Direction-of-Arrival Estimation for Deterministic Networks

Direction-of-arrival (DOA) estimation is a key physical-layer technique for guaranteeing the stringent latency and reliability requirements of emerging deterministic networks. This paper investigates a generalized dual-subarray linear array model in which the inter-subarray displacement vectors can be arbitrarily specified, and devises two subspace-based DOA estimation schemes tailored to this setting. The first scheme is a spectral-search (SS) estimator that exploits the structured relationship between the signal subspaces of the two subarrays via a parametrized phase-rotation matrix, where the DOAs are obtained by searching over angles that induce a rank deficiency in a residual matrix constructed from the estimated signal subspace, yielding a high-resolution spectral function analogous to, but more flexible than the conventional schemes. The second scheme is a search-free (SF) estimator that, under mild geometric assumptions on the linear array, reformulates the same criterion as a polynomial in a complex exponential variable and recovers the DOAs from the roots closest to the unit circle, thereby eliminating grid search and significantly reducing computational complexity. Simulation results are provided for deterministic-network scenarios to show that, for a 2 × 5-element dual-subarray and moderate SNRs, the proposed SF-based estimator achieves an RMSE below 1 while conventional MUSIC exhibits an RMSE around 10, demonstrating roughly an order-of-magnitude accuracy gain and confirming the superior performance and scalability of the proposed SS and SF schemes.