A Continuous Optimization Framework for Joint Pulse Radar Signal Association and Localization in Cluster Reconnaissance

A critical challenge in collaborative cluster reconnaissance is effectively associating pulses among various reconnaissance units. Conventional time-difference histogram methods have limitations under operational constraints such as long distances between units, high pulse repetition frequency signals, and less-than-ideal emitter positioning. These methods show sensitivity to histogram binning configurations and suffer from reduced association accuracy. This study proposes a unified framework for jointly resolving the tasks of pulse association, sorting, and localization. To overcome the computational inefficiency in discrete sequence search methodologies, we reformulate the joint optimization problem using continuous relaxation techniques. This results in a non-convex nonlinear programming model that we solve through gradient descent optimization. Simulation results demonstrate that our continuous optimization algorithm performs comparably to discrete optimization approaches while reducing the average running time from 1.1 seconds to <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$4.7-6$</tex> milliseconds under the same conditions. This significant reduction greatly enhances the real-time applicability of the method.