NeTS-NBD: Towards a Disconnection-Tolerant, Opportunistic Internet

该项目考虑通过扩展互联网以支持大规模断开容忍和机会主义连接来扩大互联网的覆盖范围。这种网络称为DTN或DTON。这种范式扩展对于：空中/汽车，农村/野外和紧急/战时DTN是有价值的。当前的网络协议经过优化，可以在网络大部分时间处于连接状态，很少断开连接的情况下运行。这个项目系统地探讨了断开连接是常态，节点/集群之间的连接是机会主义的情况。智力上的优点。由于在DTON中很难维护结构，因此该项目使用非结构化的构建块。路由的基本问题是使用有效的，非结构化的方法，主要是随机游走。虽然非结构化方法在特殊的底层图结构中工作得很好，但如何使用非结构化搜索来实现点对点连接（相当于“大海捞针”的间接性）是一个开放的问题。DTON建模使用一个新的组合对象：时间演化图代数张量表示。随机游走的实际协议使用了弱状态的新概念，这种新概念是局部学习或全局传播的。信息理论技术被用来开发网络有效吞吐量和开销的界限。更广泛的影响包括DTON在各种动态网络中的应用，如高速公路交通管理和生物系统。将互联网的覆盖范围扩大到空中、荒野和无基础设施的地方，其经济和社会影响是巨大的。PI正在向行业和国家实验室传播这些想法的初步构想。PI整合了他们课程中的材料，并为该项目开设了一个新颖的网络博客（博客），以便更广泛地传播。

This project considers broadening the reach of the Internet by extending it to support large-scale disconnection tolerant and opportunistic connectivity. Such networks are called DTNs or DTONs. This paradigm extension is valuable for: airborne/automobile, rural/wilderness and emergency/wartime DTNs. Current networking protocols are optimized to function when networks are connected most of the time, and disconnected rarely. This project systematically explores situations where disconnection is the norm, and connectivity between nodes/clusters is opportunistic. Intellectual merit. Since structure is hard to maintain in DTONs, the project uses unstructured building blocks. The fundamental problem of routing is accomplished using efficient, unstructured methods, primarily random walks. While unstructured methods work well in special underlying graph structures, it is an open problem of how to use unstructured search for point-to-point connectivity (equivalent to "needle-in-a-haystack" indirections). DTONs are modeled using a new combinatorial object: time-evolving graphs represented algebraically as tensors. The actual protocols for random walks are biased using the new notion of weak-state that is learned locally or disseminated globally. Information-theoretic techniques are used to develop bounds for network goodput and overhead. Broader impact includes DTON applications across a variety of dynamic networks such as highway traffic management and biological systems. The economic and social impacts of extending the reach of the Internet to airborne, wilderness and infrastructure-free locations are immense. The PIs are disseminating preliminary formulations of such ideas to the industry and national laboratories. The PIs integrate material across their courses, and operate a novel weblog (blog) for the project for broader dissemination.