Random Forest Bulk Filter for Radar Clutter Mitigation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
High-resolution radar systems face challenges in distinguishing targets of interest from clutter due to the large number of returns, which burdens computational resources and can lead to spurious tracking results and inefficient resource allocation.
Innovation Solution
The use of a random decision forests-based bulk filter that employs classification and regression trees (CART) ensembles to filter out clutter objects while identifying targets of interest, optimizing resource allocation and improving the efficiency and reliability of radar systems by eliminating unnecessary processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high-resolution radar systems process all returned signals, then target detection capability is improved, but computational resource consumption increases significantly
Solution Approach 1:
The patent segments the signal processing task by introducing a bulk filter that divides returned signals into clutter signals and potential target signals. The bulk filter processes signals in bulk using vectorized operations, separating obvious clutter from potential targets before applying more computationally intensive tracking algorithms only to the latter, thereby reducing overall computational resource consumption while maintaining target detection capability.
Solution Approach 2:
The patent extracts and removes clutter signals from the processed signal stream using the bulk filter. By identifying and eliminating clutter returns early in the processing chain, the system prevents unnecessary computational resources from being consumed by tracking algorithms on non-target objects, thus resolving the contradiction between detection capability and resource consumption.
2Speed
If single-feature criteria are used for resolving targets from clutter, then processing speed is improved, but target identification accuracy deteriorates
Solution Approach 1:
The patent changes the parameter of feature selection by transitioning from single-feature criteria to multi-feature criteria. The bulk filter evaluates multiple features simultaneously (range, velocity, acceleration, radar cross-section, etc.) to distinguish targets from clutter. This multi-feature approach maintains processing speed through vectorized operations while significantly improving target identification accuracy by considering multiple characteristics of returned signals.
3Reliability
If computational resources are allocated to process all returned signals, then clutter rejection capability is improved, but resource allocation efficiency deteriorates
Solution Approach 1:
The patent applies preliminary action by implementing the bulk filter before the tracking subsystem processes signals. The bulk filter performs preliminary clutter rejection by identifying and filtering out clutter returns based on multi-feature analysis. This preliminary filtering reduces the number of signals that reach the tracking subsystem, improving resource allocation efficiency while maintaining clutter rejection capability through the bulk filter's specialized optimization for this task.
Solution Approach 2:
The bulk filter serves multiple functions: it performs clutter rejection, preliminary target identification, and reduces the workload for subsequent tracking algorithms. By consolidating these functions into a single multi-functional component that processes signals in bulk, the system achieves both effective clutter rejection and efficient resource allocation, as the bulk filter is optimized to handle multiple tasks simultaneously through vectorized operations.
Data Source
AI summary
A signal processing system includes a random decision forests-based bulk filter. The signal processing system further includes one or more sensors (e.g., radar systems, satellite systems) that interface with the signal processing system via a radio frequency (RF), electro-optical (EO) or infrared (IR) exchange process, a computationally inexpensive object detector, a computationally inexpensive state estimator, then the random decision forests-based bulk filter, a computationally expensive object tracker, a computationally expensive object discriminator, and a report. Further, the random decision forests-based bulk filter features a filter training optimization process.