Towards Optimal Data Transmission by Network Coding

Abstract

Communication protocols based on coded schemes and, in particular, network coding promise increased efficiency for future networks. The field is still young. Nevertheless, many important results have been already obtained. In this thesis, network coding is studied in three directions: In one direction, wireless network coding is studied. Wireless communication involves sharing of scarce resources, and special care is required to maximize the utilization of the medium. Network coding, in that sense, has the potential to improve the capacity. There are many issues involved in modeling a wireless network. Adapting all of these issues to network coding is a difficult task. We have considered a simplified model that takes into account the broadcast property and the transmission power. Based on this setting, we analyze some aspects of network coding in wireless networks. In this study, optimization algorithms were needed to compute special flow graphs for network coding. We propose special optimization algorithms for that purpose. We have studied the type of cycles that appear in wireless broadcast networks, and through simulation we observe the penalty of avoiding cycles in wireless network coding. In another direction, the delay is studied from the queueing theoretic point of view, for code based communication protocols. We have improved previous results and propose a better measurement for delay. Finally, we study secure network coding. For secure network coding, a new attack model is studied and the secrecy capacity is improved by a concatenated secret sharing scheme.