1) Initialize at the 0th iteration |
 ∙ BS1, set \(p(x^{1}_{k}) = 1/M\) and calculate p(x k |y 1) and \(\mathbf {\Lambda }^{1}_{k}\) according to Table 1 using local data y 1. |
 ∙ BS2, set \(p(x^{2}_{k})\) based on \(\mathbf {\Lambda }^{1}_{k}\), calculate p(x k |y 1:2) according to Table 1 using local data y 2, update \(\mathbf {\Lambda }^{2}_{k}\) by (28). |
 ∙ if |p(x k |y 1)−p(x k |y 1:2)|<ε, go to 3); otherwise go to 2). |
2) Iterate until APP p(x k |y 1:2) converges. And at the t-th iteration, perform for i=1:2 |
 ∙ according to Table 1, update p(x k |y 1:2) using data y i with prior \(p(x^{i}_{k})\) obtained from \(\mathbf {\Lambda }^{s}_{k}\), s≠i. |
 ∙ update \(\mathbf {\Lambda }^{i}_{k}\) according to (30). |
3) Detect x k according to (3). |