Cooperative Signal Sorting Method Based on Multi-station TDOA and Adaptive Stream Clustering

‍ ‍Cooperative signal sorting based on multi-station time difference of arrival is an effective method for dealing with the sorting of inter-pulse waveform agility radar signals. Existing batch processing methods for multi-station time difference of arrival cooperative signal sorting lack real-time capabilities and fail to effectively mine the dynamic characteristics of radar pulse streams. This paper proposes a multi-station time difference of arrival cooperative signal sorting method based on adaptive stream clustering. The stream clustering algorithm is introduced into the realm of multi-station time difference of arrival cooperative signal sorting， exploiting its capability for continuous and rapid analysis of data to unearth the dynamic characteristics of radar pulse streams. When applied to the multi-station time difference of arrival cooperative signal sorting， it was discovered that the existing stream clustering algorithm tends to resort to increasing batch sizes in sorting in the presence of an imbalance in radar signal sample densities. Therefore， an enhancement is made to the stream clustering algorithm. By using the estimated error of multi-station time difference measurements for adaptive detection of the clusters formed by stream clustering， it was found that radar clusters were increasing in batch. Adaptive merging of these increasing-batch radar clusters effectively suppresses the occurrence of this phenomenon， thereby addressing the issue of increasing-batch in multi-station time difference of arrival cooperative signal sorting. Moreover， by utilizing multi-station time difference measurement error estimates to set clustering thresholds adaptively online， the probability of pulses being correctly assigned to radar clusters is increased， and the number of iterations required by the online clustering algorithm is decreased， thus effectively enhancing the computational speed of the algorithm. Utilizing a time-decay window model for unknown cluster detection allows for time-weighted evaluation of each data point arriving at the effective clustering repository. Early-arriving data points are given smaller weights， whereas recent arrivals are weighted more heavily. By calculating the total weighted value within the effective clustering library， the response speed for updating the clustering summary is enhanced， avoiding the failure of clustering due to an insufficient number of effective samples and improving the adaptive nature of the sorting algorithm to its environment. On the basis of the Stream Affinity Propagation algorithm and the aforementioned improvements， Adapt-Stream Affinity Propagation is proposed. Simulation results indicate that， compared to the existing multi-station time difference of arrival cooperative signal sorting algorithms， Adapt-Stream Affinity Propagation can suppress increasing-batch issues and effectively sort the correct radar signals when dealing with the imbalanced and complex inter-pulse waveform agility signals that are mixed together. Facing the evolution process of multiple complex agility radars' appearance， deactivation， and reactivation， the algorithm can precisely detect and display dynamic evolution characteristics， with results evolving in correspondence to the evolution of radar pulses. When dealing with multiple complexes inter-pulse waveform agility signals， the complexity of Adapt-Stream Affinity Propagation is significantly reduced compared to the existing multi-station time difference of arrival cooperative signal sorting algorithms， notably accelerating processing speed in high-density pulse sample scenarios， thereby better ensuring the real-time processing of multi-station time difference of arrival cooperative signal sorting. Even in situations of large multi-station time difference measurement errors and high pulse loss rates， the algorithm ensures the correctness and usability of the results， achieving a high sorting accuracy rate. Therefore， in multiple simulation comparisons， the Adapt-Stream Affinity Propagation algorithm proves to be an effective method for processing inter-pulse waveform agility radar signals in multi-station time difference of arrival cooperative signal sorting.