Over the past decade, network coding (NC) has emerged as a new paradigm for data communications and has attracted much popularity and research interest in information and coding theory, networking, wireless communications and data storage. Random linear NC (RLNC) is a subclass of NC that has shown to be suitable for a wide range of applications thanks to its desirable properties, namely throughput-optimality, simple encoder design and efficient operation with minimum feedback requirements. However, for delay-sensitive applications, the mentioned advantages come with two main issues that may restrict RLNC usage in practice. First is the trade-off between the delay and throughput performances of RLNC, which can adversely affect the throughput-optimality of RLNC and hence the overall performance of RLNC. Second is the usage of feedback, where even if feedback is kept at minimum it can still incur large amount of delay and thus degrade the RLNC performance, if not optimized properly. In this thesis, we aim to investigate these issues under two broad headings: RLNC for applications over time division duplexing (TDD) channels and RLNC for layered video streaming. For the first class of problems, we start with the reliable broadcast communication over TDD wireless channels with memory, in the presence of large latency. Considering TDD channels with large latency, excessive use of feedback could be costly. Therefore, joint optimization of feedback rate and RLNC parameters has been studied previously for memoryless channels to minimize the average transmission time for such settings. Here, we extend the methodology to the case of channels with memory by benefiting from a Gilbert-Elliot channel model. It is demonstrated that significant improvement in the performance could be achieved compared to the scheme which is oblivious to the temporal correlations in the erasure channels. Then, keeping our focus on network coded TDD broadcast systems with large latency, we consider delay sensitive applications and study the issue of throughput and packet drop rate (PDR) optimization as two performance metrics when the transmission time is considered fixed. We propose a systematic framework to investigate the advantage of using feedback by comparing feedback-free and feedback schemes. Furthermore, the complicated interplay of the mean throughputs and PDRs of users with different packet erasure conditions is discussed. Then, to better analyze the throughput performance of the proposed feedback-free scheme, we formulate the probability and cumulative density functions of users' throughputs and utilize them to investigate the problem of guaranteeing the quality of service. Finally, it is shown that the optimized feedback-free RLNC broadcast scheme works close enough to an idealistic RLNC scheme, where an omniscient sender is assumed to know the reception status of all users immediately after each transmission. For the second class of problems, we consider transmitting layered video streams over heterogeneous single-hop wireless networks using feedback-free RLNC. For the case of broadcasting single video stream, we combine RLNC with unequal error protection and our main purpose is twofold. First, to systematically investigate the benefits of the layered approach in servicing users with different reception capabilities. Second, to study the effect of not using feedback, by comparing feedback-free schemes with idealistic full-feedback schemes. To this end, we consider a content-independent performance metric and propose a general framework for calculation of this metric, which can highlight the effect of key parameters of the system, video and channel. We study the effect of number of layers and propose a scheme that selects the optimum number of layers adaptively to achieve the highest performance. Assessing the proposed schemes with real H.264 test streams, the trade-offs among the users' performances are discussed and the gain of adaptive selection of number of layers to improve the trade-offs is shown. Furthermore, it is observed that the performance gap between the proposed feedback-free scheme and the idealistic scheme is small and the adaptive selection of number of video layers further closes the gap. Finally, we extend the problem of layered video streaming to the case of transmitting multiple independent layered video streams and demonstrate the gain of coding across streams (i.e., inter-session RLNC) over coding only within streams (i.e., intra-session RLNC).