Towards an overlay network architecture

自组织覆盖网络已成为大规模部署网络服务的一种强大的新范例，但该技术的全部潜力在很大程度上仍未得到开发。该项目将推进自组织虚拟覆盖网络的最新技术，并表明自组织网络的设计原则可以演变为未来通信网络和服务的新协议架构的蓝图。我们将为完全基于自组织覆盖网络原理的网络架构开发算法和协议。通过对测试平台实现的分析方法、模拟和测量，我们将对自组织网络的能力和局限性提供新的见解。该项目解决的一个关键挑战是开发算法和协议，这些算法和协议可以在多个异构基板网络存在的情况下保持连接的覆盖图。为了证明自组织网络可以在系统架构的较低层提供高延迟和吞吐量性能，我们提出开发具有硬件支持自组织覆盖网络协议的“覆盖交换机”。我们将探索使覆盖网络扩展到数百万节点的算法。此外，我们将设计和评估覆盖网络的算法，从而有效利用网络资源。

Self-organizing overlay networks have emerged as a powerful new paradigm for deploying network services on a large scale, yet the full potential of this technology remains largely unexplored. This project will advance the state-of-the-art of self-organizing virtual overlay networks and show that design principles for self-organizing networks can be evolved into a blueprint for a new protocol architecture for future communication networks and services. We will develop algorithms and protocols for a network architecture that is entirely based on the principles of self-organizing overlay networks. Using analytical methods, simulation, and measurements of testbed implementations we will provide new insights into the capabilities and limitations of self-organizing networks. A key challenge addressed in this project is the development of algorithms and protocols that can maintain a connected overlay graph in the presence of multiple heterogeneous substrate networks. To show that self-organizing networks can deliver high delay and throughput performance at the lower layers of the system architecture, we propose the development of an `overlay switch' with hardware support for self-organizing overlay network protocols. We will explore algorithms that enable overlay networks to scale to millions of nodes. In addition, we will devise and evaluate algorithms for overlay networks that result in an efficient use of network resources.