Distributed expectation-maximization algorithm for DOA estimation in wireless sensor networks

The one‐bit diffusion for distributed beamforming over cell free wireless networks

“…We assume the desired signal to arrive at 45 • , with three interfering signals arriving at −20 • , −52 • , and 72 • . These angles can be random, but the angular resolution of the arrays is crucial for distinguishing between them [31][32][33][34][35]. According to [32] and [34], the angular resolution of the array depends on the number of its elements and the algorithm used.…”

Direction of arrival estimation for the uniform or non‐uniform noise with adaptive expectation maximisation algorithm

“…To reduce the computational complexity and improve the estimation performance, the EM algorithm can be utilised. However, as aforementioned, the DOA estimation performance of the conventional EM estimator [10–13] is very sensitive to the initial noise ratio. Thus, to deal with this problem, we propose the AEM algorithms for the UN and NUN, respectively.…”

“…complete data, and then estimate the parameters of each signal component separately. Clearly, this algorithm has been successfully used in the DOA estimations for deterministic and stochastic sources in the presence of the UN [10–13]. However, these aforementioned EM DOA estimators assume the noise power ratio for each component constant and only update DOA in the iteration process.…”

“…Obviously, the DOA estimation performance is sensitive to this initial noise ratio. More worse, it seems difficult to choose an appropriate ratio for yielding the best performance in terms of the estimation accuracy and convergence rate (see [10–13]), thus we normally choose it in a blind way.…”

“…Thus, the noise power ratio for each component would be adaptively changed in the iteration process. This is a big difference from the conventional EM algorithm [10–13], in which we have to specify the noise power ratio for each component and keep it constant in the whole iteration. More importantly, with our proposed AEM algorithm, much better DOA estimation performances can be obtained in the UN or NUN case.…”