1: Start with $V^k:M^k\times D^k$ .

2: Begin iteration.

3: Compute interference plus noise covariance matrix for $B_d^k$ for stream d at receiver k, $\forall k\in \left\{1,\cdots ,K\right\},d\in \left\{1,\cdots ,D^k\right\}$ .

4: Calculate receive combining vectors $u_d^k$ at receiver k, $\forall k\in \left\{1,\cdots ,K\right\},d\in \left\{1,\cdots ,D^k\right\}$ .

5: Reverse the communication direction and use the receive combining vectors as precoding vectors: $\overleftarrow{V^k}=U^k,\forall k\in \left\{1,\cdots ,K\right\}$ .

6: In the reciprocal network, compute interference plus noise covariance matrix for $\overleftarrow{B_d^k}$ for stream d at receiver k, $\forall k\in \left\{1,\cdots ,K\right\},d\in \left\{1,\cdots ,D^k\right\}$ .

7: Calculate receive combining vectors $\overleftarrow{u_d^k}$ , $\forall k\in \left\{1,\cdots ,K\right\},d\in \left\{1,\cdots ,D^k\right\}$ .

8: Reverse the communication direction and use the receive combining vectors as precoding vectors: $V^k=\overleftarrow{U^k},\forall k\in \left\{1,\cdots ,K\right\}$ .

9: Repeat until convergence, or the number of iterations reaches a limit defined earlier.