| dc.description.abstract | Nowadays, many enterprises have already published the multiparty conference real-time applications such as Skype, LINE, ZOOM, and Web Real-Time Communication (WebRTC). Among these multiparty conference applications, there are two kinds of architectures: centralized and decentralized architectures.
Currently, the popular technology is cloud computing, which is centralized architecture. It uses servers to relay the video and audio data among devices. Such architecture can reduce the bandwidth and CPU loading of each participant in the conference, and support large amount of participants in a conference by powerful servers. As the number of participants increases, the server also requires more computing power.
In contrast, with decentralized architecture, each participant in conference directly exchanges the data of video and audio with each other without the assistance from server. Although it does not need the cost of maintaining server, the bandwidth and computation power required increases as the number of participants increases. Consequently, the number of participants is limited by the resource limitation.
In this thesis, we compare and analyze the centralized and decentralized architectures, integrate the advantages of centralized and decentralized architectures, and utilize the mixing and demixing technologies for audio traffic. More specifically, this thesis proposes a new network topology for mobile devices to support multiparty conference. | en_US |