Search

High resolution direction of arrival estimation for closely spaced target in low signal to noise ration (SNR) is an active research area in underwater acoustic and SONAR signal processing The resolution performance of classical/traditional subspace based algorithms including MUSIC and ESPRIT procedures degraded under sever scenarios of low signal to noise ratio (SNR), less number of samples/snapshots and short distant targets. In this work, optimization courage of swarming intelligence of grey wolf optimization (GWO) is carried out for viable DOA estimation in the challenging instances of worst underwater scenarios using a uniform linear array (ULA). The high resolution DOA for short distant emitters is achieved using less number of samples vitally with GWO by inspecting the global minima of the highly nonlinear objective model of ULA having multi local minimas. The performance is analyzed for different number of targets employing estimation accuracy, resolution certainty, variability analysis, frequency distribution of RMSE, robustness against different SNR of additive-white Gaussian measurement noise and comparative studies with MUSIC and Particle Swarm Optimization PSO with the validation strength of Cramer Rao bound reveals the valuation of the scheme. [ABSTRACT FROM AUTHOR]

A novel intelligent computation mechanism is proposed for the estimation of channel state information in MIMO architecture. In this computing technique, the global optimizer Pattern Search algorithm is used to find the accurate channel state information which is based on coefficients of the channel matrix. The pattern search algorithm is exploited for the square channel matrix which consists of an equal number of array elements at both transmitter and receiver end of the communication system. Rayleigh fading channel modeling is explored in this work for evaluating the response of received signal over channel matrix of different dimensions. Also, two cases of measured white Gaussian noise are applied for evaluating the trend of algorithm which are without noise and 20 dB of noise. A Mean Square Error based evaluation function is formulated for assessing the true response of channel coefficients with the estimated one. This error function is effective enough to reach the best coefficient with only one symbol. The efficiency and performance of the pattern search algorithm are examined with different types of analysis like accuracy, convergence, and computational complexity on the bases of statistical parameters. A sufficient number of absolute Monte-Carlo simulations are practiced for the extensive statistical study of the proposed algorithm. [ABSTRACT FROM AUTHOR]