Physically-Coupled Cognitive Perceptual Systems
This theme consists of three applications that emphasize key requirements of physically-coupled cognitive perceptual systems. They serve as key integrator and demonstrator of technical advances pursued throughout the CONIX team. Each application was chosen to highlight unique operating points in each CONIX design dimension. The Smart and Connected Communities application highlights problems of network size, long distances, and human time-scales. The Interactive Mixed Reality Systems application focuses on short distances, tight latency, and large bandwidths. The Enhanced Situational Awareness application spans these two design points in terms of latency, bandwidth, scale, and distance, while additionally requiring complex spatiotemporal choreography.
Mixed Reality Systems
This application aims to merge physical and virtual elements in AR/VR systems with a collaborative digital teleportation application. Unlike existing AR telepresence applications, we will focus on tightly integrating physical components in the user’s space, including their own bodies and facial expressions, with the virtual world providing two-way remote interaction. This requires extremely low-latency local control mixed with low-latency or latency-masking wide-area connectivity.
Enhanced Situational Awareness
This application will focus on creating on-demand live sensory information feeds for decision-makers from a large-scale swarm of cooperating drones and ground-based autonomous systems controlled by an individual or small group of human agents. This application will highlight the ability of the CONIX distributed architecture to support real-time collaboration among humans, algorithms, and machines in a safe, robust, and secure manner in an information-rich, rapidly-evolving tactical environment.
Smart and Connected Communities
This application will look at mechanisms for managing and processing millions of sensors’ feeds in urban environments. This will involve deploying highly reconfigurable CONIX edge devices that will monitor and visualize the flow of pedestrians through urban spaces. At scale, this technology would aid public safety officials, urban planners, and stakeholders interested in infrastructure management.
Platforms, Programming and Synthesis
This theme focuses on:
- Abstractions for programmers to develop and verify high-level application specifications conveniently separable into functionality specifications and end-to-end policies governing security, privacy, reliability, and performance;
- Synthesis tools and runtimes for translating this high-level specification into a distributed program and mapping it onto heterogeneous in-network resources;
- Program transformations to help meet the latency and reliability specifications; and
- Custom heterogeneous platforms with support for accelerators, introspection, and intermittent execution.
Security, Robustness and Privacy
This theme focuses on four fundamental capabilities:
- Enforcement mechanisms supplied by trustworthy foundations,
- Resilient and secure network features that dynamically customize the network’s security posture to the current operating context,
- Secure programming systems that prevent or guide developers around the common pitfalls of secure application today, and
- Learning capabilities that inform the configuration of these policies and enforcement mechanisms that entail both understanding the system state as well as proactively mitigating advanced and strategic adversaries.
This theme focuses on supporting intelligent networks by integrating machine learning, AI, adaptive control capabilities, and spatiotemporal awareness into the system stack by:
- Refining existing and developing new and specialized machine learning approaches to automatically tune run-time system parameters,
- Develop techniques for predictable ultra-low latency and precise coordination of actions across a network using optimized positioning of data and computation, speculative computation and communication and programming abstractions and
- Developing scientific principles and design methodologies for networked control over distributed computing substrates.