A Disciplined convex programming error problem {convex} >= {real constant}

 cvx_begin sdp
    variable W1(Nt,Nt) complex hermitian
    variable W2(Nt,Nt) complex hermitian
    variable t
    max  t     
    
    subject to
    
    t1=real(log_det(n1*eye(Nb,Nb)+H*W1*H'+(H*W2*H')));
    t2=real(log_det(n2*eye(Ne,Ne)+G*W2*G'));
    t3=real(log(det(n1*eye(Ne,Ne)+G*V*G')));
    t4=real(trace((inv(n1*eye(Nb,Nb)+H*V*H'))*H*W2*H'));
    t5=real(trace((inv(n1*eye(Nb,Nb)+H*V*H'))*H*V*H'));
    t6=real(log(det(real(n2*eye(Ne,Ne)+G*W*G'+G*V*G'))));
    t7=real(trace((inv(n2*eye(Ne,Ne)+G*W*G'+G*V*G'))*(G*W1*G'+G*W2*G')));
    t8=real(trace((inv(n2*eye(Ne,Ne)+G*W*G'+G*V*G'))*(G*W*G'+G*V*G')));
    (t1+t2-t3-t4+t5-t6-t7+t8)>=t;
    real(trace(W1+W2))<=P;
    W1>=0;
    W2>=0;
    norm(H*W1*H')+norm(G*W2*G')>=gama;
cvx_end

it turns out like this:

Disciplined convex programming error:
Invalid constraint: {convex} >= {real constant}

= (line 21)
b = newcnstr( evalin( ‘caller’, ‘cvx_problem’, ‘[]’ ), x, y, ‘>=’ );

TAES_Main (line 112)
norm(HW1H’)+norm(GW2G’)>=gama;

You cannot have

norm(something)>= constant

in a convex model. Since is obviously nonconvex set and hence cvx is the wrong tool.

Thanks,I understand what your meaning,but i still have no idea about how to solve the problem of the parper,could you please give me some advice,thanks.
MAX SR
W1,W2
s.t. SINR>=a1
Tr(W1)+Tr(W2)<=P
W1>=0,W2>=0
and SINR=||HW1H’||+||GW1G’||
SR is convex