UDM-RBM-Based DOA Estimation in Alpha Unstable Impulse Noise and Multipath Signals
Harikrushna Gantayat, Trilochan Panigrahi, Pradyumna Patra · 2022 6th International Conference on Electronics, Communication and Aerospace Technology · 2022
The source signals' Direction of Arrival Estimation (DOAE) has been one amongst the critical issues in sensor array processing. Numerous high-resolution DOAE mechanisms have been propounded; however, in these frameworks, a general issue noted is that the additive noise is assumed to be Gaussian distributed. To overcome these issues, this paper proposes a novel Uniform Distribution Modified Restricted Boltzmann Machine (UDM-RBM)-centric DOAE model. Here, a signal model is selected initially; also, the Covariance Matrix (CM) of an array is attained. Subsequently, for all frequencies, a steering vector grounded on the Direction of Arrival (DOA) and Short Time Fourier Transform (STFT) vector formulation is computed. After that, by utilizing Average Mean Modified Gaussian Naive Byes-Feature Mapping (AMM-GNB-FM), both values are given for feature mapping 1. Conversely, as of the input signal, the noise is found; then, it is injected into the impulse noise vector function. This vector is given for feature mapping 2. Now, both feature map 1 and 2 are amalgamated; then, utilizing Chebyshev Distance-centric Ebola Optimization Algorithm (CD-EBOA), they are given to feature dimensionality reduction. After that, utilizing UDM-RBM, the reduced features are given to train the classifier for DOAE. Hence, the experiential outcomes display an efficient DOAE for the proposed mechanism when analogized with the prevailing approaches.