Quantum information theory is an important element of quantum computing and quantum communication systems. Whenever a quantum computer needs to send an output state to another party, or two parties need to establish quantum entanglement or secure keys via quantum communication, a quantum channel is inevitably involved. Hence it is absolutely important to understand the properties of quantum channels for the purpose of communication. Here, quantum entanglement plays a huge role. Pre-shared entanglement could enhance the capacity, whereas entanglement across inputs could render the capacity formulae impossible to compute. The first part of this thesis seeks to address this issue, by studying the additivity properties in the communication of classical and quantum information, with or without entanglement assistance. I also study the reverse problem that, given a channel capacity, what can be said about the quantum channel itself. Quantum information theory also serves as an important tool in understanding other systems, for example, black holes. In this thesis, I model a closed random system by a unitary channel, and study how typical unitary channels process information. This provides huge insight into the strength of generalized entanglement measures, and the hierarchies in the complexity of information scrambling.