| With the development of wireless networks and wireless multimedia applications, a variety of wireless multimedia system came into being, such as a wireless streaming media system called CWMS (Clear Wireless Multimedia System). However, wireless environment cannot guarantee the bandwidth, delay, jitter and packet loss rate, thus, we need to focus more on the quality of media.According to the above ideas, we proposed a set of wireless multimedia QoS solutions. This design is mainly based on the transmission procedure of a media stream. We firstly find out the main factors that affect the wireless multimedia quality: the performance and efficiency of encoding, the quality assurance during the stream's transmission. Afterwards, our QoS architecture could be work out in response to the two factors and meanwhile the characteristics of the existing CWMS. Thus, our proposal target in two major areas:the error recovery mechanism of encoding; the more suitable transport protocol. In transport layer, we decided to use RTP/TCP protocol instead of the UDP protocol, besides, in order to comply with our "centralized-distributed" architecture of CWMS, we give RTSP and RTP separated TCP connections, this new protocol we designed is called double TCP protocol. In application layer, we add FEC algorithm, with the help of the redundant code to realize the packet loss recovery. This QoS architecture on one aspect, has considered the exclusive features of CWMS, such as the "centralized-distributed" architecture. On the other aspect, we take into account the scalability and flexibility of QoS architecture, for example, FEC and double TCP protocol could both be encapsulated as far as possible, to meet different requirements of varying systems and platforms.At last, we do a more detailed simulation using NS2, the results shows our QoS architecture really works and have good improvement for wireless multimedia quality. And in the follow-up experiments, we focus on the efficiency assessment, and do a series of analysis. Hereafter, we find out the key factors that affect the processing time, those results may propose some direct recommendation for the following researches. |