Nonconvex Distributed Optimization Based Power Allocation for Maximizing DL-UL Total Sum Rate in Dynamic TDD Systems

We investigate the power allocation problem of a dynamic time division duplexing based heterogeneous network comprising downlink (DL) macro base stations (MBSs) and uplink (UL) small base stations (SBSs). In such networks, beyond intracell interference, the asynchronous DL transmission of MBSs and UL transmission of SBSs introduce additional interference known as cross-link interference. This interference occurs not only between MBSs and SBSs, but also between macro-cell user equipment and small-cell user equipment (SUE). To maximize the sum rate of UL and DL, we formulate a non-convex distributed optimization problem where power allocation variables of both DL and UL are to be optimized. We propose a power allocation algorithm relying on the Lagrange method with logarithmic barrier. Simulation results demonstrate that our proposed algorithm outperforms the representative benchmark schemes.