The paper presents an overview of the electrical bio-impedance (EBI) signal decomposition into its cardiac and respiratory components. This problem mainly originates from the nonstationarity of the signal components and overlapping of their harmonic spectra. In the introductory part of the paper, an overview of the bio-impedance signal decomposer (BISD), as a solution of the problem, is accompanied with an introduction to a cardiac BI signal model, which is constructed from the components of the application-specific orthonormal basis. In the main part of the paper a semi-synchronous cardiac signal amplitude estimator, which is based on the cardiac signal model and on the proposed extrema searching algorithm, is proposed. After that, the cardiac signal lock-in detection algorithm is proposed. Finally, a conditioning of the estimated cardiac signal frequency is discussed. The proposed amplitude estimator, lock-in detector and frequency conditioning increase twice the reaction speed of the BISD to the input EBI signal. The proposed version of the BISD estimates the cardiac signal amplitude during only a few cardiac periods, even if very large difference between amplitudes exists in different conditions. As a result, the entire BISD becomes locked during 8 s (including 4 s of soft start). The proposed improvements allowed reducing the latency of the BISD from 2 to 1 s.