1: \begin{abstract}

2: To ensure the system stability of the $\mathcal{H}_{2}$-guaranteed cost optimal decentralized control problem (ODC), we formulate an approximate semidefinite programming (SDP) problem based on the block diagonal structure of the gain matrix of the decentralized controller. To reduce data storage and improve computational efficiency, we apply the property of the Kronecker product to vectorize the SDP problem into a conic programming (CP) problem. Then, a proximal alternating direction method of multipliers (PADMM) is proposed to solve the dual of the resulted CP. By establishing the equivalence between the semi-proximal ADMM and the (partial) proximal point algorithm, we find the non-expansive operator of PADMM so that we can use Halpern fixed-point iteration to accelerate the proposed algorithm. Finally, we verify that the sequence generated by the proposed accelerated PADMM has a fast convergence rate for the Karush-Kuhn-Tucker residual. Through numerical experiments, it has been demonstrated that the accelerated algorithm surpasses the renowned COSMO, MOSEK, and SCS solvers in efficiently solving large-scale CP problems, particularly those arising from $\mathcal{H}_{2}$-guaranteed cost ODC problems.

3: \end{abstract}

4: