Estimation of lfm radar target parameters with high resolution using Deterministic compressed sensing in noisy environment

Abstract

ABSTRACT newlineMost of the contemporary radar systems use wide newlinebandwidths, normally ranging from hundreds of MHz to GHz, so as newlineto attain acceptable range resolution. The conventional Matched newlineFilter (MF) needs the signal to be sampled at or above the Nyquist newlinerate which is twice the signal bandwidth. Thus the radar receiver newlinedemands high-cost, high speed analog to digital converters (ADCs). newlineBesides, vast sampled data produced leads to increase in storage newlinespace and power consumption. newlineIn addition pulse compression MF produces very high newlinesidelobes due to the correlation function between transmitted and newlinereceived signals which may mask weak targets or might cause false newlinealarms. Furthermore, using MF both the time delay and Doppler newlineshift cannot be accurately measured due to time frequency newlineunambiguities. newlineThis research work considers the Compressive Sensing (CS) newlineframework for radar system design, replacing the MF at receiver newlineand reducing the ADC bandwidth from Nyquist rate to information newlinerate, simplifying the hardware design. Compression while sensing newlineitself leads to reduction in data generated/stored and reduces the newlineneed for high rate ADC s. newlineThe research accelerating factors in CS are potential to newlinereconstruct from minimum number of measurements, performance newlineguarantees, speed, complexity and noise robustness. The thesis newlinemain aim is to propose CS approach for the estimation of radar newlinetarget parameters such as delay and Doppler shift with high newlineresolution using minimum number of measurements possible yet newlineachieving noise robustness. newlinei newline newlineAndhra University, Visakhapatnam newlineCS permits the recovery of a sparse signal from far less newlinenumber of samples measured from received echo. The prerequisite newlineof CS for accurate recovery is that the measurements need to be newlineacquired randomly. Random matrices have low coherence with any newlinedictionary, but as these are unstructured their elements are highly newlineuncertain which require huge memory and are costly to implement newlinein hardware. newlineA computationally fast and efficient DFT deterministic matrix newlineis construct