CAREER: Next Generation Network Coding: Distributed Design Via Coded Feedback

互联网和无线网络对异质服务的需求呈指数级增长，这对下一代网络的设计提出了新的要求和巨大的挑战，以实现更高的吞吐量、更好的抗攻击能力和更强的敏感信息安全性。最近，一种新的网络编码范式已经出现，它可能以最优的方式同时实现上述目标。然而，为了充分实现其预期的好处，需要一种全新的网络编码设计方法，在实现最佳性能和保持低复杂性的同时，对不断变化的网络环境做出快速反应。为此，研究人员研究了编码反馈，它通过结合网络编码独特的几何结构来概括确认非编码流量的传统反馈机制。本研究首先探讨了编码前向包和反馈包之间的联合代数性质，为后续开发新一代网络编码的高效网络协议奠定了基础。特别是，研究人员对编码反馈的作用感兴趣，编码反馈将非编码网络资源分配算法转换为面向编码的网络资源分配算法。然后，编码反馈携带的丰富信息被用于设计协议，在自适应网络变化的同时提供增强的全局和局部错误保护和数据安全性。该方案的理论研究紧跟实际情况，包括无线网络的动态信道质量和干扰，以及协议的复杂性。系统级实现被用来验证和进一步提炼该方案的理论结果。编码理论、图论和网络优化的协调方法为普渡大学的学生提供了有价值的跨学科培训。

The exponentially increasing demands for heterogenous servicesover Internet and wireless networks have imposed new urgencyand great challenges for designing next-generation networksthat achieve higher throughput, better resiliency to adversaryattacks, and stronger security for sensitive information.Recently, a new network coding paradigm has emerged, whichpotentially achieves the above goals simultaneously in anoptimal fashion. Nonetheless, to realize fully its promisedbenefits requires a clean-slate methodology of network codedesign that achieves optimal performance and maintains lowcomplexity while reacting quickly to the ever changing networkenvironment.To that end, the investigators study coded feedback, whichgeneralizes the conventional feedback mechanism thatacknowledges non-coded traffic by incorporating the uniquealgebraic structure of network-coding. This research firstexplores the joint algebraic properties between coded forwardand feedback packets, which are the basis for the subsequentdevelopment of new, efficient network protocols for nextgeneration network codes. In particular, the investigators areinterested in the roles of coded feedback that transform thenon-coding network resource allocation algorithms to theircoding-oriented counterparts. The rich information carried bycoded feedback is then used to design protocols that offerenhanced global and local error protection and data securitywhile self-adapting to network changes. The theoretic studiesof this program follow closely practical considerationsincluding the dynamic channel quality and the interferencenature of wireless networks and the complexity of the resultingprotocols. System-level implementation is used to verify andfurther refine the theoretic findings of this program. Thecoordinated approaches of coding theory, graph theory, andnetwork optimizations provide valuable interdisciplinarytraining for students with Purdue University.