Multiple Emitter Geolocation Using Joint Maximum Likelihood
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing geolocation techniques struggle to accurately determine the positions of multiple radio frequency emitters due to noisy bearing measurements and the limitations of conventional maximum likelihood estimators and the Stansfield approach, particularly in scenarios with closely spaced and unidentified targets.
Innovation Solution
A joint maximum likelihood optimization procedure is applied to simultaneously solve for multiple target locations using a search grid and expectation maximization processes, incorporating association probabilities and weighted least squares to refine geolocation estimates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional maximum likelihood estimators or Stansfield approach are used for single target geolocation, then the geolocation problem can be solved, but the accuracy deteriorates for multiple targets and closely spaced emitters
Solution Approach 1:
The patent segments the multiple target geolocation problem into iterative single-target subproblems. By using Expectation Maximization (EM) algorithm, the system alternates between estimating target locations (E-step) and associating bearing measurements with targets (M-step), effectively dividing the complex multi-target problem into manageable iterations that converge to a solution
Solution Approach 2:
The patent combines multiple bearing measurements from different sensors and time instances into a unified geolocation estimate. The EM algorithm merges information from all measurements collectively, considering their associations with different targets, to produce accurate geolocation results for multiple emitters simultaneously
2Ease of operation
If bearing measurements are intersected to determine emitter location, then geolocation can be obtained, but noisy measurements lead to significant error
Solution Approach 1:
The patent implements feedback through the iterative EM algorithm. In each iteration, the system refines target location estimates based on current associations, then uses these refined estimates to improve measurement associations in the next iteration. This feedback loop continues until convergence, progressively reducing errors from noisy measurements
Solution Approach 2:
The patent changes the parameter representation from direct bearing intersection to probabilistic association weights. Instead of simply intersecting noisy bearing lines, the system uses probability weights to represent the likelihood of each measurement belonging to each target, transforming the problem into a statistical optimization that is more robust to noise
3Device complexity
If Stansfield approximation fn≈sin fn is used, then the nonlinear geolocation problem is reduced to linear least squares, but significant bias error occurs
Solution Approach 1:
The patent uses a dynamic iterative approach rather than a static approximation. The EM algorithm adaptively adjusts target location estimates and measurement associations through multiple iterations, allowing the solution to evolve toward the true locations. This dynamic process captures the nonlinear relationships accurately without requiring linearizing approximations that introduce bias
Data Source
AI summary
Techniques for jointly estimating geolocations of a plurality of emitting targets. An example method includes acquiring a plurality of bearing measurements, establishing a search grid based on the bearing measurements, performing a joint maximum likelihood optimization procedure on the bearing measurements over the search grid to produce an estimated geolocation solution by solving a nonlinear multi-target geolocation description function for all targets associated with a current rank, repeating the joint maximum likelihood optimization procedure set successive ranks until a maximum rank is reached, evaluating the estimated geolocation solution sets for each rank to determine a number of targets, based on the number of targets, selecting the corresponding geolocation solution set for the rank corresponding to the number of targets, and processing the selected corresponding geolocation solution set to produce a geolocation result that includes an estimated location of each target.