Principal Investigator (PI): Tara Javidi, UCSD
Co-Principal Investigator (co-PI): Ashutosh Sabharwal, Rice
Co-Principal Investigator (co-PI): Urbashi Mitra, USC
National Science Foundation
Project Summary and Synopsis:
We propose a novel resiliency paradigm for NextG networks based on situational awareness achieved via Layering for Active Resilience and Awareness (LARA).
Taking a cue from resilient designs in other domains (e.g., aircrafts), we propose that a network resiliency paradigm needs to adopt the following three design principles: First and foremost, resilient systems assume that unforeseen disruptions can and will occur. Thus, the systems are designed to be vigilant, and the vigilance is achieved by sensing both the internal system state and the external environment. Second, the systems are smart in their vigilance by actively managing the resources to form a world-view that enables active preparedness and resilience. Third, armed with active vigilance, resilient systems are in a constant state of preparedness, continually adapting their view of the world by counterfactually reasoning about the system’s future – if a particular failure were to occur now, then what would be the best response?. We propose to adopt the above three principles for resilient network design, developing situationally aware networks in the process.
To achieve the resilient design, LARA design consists of three novel layers:
Layer 1: Joint Wireless Sensing and Communications We will develop new techniques to use wireless bands for imaging, sensing weather conditions and resilient geo-location. This layer's design and development is led by Prof Ashu Sabharwal, Rice University, with extensive input and contributions from the teams at UCSD and USC.
Layer 2: Active Inference for Resource Management. In this set of activities, new methods are proposed to allow for the application of reinforcement learning theory to ground meaningful algorithms to incorporate and orchestrate Layer 1 sensing functionalities to allow for global situational awareness. This set of activities will be led by Prof Ubli Mitra, USC, with extensive input and contributions from the teams at Rice and UCSD.
Layer 3: Consistency, Correctness and Counterfactuals for Reasoning. We will develop fundamental limits and algorithms for distribution estimation/testing, anamaly detection, and eventually counterfactual reasoning. The proposed Layer 3 will ensure that sensed information and learned models are consistent and timely in order to achieve active resilience. This set of activities will be led by Prof Tara Javidi, UCSD, with extensive input and contributions from the teams at Rice and USC.
Collaborators: This is a collaborative project involving UCSD, USC, and Rice. The lead PIs are Tara Javidi (UCSD), Ashu Sabharwal (Rice), and Ubli Mitra (USC). Our labs hold joint weekly research meetings. The aim is to both report on our findings and foster collaborations.Our team compromises of the following Graduate Student Researchers.
Talha Bozkus, USC
Zhican (West) Chen, Rice University
Yirong Chen, Rice University
Greg Fields, UCSD
Jianxiu Li, USC
