CAREER: Towards Realizing Timely Information Transfer and Processing for Networked Communication Systems

随着当今世界交换的数据量不断增长，需要有一种可靠的方法来及时处理和在最终用户之间传输信息。受评估信息新鲜度的新型信息年龄（age-of-information, AoI）度量的启发，该提案提供了在联邦学习（FL）、云计算和遥感系统中实现数据及时交换的途径。它引入了一种新的信息时效性度量，AoI，到这些系统的研究中，并研究了在网络上实现及时学习、计算和感知的各种方法。它有潜力成为各种应用的推动者，包括协作式自动驾驶、精确医疗保健等。作为本研究计划的一个组成部分，提出了一个广泛的教育计划，涉及K-12，本科和研究生水平的议程。这个项目涵盖了三个基本的信息处理类别。第一种涉及FL设置中的智能代理。提出了新的智能体调度算法和高效的模型压缩算法，以实现实时FL的概念。第二种是考虑云计算系统中的客户，在系统级约束下设计了创新的作业上传策略，以实现作业的及时完成。最后，第三类聚焦于遥感，提出了新的可变长度编码方案，以实现在噪声通信信道上的及时多进程跟踪。这个项目有可能改变信息获取和处理系统在严格的通信和资源限制下的运作方式，而教育计划提供了必要的工具来推断网络系统中信息的及时性。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

With the ever-growing amounts of data exchanged in today's world, there needs to be a reliable means for timely information processing and transfer between the end users. Inspired by the novel age-of-information (AoI) metric that assesses information freshness, this proposal offers a pathway towards realizing the ex- change of data in a timely manner in federated learning (FL), cloud computing, and remote sensing systems. It introduces a novel information timeliness metric, AoI, into the study of these systems, and investigates various approaches to achieve timely learning, computing and sensing over networks. It has the potential to become an enabler for various applications, including collaborative autonomous driving, precision health- care, and others. As an integral part of this research proposal, an extensive educational plan that involves K-12, undergraduate and graduate level agendas is proposed.This project spans three fundamental information processing categories. The first involves intelligent agents in a FL setup. Novel agent scheduling and efficient model compression algorithms are proposed for this category to realize the notion of timely FL at-scale. The second considers clients in cloud computing systems, for which innovative job uploading policies are devised under system-level constraints to achieve timely job completions. Finally, the third category focuses on remote sensing, and presents novel variable length coding schemes to enable timely multi-process tracking over noisy communication channels. This project has the potential to transform the way information acquisition and processing systems operate under stringent communication and resource constraints, while the educational plan provides the necessary tools to reason about the timeliness of information in networked systems.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.