log(b_H(n)*B_max/(b_H(n)*B_max+P_H(n)*G_H(n)))>=x_H(n)*vepsilon(n)*para;
end 
sum(f_R)<=1;
sum(f_H)<=1;
cvx_end

The scaling of the input data is horrible, which can make solver performance and conclusions unreliable.

First, fix the scaling so that non-zero input data are within a small number of orders or magnitude of one.

Second, if the model is still reported as infeasible, follow the advice in https://yalmip.github.io/debugginginfeasible , which also applies to CVX, except for section 1.

Thank you very much, Mark! The link you provided helps me a lot. Thanks again!