Permeation is a thermodynamically and kinetically driven process. For small gases permeating through polymers, permeability is a product of solubility (thermodynamic aspect) and diffusivity (kinetic aspect) of the small gas molecule through the polymeric matrix. Permeability is an important property in the design of polymer-based membranes to be used in gas separation processes. Evidently, the permeability properties of a polymeric system are intimately related to its monomer composition and/or chain conformation. Developing such models requires information on how properties (density, diffusivity, solubility, etc.) vary as a function of polymer structure and architecture (length, branching etc.). The objective of this paper is to highlight a multiscale modeling approach to predict the permeability of the polymers as a function of its branched architecture. As experimental data of these properties as a function of the structural parameters defining branched architecture is not available, molecular modeling has been used to generate pseudo-experimental data. The next step is to use the generated data to develop higher order group contribution methods that take into account the parameters that define the branched architecture of polymers. The hierarchal procedure employed to generate the pseudo-experimental data (properties of polymers) include selecting the repeat unit of the polymer in question followed by creating a simulation box with a number of chains of the polymer. An extensive molecular dynamic (MD) simulation (up to an order of 100 ns) is performed on the simulation box whose static energy is minimized a priori, to obtain initial velocities and positions of particles in the system. Monte Carlo (MC) and Molecular Dynamics (MD) are employed on the extensively equilibrated structures to model solubility parameters and diffusion coefficients respectively. Results from these calculations are then compared with known experimental values to validate the procedure. The calculations are then repeated for architectures to generate the pseudo-experimental data. Data analysis using group contribution methods yields closed-form analytical expressions, relating these properties to features of molecular structure and conformation of the polymer. This approach has been tested and validated with polyethylene (PE) for which some experimental data was also available. For the case of PE considered here, the structural parameters include the molecular length (number of carbon atoms) of the main chain backbone, the molecular length of the branches and the spacing between successive branch points along the chain (branch frequency). The procedure was repeated for polyisobutylene (PiB).