[Paper Review] Cognitive Dynamic Systems: A Technical Review of Cognitive Radar

Krishanth Krishnan, Taralyn Schwering+1 more|arXiv (Cornell University)|May 26, 2016

Radar Systems and Signal Processing9 references3 citations

TL;DR

This paper proposes a cognitive radar system modeled on the Perception-Action Cycle (PAC) and Fuster’s cognitive functions, integrating Bayesian filtering, dynamic programming, and state-space modeling for adaptive tracking. Simulation results show the Cubature Kalman Filter (CKF) outperforms UKF and EKF in root mean square error (RMSE), with EKF having the highest computational load due to Jacobian calculations.

ABSTRACT

We start with the history of cognitive radar, where origins of the PAC, Fuster research on cognition and principals of cognition are provided. Fuster describes five cognitive functions: perception, memory, attention, language, and intelligence. We describe the Perception-Action Cyclec as it applies to cognitive radar, and then discuss long-term memory, memory storage, memory retrieval and working memory. A comparison between memory in human cognition and cognitive radar is given as well. Attention is another function described by Fuster, and we have given the comparison of attention in human cognition and cognitive radar. We talk about the four functional blocks from the PAC: Bayesian filter, feedback information, dynamic programming and state-space model for the radar environment. Then, to show that the PAC improves the tracking accuracy of Cognitive Radar over Traditional Active Radar, we have provided simulation results. In the simulation, three nonlinear filters: Cubature Kalman Filter, Unscented Kalman Filter and Extended Kalman Filter are compared. Based on the results, radars implemented with CKF perform better than the radars implemented with UKF or radars implemented with EKF. Further, radar with EKF has the worst accuracy and has the biggest computation load because of derivation and evaluation of Jacobian matrices. We suggest using the concept of risk management to better control parameters and improve performance in cognitive radar. We believe, spectrum sensing can be seen as a potential interest to be used in cognitive radar and we propose a new approach Probabilistic ICA which will presumably reduce noise based on estimation error in cognitive radar. Parallel computing is a concept based on divide and conquers mechanism, and we suggest using the parallel computing approach in cognitive radar by doing complicated calculations or tasks to reduce processing time.

Motivation & Objective

• To develop a cognitive radar system inspired by human cognitive functions, particularly the Perception-Action Cycle (PAC) and Fuster’s five cognitive functions.
• To improve radar tracking accuracy by replacing traditional filters with advanced nonlinear filters such as CKF, UKF, and EKF.
• To reduce computational load and enhance noise resilience through risk management and probabilistic independent component analysis (PICA).
• To accelerate processing and improve system precision by integrating parallel computing via GPGPU in the radar’s perception-action loop.
• To design a new cognitive radar architecture that supports real-time, adaptive, and high-precision signal processing using parallel state-estimation units.

Proposed method

• Adopt the Perception-Action Cycle (PAC) as the core framework, modeling radar as a system that perceives the environment, processes information, acts, and updates its internal model based on feedback.
• Implement a state-space model with Bayesian filtering and dynamic programming to enable adaptive estimation and decision-making under uncertainty.
• Compare three nonlinear filters—Cubature Kalman Filter (CKF), Unscented Kalman Filter (UKF), and Extended Kalman Filter (EKF)—using simulation-based performance evaluation.
• Integrate risk management principles to optimize parameter control, especially for noise reduction and robustness in dynamic environments.
• Propose a new spectrum sensing model using Probabilistic Independent Component Analysis (PICA) to reduce noise based on estimation error in cognitive radar.
• Design a parallel computing architecture using General-purpose computing on Graphics Processing Units (GPGPU), incorporating dual state-estimation blocks and a decision-making unit within the feedback information module.

Experimental results

Research questions

• RQ1 How can the Perception-Action Cycle (PAC) be effectively modeled and implemented in cognitive radar to improve adaptive tracking performance?
• RQ2 What is the relative performance of CKF, UKF, and EKF in terms of RMSE and computational load for nonlinear radar tracking?
• RQ3 Can risk management principles enhance parameter control and noise resilience in cognitive radar systems?
• RQ4 How effective is the proposed PICA-based model in reducing estimation error and improving spectrum sensing accuracy?
• RQ5 To what extent can GPGPU-based parallel computing accelerate processing and improve precision in cognitive radar systems?

Key findings

• The Cubature Kalman Filter (CKF) achieved the lowest root mean square error (RMSE) among the three filters tested, outperforming both UKF and EKF in tracking accuracy.
• The Extended Kalman Filter (EKF) exhibited the highest computational load due to the need for analytical derivation and evaluation of Jacobian matrices at each iteration.
• The Unscented Kalman Filter (UKF) showed better performance than EKF but worse than CKF, with a moderate computational cost.
• The proposed PICA model is expected to reduce noise in cognitive radar by leveraging estimation error in space-time processing, though experimental validation is pending.
• The integration of parallel computing via GPGPU is predicted to significantly reduce processing time, increase system speed, and improve precision by enabling concurrent execution of complex tasks.
• The redesigned cognitive radar architecture with dual state-estimation blocks and a decision-making unit in the feedback loop is expected to enhance real-time adaptability and robustness.