The main focus of my this talk is on the design of communication algorithms for multi-terminal communications. We investigate the short-comings of the existing coding strategies used in networks, that are based on random unstructured codes. We show the necessity of certain forms of structure in optimality achieving codes, and argue that the lack of such structure in the current coding schemes contributes to their sub-optimality. We consider codes with different kinds of structures. We use these structures to develop new multi-terminal coding schemes with improved achievable regions.  The work consists of two parts. In the first part, we investigate the role of algebraic structure in the performance of communication strategies. In the second part, we consider a different structural restriction on codes used in network communication. Namely, we limit the `effective length' of these codes. We show that the large block-length single-letter coding schemes in the literature are sub-optimal in various multi-terminal communication settings.