Multiplier Offset Voltages in Adaptive Arrays

The effects of multiplier offset voltages in adaptive arrays are examined. Multiplier offset voltages arise when active circuits are used to implement the error-by-signal multipliers required in an array based on the LMS algorithm. These offset voltages are known from experimental work to have a strong effect on array performance. It is first shown how multiplier offset voltages may be included in the differential equations for the array weights. Then their effect on weight behavior is studied. It is found that the offset voltages affect the final values of the weights, but not the time constants. Furthermore, the effect they have is influenced by the amount of element noise in the array. An adequate amount of noise is necessary to minimize weight errors due to offset voltages. An example is treated to show the effect of offset voltages on the final array weights and the output signal-to-noise ratio (SNR). With offset voltages present, it is found that there is a maximum SNR that can be obtained from the array. A specific input SNR is required to obtain this maximum output SNR. Finally, it is shown that a finite operating range for the weights places a further restriction on the acceptable values of offset voltages and noise.     

Equivalent transformation for a partially adaptive array

A partially adaptive array is one in which elements of a phased array are controlled or adaptively weighted in groups or in which certain elements, called auxiliary elements, are made controllable. Mathematically, this type of array is formed by transforming all of the elements of an array by a nonsquare matrix such that the resulting output vector has a length less than the number of array elements. It is shown that there is an equivalent matrix transform that can effectively be utilized in analyzing the partially adaptive array's performance when a small number of external jammers are present. Processor implementation and convergence rate considerations lead to the desirability of reducing the dimensionality of the cancellation processor while maintaining good sidelobe interference protection. A meaningful measure of canceller performance is to compute the optimal output signal-to-noise ratio. This expression is a function of the jammer, direction-of-arrival vectors (DOAVs), jammer powers, the array steering vector, and internal noise. It is shown that if this expression is computed for the fully adaptive array then it is easily computed for the partially adaptive array by transforming the jammer DOAVs and the steering vector by the orthogonal projection matrix defined by the rows of the subarray transformation matrix and substituting these vectors back into the original expression for the fully adaptive array     

Postfiltering in Digital Beamformers

Assuming a sinusoidal signal superimposed on a narrow-band Gaussian noise as the input to a receiving array, the output power and signal-to-noise ratio of a digital beamformer with postfiltering were formulated so that subsequent calculations could be made without an analysis in the frequency domain. The formulation utilized the quantizer functions previously given by the author and certain spectral power distribution factors originally attributed to Davenport but more rigorously derived and discussed in the present work. A numerical study based on this formulation for a DIMUS array in a correlated noise field reveals that except for certain rare circumstances, postfiltering generally improves the output SNR or array gain. It is demonstrated that the amount of postfiltering gain not only varies with array input SNR but also depends strongly upon the spacing-to-wavelength ratio, and its meaningful interpretation can only be made in conjunction with both the clipping and noise correlation losses. In particular, balancing postfiltering gain against the two losses suggests that receiving arrays with element spacings smaller than one-half of the operating wavelength may be used to the advantage of system design under certain conditions.     

Theory of Adaptive Radar

This paper reviews the principles of adaptive radar in which both the spatial (antenna pattern) and temporal (Doppler filter) responses of the system are controlled adaptively. An adaptive system senses the angular-Doppler distribution of the external noise field and adjusts a set of radar parameters for maximum signal-to-interference ratio and optimum detection performance. A gradient technique for control of the radar array/filter weights is described and shown to generate weights which asymptotically approach optimum values. Simulation results illustrate the convergence rate of adaptive systems and the performance improvement which can be achieved.     

Effect of Single-Bit Digitization in Adaptive Array Control Loops

The properties of an adaptive array antenna, including transient response rate and control loop noise, depend on the intensity of the external noise field. The dependence can be reduced by 1-bit digitization of the real and imaginary parts of the signals from which the envelopes in the control loops are formed. This can be done without degrading the performance of the adaptive array.     

Effect of Envelope Limiting in Adaptive Array Control Loops

The properties of an adaptive array antenna, including transient response rate and control loop noise, depend on the intensity of the external noise field. This dependence can be reduced by envelope hard limiting in the control loops, without degrading the performance of the adaptive array.     

The Effect of Integrator Pole Position on the Performance of an Adaptive Array

The LMS adaptive array requires an integrator in each weight feedback control loop. In practice the integrator is often replaced by a low-pass filter, i.e., by a filter with a single pole at s = - ? (where s is complex frequency). The effect of this pole position on array performance is examined. It is shown that to obtain optimal performance from the array, ? must be less than k?2, where k is the loop gain and ?2 is the thermal noise power per element. When at exceeds k?2, the output signal-to-inter ference-plus-noise ratio from the array is degraded for intermediate values of interference power.     

Specification and Performance of Passive Sonar Spectral Estimators

The application of existing estimation theory to the problem of specification and performance of passive sonar spectral estimators is considered. The classification function is addressed, so that the signal is assumed to be present, and so that the energy arrival angle is known. The spatial filter considered is a line array of M equally spaced omnidirectional hydrophones. Signal and ambient noise are both zero-mean, wide-sense, stationary Gaussian random processes that differ in their spatial correlation across the face of the array. The signal is a plane wave that can be made totally spacially corrected between array elements by inserting delays between sensors to invert the signal propagation delay. The noise correlation is a function of frequency, bandwidth, element separation, and the relative time delay between sensors. Under these assumptions, the Cramer-Rao lower bound is derived for the class of unbiased estimates of signal power in a narrow frequency band at the hydrophone in the presence of correlated ambient noise of known power. The bound is examined numerically, resulting in a threshold phenomenon with M that constitutes a new design consideration. In addition, there is a striking insensitivity to realistic values of ambient noise correlation, and there are ranges in signal-to-noise ratio for which one gains more by increasing M than by increasing the bandwidth-time product. Specific processors, including a new unbiased estimator when noise power is unknown, are developed.     

Angle Estimation with Adaptive Arrays in External Noise Fields

A method of estimating the angle of arrival of a signal at an array of sensors in an external noise environment is outlined. The development is based on a maximum likelihood estimator and leads naturally to adaptive sum and difference beams which null the external noise sources. An algorithm for estimating angle of arrival, based on the outputs of adaptively distorted sum and differnce beams, is shown to perform well in the presence of sidelobe and/or main beam interference.     

Range and Bearing Estimation in Passive Sonar

Passive sonar systems are used for estimating the range and bearing of signal sources, such as ships or submarines. In this study, the Cramer-Rao bounds on estimation errors are used as measures of the accuracy of the estimates. The bounds show how parameters such as observation time, signal bandwidth, signal-to-noise ratio, and array geometry can be chosen to obtain maximum accuracy. When the array geometry satisfies certain conditions, the bound for range estimate is shown to be independent of the actual source bearing and the bound for bearing estimate independent of both the range and bearing of the source. It is also shown that the same conditions on array geometry ensure that the range or bearing estimation accuracy is not degraded when tne other pammeter is not known.     

On Optimizing Array Reception of Multipath

This paper reviews system configuration requirements and analyzes detectability performance characteristics for maximum likelihood array reception of multipath. Performance is analyzed to determine the effects of channel multipath structure (multipath delay and signal power division among the paths), space-time correlation properties of the incident processes, and the array spacing. It is shown by a series of case studies, that for single element coupling, as well as array coupling, an increased multipath delay factor results in decreased system detectability for fixed signal and noise intensity levels. The performance capacity is degraded as the available signal power tends to distribute more uniformly between the paths. These effects are attributed to the loss of effective signal energy concentration, resulting in a lower effective pre-detection signal-to-noise ratio. An investigation of the effects upon system performance, due to array element spacing, shows that performance is enhanced by increasing the spacing relative to the multipath delay factor and the reciprocal signal bandwidth. The former is the result of a more directive detectability (beam) pattern arising from the increased spacing. In effect, with increased spacing, the main lobe of the pattern is narrowed, while the side lobes are optimally suppressed by the required noise related array element link, frequency filters (weights).     

Noise factor and antenna gains in the signal/noise equation for over-the-horizon radar

A recommended form of the signal-to-noise equation that includes both internal and external system noise and signal/noise processing losses is discussed. The recommended form conforms to the internationally accepted definition of system operating noise factor but is extended to include signal/noise processing. The predetection signal-to-noise ratio (SNR) of a radar or communication system is proportional to the power gain of the transmit antenna and the directive gain of the receive antenna, and is inversely proportional to the operating noise factor of the receiving system. The operating noise factor is approximately equal to the product of the external noise factor and the signal/noise processing factor when the system is external noise limited, as is usually the case for over-the-horizon (OTH) radar.<>     

A Superlimiting Phased-Array Receiving System in a Two-Source Environment

The performance of a uniformly spaced phased-steered line array with element channel superlimiting is studied for far-field sources consisting of 1) two sinusoidal signals with different frequencies and angular locations, and 2) a sinusoid and a noise signal at different angular locations. Attention is focused on the nonlinear case where internal noise is negligible compared to both input signals. The analysis for the two-sinusoid case gives the precise frequencies, positions, and amplitudes of all apparent sources. In addition to the two active sources, the array output has an array of images arranged symmetrically in sine space about the larger input, at intervals equal to the spacing between the two active sources. For the case of a separated sinusoid and a noise source, the analysis shows that the angular positions and average powers of the array outputs duplicate the double-sinusoid results, but the images have noise-like spectra. The analyses are confirmed by experimental results obtained with a 60-element superlimiting X-band array.     

Adaptive Interference Cancellation in Nyquist Rate Scanned Arrays

A concept for adaptive directional filtering in a Nyquist rate scanned array based on time-domain sampling in the array sum channel is presented. By this method the adaptation rate may be independent of the scan rate in contrast to arrays with adaptive weight setting in the array element channels.     

Adaptive Radar Beacon Forming

Propagation errors along paths between an array radar and a distribution of targets cause degradations in angle measurements and detection range. The overall objective of the research described in this paper was to analyze and demonstrate the use of conjugate reflections for compensating adverse effects of path errors. The effect of reflecting the conjugate of an incident wave is described mathematically and is demonstrated by computer simulation. Repeated conjugate reflections are shown to result in the formation of a single beam usually focussed on a target highlight. Echoes from this spatial reference, or "beacon" are shown to provide the means by which aperture phase errors may be effectively compensated. Results of radar simulations include two-way patterns computed for an example involving a distribution of three-point targets and half-wave-length Gaussian aperture errors. Without compensation a gain loss of 12 dB is computed; with error correction, based on echoes from an adaptively focussed beam, the two-way pattern is within a small fraction of a dB of the ideal pattern. The effect of noise on adaptive beacon forming was considered for a case involving one target. Repeated conjugate reflections improve signal-to-noise ratio as long as the effect of noise is less than the effect of aperture dephasing on the power reflected back to the target. An example is presented in which signal-to-noise ratio at the output of the receiver combining network is increased from 4 to 11.8 dB.     

Probability of Detecting a Fluctuating Target Immersed in Both Noise and Clutter

The probability of detecting either a Swerling 1 or Swerling 2 target immersed in both Rayleigh-distributed noise and log-normally distributed clutter is calculated. Results are presented which demonstrate the effect of noise-to-clutter ratio, signal-to-noise ratio, and number of pulses integrated on the detection statistics.     

Instrumental variable adaptive array processing

An instrumental variable (IV) approach is presented for estimating the weights of an adaptive antenna array. Theoretical analysis of the IV method shows that the antenna gain weights are independent of finitely correlated noise, so that unbiased estimation of signal arrival angles is possible. Only matrix inversions are required to compute the weight estimates. In this sense, the IV method provides performance comparable with eigenvector techniques but with lower computational burden. Both minimal and overdetermined IV estimators are derived. The overdetermined estimators give the same theoretical array weights as minimal estimators, but yield more accurate weight estimates in real data situations. Simulation results are presented to compare these IV methods with one another and with conventional matrix inversion weight estimators. In these examples it is seen that IV methods are able to resolve closely spaced interference sources when conventional matrix inversion techniques cannot. It is also shown that overdetermined methods are capable of providing weight estimates with lower variances than those of minimal methods     

Sonar Array Detection of Gaussian Signals in Gaussian Noise of Unknown Power

A statistical test is postulated for detecting, with an M-element hydrophone array, a Gaussian signal in spatially independent Gaussian noise of unknown power. The test is an extension of the uniformly-most-powerful (UMP) unbiased test for a two-element array. The output signal-to-noise ratio of the test is calculated and, for a large number of independent space-time samples, is shown to be no better than a mean-level detector (MLD). Receiver operating characteristic curves (ROC) for the MLD are computed and compared to the ROC curves for the optimum (Bayes) parametric detector. The input signal-to-noise power ratios required to provide a detection probability of 0.5 differ by less than 0.2 dB for a fifty-element array with wide variation in false-alarm probability and time-bandwidth product. This result suggests that both the extended bivariate UMP unbiased test and the MLD perform close to the unknown UMP unbiased test for independence of a multivariate Gaussian distribution.     

Adaptive array CFAR detection

Presented here is a large class of adaptive array detection algorithms with constant false alarm rate (CFAR), so that the false alarm rate can be set to any preassigned number without knowledge of the noise covariance matrix. This class map incorporate any usual method of cell averaging and any method for array weight vector synthesis. A sufficient condition for CFAR is derived, which is easy to satisfy in practice. Basic system parameters are discussed. An example of detection performance for a simple cell-averaging detector, in which the array weight vector is synthesized by the method of diagonal loading, is provided using Monte Carlo simulations     

Rapid Convergence Rate in Adaptive Arrays

In many applications, the practical usefulness of adaptive arrays is limited by their convergence rate. The adaptively controlled weights in these systems must change at a rate equal to or greater than the rate of change of the external noise field (e.g., due to scanning in a radar if step scan is not used). This convergence rate problem is most severe in adaptive systems with a large number of degrees of adaptivity and in situations where the eigenvalues of the noise covariance matrix are widely different. A direct method of adaptive weight computation, based on a sample covariance matrix of the noise field, has been found to provide very rapid convergence in all cases, i.e., independent of the eigenvalue distribution. A theory has been developed, based on earlier work by Goodman, which predicts the achievable convergence rate with this technique, and has been verified by simulation.