Cognitive Inference and Signal Processing in Adversarial Uncertain Environments

为了收集数据、处理数据并使其易于获得广泛的潜在应用，各种网络技术正在以高度分布式的架构出现，以应对数量和类型不断增加的异构传感器。提出的研究计划（RP）的目标是发明和分析用于处理数据的认知分布式算法。传统方法在实践中经常失败，因为它们在检测、分类、分析、汇总、融合和保护网络信号和信息时没有考虑不确定性和对抗环境（u&ae）。认知使系统能够随着环境的变化而智能地、动态地改变、重新配置和增强自己。我们的目标是开发一种新的系统范式来解决u&ae中遇到的基本认知问题，例如设计分布式复合决策。我们将解决在特定领域遇到的挑战，包括：认知无线通信、认知无源雷达、压缩感知和稀疏信号处理、心房颤动映射、机器学习、超光谱成像和递归贝叶斯估计。这项研究的成果是以新方法的形式来增强对环境的传感和监测，例如使用多种类型和数量的分布式传感器（如雷达、相机、电极）对源和物体进行定位和检测。******该RP将以先进和低成本的认知信号处理算法的形式产生解决方案，并发明新的方法，使关键系统能够在u&ae中学习和认知操作。这些解决方案将应用于能源、通信、汽车、雷达、安全、生物医学、金融和大数据等各个行业。这些部门的进步将对加拿大的经济和就业产生重大影响，因为相关部门每年产生大量收入。加拿大的工业也将从RP的成果中受益，因为认知范式促进了更有效的资源利用，而他们在提高效率方面的成功将使他们更具竞争力。此外，6名博士、4名硕士和5名本科生将接受前沿技术培训，以帮助满足该领域日益增长的专业需求。******这项研究的及时性可以从最近智能、机器学习和信号处理领域的多个行业创造的大量就业机会中得到证明。HQP培训，以填补这些必要的角色是至关重要的，以支持在加拿大的商业有利可图的投资。

To collect data, process and make it available and readily accessible for a wide range of potential applications, various cyber technologies are emerging in a highly distributed architecture for increasing numbers and types of heterogeneous sensors. The goal of the proposed research program (RP) is to invent and analyze cognitive distributed algorithms for processing data. Traditional methods often fail in practice because they are not designed to take into account uncertain and adversarial environments (U&AEs) in detecting, classifying, analyzing, summarizing, fusing and protecting signals and information over networks. Cognition allows the systems to intelligently and dynamically alter, reconfigure and enhance themselves as the environment changes. We aim to develop a new systematic paradigm for solving fundamental cognitive problems encountered in U&AEs such as designing distributed composite decision making. We will resolve challenges encountered in specific areas including: cognitive wireless communication, cognitive passive radars, compressed sensing and sparse signal processing, atrial fibrillation mapping, machine learning, hyper-spectral imaging and recursive Bayesian estimation. The outcomes of this research are in the form of novel methods for enhanced sensing and monitoring of the environment such as localization and detection of sources and objects using multiple types and number of distributed sensors such as radars, cameras, electrodes. ******This RP will yield solutions in the form of advanced and low-cost cognitive signal processing algorithms and invent new methods to enable critical systems to learn and cognitively operate in U&AEs. These solutions will have applications in various industries such as: energy, communication, automotive, radar, security, biomedical, financial and big data. Advances in these sectors will have a significant impact upon the Canadian economy and employment, as relevant sectors generate significant annual revenue. Canada's industries will also stand to benefit from the outcomes of this RP because the cognitive paradigm promotes more efficient resource use, and their success in increasing efficiency will make them more competitive. In addition, six PhD, four MSc and five undergraduate students will be trained in leading-edge technology to help meet the growing demand for expertise in this area. ******The timeliness of this research is evidenced by recent massive job creation by multiple industries in the areas of intelligence, machine learning, and signal processing. HQP trained to fill these necessary roles are critical to supporting profitable investment in Canadian business.