Project: Prudent Sampling

Name: Prudent Sampling

TItle: Prudent Sampling for Cyber-Physical Systems

Description: In Cyber-Physical Systems the physical world influences the state of the various computation, communication, and storage processes via the sensing interface at which measurements of physical world variables are acquired and digitized for subsequent use in application-specific detection, estimation, inference, and control processes. Although its role is fundamental to the Cyber and the Physical, this interface receives little attention in system level design analysis, optimization, and verification. Typically, the interface between the continuous and the discrete halves is abstracted and idealized.

The ability to abstract the physical-world sampling interface stems from the perfect reconstruction assured by Shannon Sampling and enables designers to separate its concerns from the rest of the system. But, this relies on several implicit assumptions: that the act of sampling the physical world is cheap relative to the rest of the system, that Nyquist sampling is the best one could do, and that the performance and correctness of the rest of the system is decoupled from how the sampling interface is designed. These assumptions are often not true in practice.

Profligacy in sampling leads to a variety of energy, processing, communications, and even security bottlenecks at the system level. Our research investigates the impact on these bottlenecks of mechanisms that optimize the sampling and processing using adaptation and context-awareness while exploiting recent theoretical and embedded platform technology advances. Our goal in studying this is two-fold. First, we seek to systematically and jointly optimize and manage the various phases of the entire physical-to-cyber information-acquisition-and-processing pipeline for specific application objectives and system resource constraints. Second, we seek to develop methods to predict and validate the performance, resource requirements, and correctness of systems that make use of sophisticated sampling strategies with optimized analog, computation, communication, and storage processes.

Status: Active Project

Main Research Area: Sensor and Actuator Networks