The forward-backward (FB) charged particle multiplicity correlations between windows separated in rapidity and azimuth are analyzed using a model that treats strings as independent identical emitters. Both the short-range (SR) contribution, originati
ng from the correlation between multiplicities produced from a single source, and the long-range (LR) contribution, originating from the fluctuation in the number of sources, are taken into account. The dependencies of the FB correlation coefficient, $b$, on the windows rapidity and azimuthal acceptance and the gaps between these windows are studied and compared with the preliminary data of ALICE. The analysis of these dependencies effectively separates the contributions of two above mechanisms. It is also demonstrated that traditional definitions of FB correlation coefficient $b$ have a strong nonlinear dependence on the acceptance of windows. Suitable alternative observables for the future FB correlation studies are proposed. The connection between $b$ and the two-particle correlation function, $C_2$, is traced, as well as its connection to the untriggered di-hadron correlation analysis. Using a model independent analysis, it is shown that measurement of the FB multiplicity correlations between two small windows separated in rapidity and azimuth fully determine the two-particle correlation function $C_2$, even if the particle distribution in rapidity is not uniform.
