In this project, we explore network tomography based on network coding principles. We deal with multimedia transmission which is characterized by strict timing constraints. Our goal is to integrate network tomography within efficient multimedia delivery techniques that exploit network resources to maximize the end video quality of the receiving clients experiencing the multimedia data. In our approach, network coding is used to detect prospective bottlenecks in the network. Since multimedia data consists of segments exhibiting unequal importance for the overall reconstruction quality of the media presentation, we employ prioritized network coding that provides unequal error protection to the data commensurate with its importance. In our system, the detection of network bottlenecks is done in a receiver-driven fashion by the clients. Each client selectively detects its own bottlenecks for receiving the data such that its own video quality is maximized, e.g., low-bandwidth and high-bandwidth clients may not perceive the same network bottleneck in regard to data delivery. We devise a protocol for network link (flow) redirection for avoiding the bottlenecks once they are detected. The redirection is done such that the overall system performance is retained at the maximum level. We design orthogonal codebooks that are used for network coding the multimedia data by the respective sources. This enhances the performance of our network tomography techniques as it allows us to identify the network routes that the data associated with each multimedia source is sent over. Furthermore, we study the performance of our techniques in actual network scenarios. Among others, we examine how performance is affected by a random allocation of network coding codebooks in contrast to an actual orthogonal design. In conjunction, we study the implementation aspects of a tracking server that determines coherently the orthogonal codebooks for each media source involved in data transmission over the network.