Research On Cognitive Service Of Initiative Multimedia Streaming In MANETs

Provision of peer-to-peer-based (P2P) video-on-demand (VoD) services with initiative multimedia streaming characteristics in mobile ad hoc networks (MANETs) to support initiative and interactive accessing video resources of large-scale mobile users has become one of key technologies of future mobile multimedia streaming. However, the deployment of P2P VoD system in MANETs is a significant challenge.(1) VoD services support the asynchronous access and initiative changing for current playing content, so as to lead to high jitter level between logical links of nodes in P2P networks and bring negative effects for user playback continuity.(2) The existing P2P technologies which have the following problems:rigescent networking strategies, high-cost maintenance of networks, low-efficiency resource searching, cannot meet the demand of high-efficiency management and distribution of video resources in mobile environment.(3) Due to the shortage of existing studies on the strategies of perceiving variation of network environment, the mobile nodes do not have the capacities of adapting dynamic network environment, the static mechanisms of cooperation between mobile nodes leads to deficient supply of available video resources, high-cost resource fetching and low-efficiency sharing. Aimed to the above problems, this paper researches the following key technologies:perception of user playback behavior, distribution infrastructure of video resources and perception of mobility of mobile nodes, and obtains the following contributions.(1) The paper designed a new reliability-oriented perception and prediction model of user playback behavior. Perceiving user playback behaviors which have initiative and interactive characteristics and reflect the variation of user interest for video content and jitter level of request and utilization of resources is cognition-driven source for the research on initiative multimedia streaming. The paper first proposed a prediction algorithm of resource requirement, which can accurately predict future playback content of users. The predicted results are pre-downloaded into local buffer to ensure the playback continuity. The estimation model of user playback state is proposed. By making use of the estimation model to identify the users with reliable state based on their historical playback records, the stability of logical links between supplier and receiver of resources can be ensured.(2) The paper proposed a novel mobile P2P-based management and distribution mechanism of video resources. The constraint in storage capacity of mobile devices and dynamic variation of resource requirement bring negative effects for construction of mobile P2P networks and distribution of video resources. The paper first proposed a mobile P2P networking strategy and a searching resource algorithm, which constructs the micro-community-based mobile P2P networks in terms of historical and current playback content. By making use of the prediction model of user playback behavior, the logical relationships between micro-communities are constructed to obtain efficient searching resources. The maintenance strategy of mobile P2P networks is proposed, which relies on maintaining the node with reliable state to reduce the maintenance cost.(3) The paper designed a novel sharing and cooperative fetching mechanism of video resources based on perceiving dynamic variation of network environment. The dynamic network topology structure leads to unreliable link and low performance of video data delivery, so as to bring negative effects for smooth playback experience of users. The paper first designed a sharing mechanism of video resources based on cross-layer perception and neighbor cooperation. By perceiving the variation of communication quality of one-hop neighbors, the estimation model of resource fetching cost is proposed. The construction of resource-centric cluster structure can enhance the efficiency of resource sharing and performance of video data delivery. The estimation model for the stability of mobility and communication quality of one-hop neighbors is proposed. The predicted playback content is cooperatively download into local buffer with the help of one-hop neighbor which has high stability of mobility and communication quality to, reducing negative effects of video data delivery from mobility of mobile nodes and variation of communication quality.