Broadband Source DOA Estimation Based on Incoherent Signal Subspace (ISM) Methodology

In broadband source direction-of-arrival (DOA) estimation methods, the Incoherent Signal Subspace (ISM) technique enhances resolution by analyzing signal subspaces across different frequency bands. A key improvement involves integrating the modified MUSIC algorithm with data matrix conjugate reconstruction into the ISM framework. This integration not only strengthens the algorithm's resolution capability for conventional signals but also overcomes the limitation of traditional methods in handling coherent sources. Implementation-wise, this typically involves segmenting broadband signals into narrowband components using FFT, followed by subspace decomposition per frequency bin through eigenvalue decomposition of covariance matrices. The Coherent Signal Subspace Method (CSM) achieves frequency-domain energy concentration through focusing matrix construction, with performance hinging on two critical elements: focusing matrix design and focal frequency selection. Research demonstrates that employing optimal focusing matrix criteria and dynamic focal frequency strategies significantly improves angle estimation accuracy. Particularly under colored noise conditions, traditional eigenvalue decomposition-based algorithms face dual challenges of high computational complexity and strong noise sensitivity. In code implementation, the focusing matrix can be designed using techniques like signal subspace alignment or maximum energy concentration principles, often solved via optimization algorithms. To address this bottleneck, the improved TCT algorithm based on the propagator method demonstrates unique advantages. This approach directly estimates the noise correlation matrix using the propagator operator, bypassing complex eigenvalue decomposition operations to construct an efficient fast TCT focusing matrix. The innovative architecture maintains the TCT algorithm's robustness against colored noise while reducing computational complexity to practically applicable levels. From an implementation perspective, the propagator method avoids full covariance matrix decomposition by constructing a signal subspace approximation through linear operations, significantly speeding up processing. Simulation results verify that this hybrid algorithm maintains estimation accuracy while achieving remarkable computational efficiency improvements, providing a new technical pathway for real-time broadband signal processing in practical systems.