It is known, that coherent chaotic communication systems are more vulnerable to noise in the transmission channel than conventional communications. Among the methods of noise impact reduction, such as extended symbol length and various digital filtering algorithms, the optimization of the synchronization coefficient may appear as a very efficient and simple straightforward approach. However, finding the optimal coefficient for the synchronization of two chaotic oscillators is a challenging task due to the high sensitivity of chaos to any disturbances. In this paper, we propose an algorithm for finding the optimal synchronization parameter K_opt for a coherent chaos-based communication system affected by various noises with different signal-to-noise ratios (SNR). It is shown, that under certain conditions, optimal $K$ provides the lowest possible bit error rate (BER) during the transmission. In addition, we show that various metrics applied to the message demodulation task propose different noise immunity to the overall system. For the experimental part of the study, we simulated and physically prototyped two chaotic communication systems based on well-known Rossler and Lorenz chaotic oscillators. The microcontroller-based prototype of a chaotic communication system was developed to investigate the influence of noise in the real transmission channel. The experimental results obtained using the designed hardware testbench are in good correspondence with the theoretical propositions of the study and simulation results. The suggested evaluation metrics and optimization algorithms can be used in the design of advanced chaos-based communication systems with increased performance.