Science of Tracking, Control and Optimization of Information Latency for Dynamic Military IoT Systems
Abstract
Project SummaryApproved for public release. The military Internet of Things (IoT), that interconnects a plethora of devices ranging from sensors to autonomous ground, air, and sea vehicles, will provide significant competitive advantages in tomorrow~s battlefields and military systems. However, the effective deployment of a military IoT hinges on its ability to provide low-latency communicationacross its devices. In an IoT, latency can no longer be measured as a simple packet delay, as done in conventional communication systems. Instead, it must be quantified as a measure of ~information freshness~, using the so-called age of information (AoI). Although the AoI has recently attracted attention, existing works cannot meet a number of IoT-specific challenges: 1) Information agesdifferently and dynamically across a distributed IoT, due to the system~s heterogeneity and dependence of its information on physical dynamics. Using a single notion of AoI, as done in prior art, to model such a multi-modal aging process is simply not adequate, 2) Latency or AoI in an IoTsystem cannot be decoupled from the associated value of information and latency costs of information collection, 3) Information freshness in an IoT can be correlated and synchronized across devices that share a common physical medium, 4) Minimizing AoI in an IoT must be cognizantof the massive scale of the system and the unique, time-varying per-device physical dynamics, and 5) A military IoT is vulnerable to security threats that can significantly increase the AoI of its information transmission and collection processes. To overcome these challenges, the goal of this MURI effort is to develop a foundational framework for guaranteeing low latency and informationfreshness in a military IoT by introducing a fundamentally novel notion of ~multi-mode AoI~ that is cognizant of the dynamic and heterogeneous information aging processes in a military IoT, and, then, developing a suite of new tools for characterizing, optimizing, and enhancing this multi-modeAoI in large-scale military IoT systems. Borne out of a collaboration between researchers in wireless networks, information theory, signalprocessing, game theory, optimization, computer science, and security, this research will result in several latency-centric innovations: 1) Novel fundamental latency metrics, using the notion of amulti-mode AoI, that can quantify how different military IoT information ages differently according to heterogeneous dynamics, inferred from physical and communication processes, 2) Efficientcentralized optimization algorithms for minimizing the multi-mode AoI while taking into account real-world IoT system sampling and wireless transmission models, 3) Fundamental modeling and characterization of the latency costs for information collection and measurements using new ideas that build on the Heiseneberg uncertainty principle, 4) Novel decentralized IoT network controland tracking solutions for optimizing the tradeoff between multi-mode AoI, value of information, and cost of information, 5) Fundamental game-theoretic framework, that advances novel tools from mean-field type game theory, to characterize optimal (multi-mode AoI minimizing) resource and information management strategies for a massive-scale IoT under physical dynamics~ constraints, 6) Reinforcement learning algorithms that integrate ideas from neural networks to enable the IoT system to learn its multi-mode AoI metrics and optimize them, 7) New security mechanisms to protect the multi-mode IoT information freshness from jamming and replay attacks, and 8) Extensivesimulations and experimental demonstrations that validate the proposed framework in two real-world IoT domains: autonomous vehicles and drone systems. Ultimately, the anticipated research outcomes will equip the DoD with a foundational science of information latency broadly applicable across multiple military applications.1
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Sep 30, 2019
- Source ID
- N000141912621
Entities
People
- Jeffrey H. Reed
Organizations
- Office of Naval Research
- United States Navy
- Virginia Tech