Optimal Power Allocation in Zero-Forcing Assisted PMSVD-based Optical MIMO Systems

The multiple-input multiple-output (MIMO) technology has attracted a lot of interest in the optical fiber community: MIMO allows higher data rates with no loss regarding bandwidth and transmit power. The inherent interferences within a MIMO system call for appropriate signal processing strategies. Polynomial schemes such as polynomial matrix singular value decomposition (PMSVD) have attracted the research interest since they do not require any block transmission compared with schemes such as spatio-temporal vector coding (STVC). In comparison to SVD-based STVC schemes, where the whole MIMO system is transferred into a number frequency non- selective non-interfering single-input single-output (SISO) systems, PMSVD transfers the MIMO system into independent SISO layers exhibiting a frequency-selective characteristic. In this work optimal power allocation solutions in PMSVD-assisted optical MIMO systems with layer-based zero-forcing equalization are investigated. As shown by our computer simulations and testbed measurements, optimal power allocation (PA) leads to the best BERs at the cost of a high complexity compared to suboptimal PA solutions. However, as highlighted by our simulation results, suboptimal PA schemes seem to be a good alternative for achieving nearly optimum bit-error rate (BER) at a significantly lower complexity.