Coordinationless Coordinated Fastlane Network Service in Wireless Multimedia Sensor Networks

摘要

Motivated by the observation that there is an increasingly aggravated conflict between the expanding sensor capabilities for sensing service/content and the limited network performance in wireless multimedia sensor networks (usually caused by cost, congestion, packet loss, resource limit, etc.), we provide a bottom-up approach to speedup the network by leveraging a group of nodes' (fog nodes) capacity and piling them up as a considerable amount for fastlane service. Specifically, our approach is based on the following designs: 1. A novel way for capacity cumulation based on temporal and spatial diversity of network content/service access: there always exists redundant resources and ability among other sensors in time and spatial domain that can be piled up as a considerable amount, though each piece may be really small; 2. Content/Service Recoding, it is often the case that some content/service access require various resources (e.g., bandwidth) at a sensor who cannot meet. we divide content/service into ultimate small slices; 3. fog computing based network virtualization and distributed coordination enable the fog nodes be efficiently connected, and the recoded fragmented content/service can pass through the formed virtual network. Based on theoretical analysis, it is shown that our approach could achieve global optimal request/resource allocation in a distributed manner and maximize the network service ability. The extensive experimental study conducted in various scenarios (TCP, UDP, and both, etc.) also shows that the proposed approach could adapt to different networking environments and achieve significant improvement of quality of service (QoS) and experience (QoE). Furthermore, our approach is shown to be able to achieve linear speedup as the number of fog nodes increases.