In congested or contested spectrum, radar is costly to operate due to high power usage, low spectral efficiency, and low covertness compared to passive sensors. For this reason, this work proposes a multi-mode radar sensing strategy, in which the sensors can choose between a monostatic radar mode and a passive electronic support measure (ESM) spectrum sensing mode. In ESM mode, a target is localized with a network of multi-mode sensors, which creates opportunities to reduce radar measurements. Radar and ESM measurements are rigorously compared using the Cram & eacute;r-Rao bound to quantify the localization error of each mode. The best mode for each sensor is chosen using a restless contextual combinatorial multi-armed bandit (RCC-MAB) online learning algorithm. The RCC-MAB increases the flexibility of the network by adapting to the target in real-time based on recent radar and ESM measurements. Two variants, the & varepsilon; -Greedy and covert RCC-MABs, were created to fulfill different tracking objectives. The & varepsilon; -Greedy RCC-MAB variant seeks to minimize the tracking error by selecting the best sensing modes based on the quality of previous measurements and the current context of the tracking filter. The covert RCC-MAB variant significantly reduces radar usage to stay covert or minimize access to a shared spectrum by only exploring radar measurements when the tracking error approaches a predefined maximum error. The & varepsilon; -Greedy RCC-MAB consistently achieved the lowest tracking error of the tested mode controllers, 58% and 16% lower than a single-mode radar and ESM network, respectively, when the radio emissions of opportunity (REO) were available during 50% of measurement opportunities. In the same scenario, the covert RCC-MAB had 55% lower tracking error than single-mode radar, while using 82% less radar than the & varepsilon; -Greedy RCC-MAB.

We propose joint radar and communications bandwidth and carrier frequency optimization and allocation (BWCFOA), and radar waveform optimization algorithm for a distributed radar network. The objective is to maximize harmonicmean of the channel capacity while ensuring that the signal-to-clutter-and-interference-plus noise ratio (SCINR) at each radar node remains above a certain threshold and the minimum range resolution requirement is attained. This optimization is nonconvex and intertwined with the problems of optimizing: a) radar waveforms, b) both radar and communications bandwidths and carrier frequencies, and c) their allocations to the nodes. The radar waveforms are optimized locally at each node utilizing manifold optimization, and the resulting outcome is utilized by the central coordinator to solve the BWCFOA problem under a framework of alternating optimization. For a given allocation, the bandwidth and carrier frequency optimization (BWCFO) subproblem is formulated as a geometric programming problem, and for the given bandwidths and carrier frequencies, the allocation sub-problem is formulated as a mixed integer (binary) convex programming problem (MICP) utilizing McCormick relaxation technique. The two sub-problems are iteratively solved until some desired convergence accuracy is achieved. Computer simulations show that the proposed waveform optimization and BWCFOA (WO+BWCFAO) outperforms random waveform with BWCFAO (RW+BWCFOA) and waveform optimization with equal bandwidth allocation (WO+EBWA) algorithms.

Over time spectrum usage has been increasing at an exponential rate. This increased usage has presented radars with increasingly congested spectral environments. Congested spectral environments can cause radars many issues such as radar processing errors due to signal distortion through saturation of the low-noise amplifier (LNA) and decreased signal-to-interference-plus-noise ratio (SINR). Excessive interference power on a receiver's front end can cause nonlinear distortion of a received waveform. A useful solution to these issues for ultra-wideband radars (UWB) are notch filters, which effectively block out known interferers prior to the LNA. Adding a notch filter prior to the LNA avoids saturation and alleviates other radar processing issues. While adding a notch filter into a UWB radar system may help block out interferers as well as their undesired effects, the notching will introduce new issues mainly when detecting a target. This presentation discusses the challenges introduced to the UWB radar system when performing target detection techniques, as well as possible solutions to alleviate this issue.

The demand for spectrum is growing and to fully utilize this limited resource dynamic spectrum-sharing (DSS) systems need to be developed and deployed at scale. In 2021, Verizon spent $45.5 billion for a subset of spectrum licenses in the C-Band [1]. This demonstrates the economic impact of increasing the number of services that can use a given frequency at the same place and time. Sensing and reconfigurable circuitry can be used in DSS systems to prevent interference between concurrent services by providing spectrum agility in four domains: time, frequency, direction, and power. Directional modulation enables a dualfunction radar communication systems in which a single antenna array can transmit both communication and radar waveforms in different directions simultaneously.

As demand for spectrum has increased, it has become a more scarce and valuable resource. Even so, the management of spectrum remains largely rigid. Consistent with the management approach, technology used to implement wireless transmitters is often rigid and unable to maximize performance while adjusting operating frequency. To move towards a future of more flexible spectral co-existence, there must be innovation in the underlying technologies. For instance, while the signals in the elements of a traditional, fixed-geometry phased array can be phase-shifted to change beam direction, there are many other parameters that are typically fixed in the design and cannot be altered during use while maintaining acceptable performance, such as frequency, bandwidth, beam pattern, and power. While sparse arrays can provide more control over array geometry, they are very application specific and may not be applicable in general use cases. Instead, a more adaptable and reconfigurable array that can change its electrical and physical properties to meet varying specifications could provide a much more flexible solution for dynamic spectrum use scenarios.

We propose joint radar and communications band-width and carrier frequency allocation (BWCFA), and radar waveform optimization algorithm for a network consisting of a central coordinator and distributed radar nodes that operate in a monostatic mode. Considering that target related information (TRI) acquired by a node, which observes low signal-to-clutter and noise ratio (SCNR), can be equally important as that acquired by the node that observes high SCNR, maximization of the minimum of the SCNRs (Max-min approach) is proposed under the constraints on minimum range resolution and communications capacity, and the available system bandwidth. This optimization is non-convex and intertwined with the problems of optimizing radar waveforms, assigning bandwidths and carrier frequencies to nodes, and determining their optimum values. However, assuming that each node allocates contiguous and non-overlapping bandwidths for its radar and communications operations, considering that SCNR is a monotonically decreasing function of bandwidth and carrier frequency, and utilizing explicit relations between bandwidth and carrier frequencies for radar as well as communications, we approximately solve the overall joint optimization problem in two steps. The radar waveforms are first optimized, and then the resulting objective function is utilized for solving BWCFA with the geometric pro-gramming (GP). Computer simulations show that the proposed waveform optimization and BWCFA (WO-BWCFA) outperforms random waveform with BWCFA (RW-BWCFA) and waveform optimization with equal bandwidth allocation (WO-EBWA) algorithms.

Tunable notch filters (TNFs) can be used in radar receivers to reject interference in crowded, dynamic spectral environments. Given many spectrum reallocations of radar bands, causing increasingly congested spectrum near radar operating frequencies, interfering signals can cause radar receiver low-noise amplifier (LNA) saturation, resulting in errors in radar signal processing. An overview of TNF technology and its benefits and challenges for application to radar systems are presented. TNFs are needed to efficiently block the high interference power levels presented to radar receivers. Additionally, impedance mismatches within the receiver chain can cause multiple signal reflections, resulting in ghost targets. Impedance matching, if included in the TNFs, could solve the problem of unwanted perceived targets due to multiple reflections. State-of-the-art TNFs exhibit passband loss ranging from 0.5 dB to 3.0 dB, notch depth ranging from 12 dB to 53 dB, 3 dB bandwidth (BW) ranging from 3% to 7%, and notch tunability ranging from 1.07:1 to 1.45:1. The commercial off the shelf (COTS) market is limited to only single-notch TNFs. Given the status of available TNF technology, multiple areas of potential innovation for TNFs related to radar system improvement are discussed.

As more radar and wireless communication systems access the electromagnetic spectrum, interference becomes more common. To increase robustness to interference, a fully reconfigurable array topology and optimization methodology is presented which can enable spectral mobility through live impedance tuning and signal equalization. Pulse-to-pulse optimizations allow efficiency maximization while maintaining array pattern integrity. This array is compatible with advanced digital beamforming techniques such as directional modulation, is highly modular, and is designed for forward compatibility. A demonstration of the proposed optimization for dual-function radar-communications is presented in simulation with a linear sixteen-element array, but it is easily scalable to larger arrays.

In this paper, we propose joint bandwidth and carrier frequency allocation algorithm for a network consisting of a central coordinator and distributed radar nodes, each operating in a monostatic mode. With an objective of enabling poor performing radar nodes, that observe low target signal-to-noise and interference ratio (SINR) values, benefit from distributed collaboration, we propose to maximize harmonic mean of the node SINRs under total bandwidth and individual node's range resolution (RR) constraints. This optimization is non-convex, but we solve it efficiently utilizing an explicit relationship between bandwidth and carrier frequencies, and the fact that each node's SINR is a monotonically decreasing function of bandwidth and carrier frequency allocated to the node. We propose an iterative method that successively approximates the optimization with a geometric programming (GP) problem. Computer simulations show that the proposed method significantly outperforms equal bandwidth allocation (EBWA) method and enables poor performing nodes to enhance their individual SINRs significantly.

Cognitive Radar Networks were proposed by Simon Haykin in 2006 to address problems with large legacy radar implementations - primarily, single-point vulnerabilities and lack of adaptability. This work proposes to leverage the adaptability of cognitive radar networks to trade between active radar observation, which uses high power and risks interception, and passive signal parameter estimation, which uses target emissions to gain side information and lower the power necessary to accurately track multiple targets. The goal of the network is to learn over many target tracks both the characteristics of the targets as well as the optimal action choices for each type of target. In order to select between the available actions, we utilize a multi-armed bandit model, using current class information as prior information. When the active radar action is selected, the node estimates the physical behavior of targets through the radar emissions. When the passive action is selected, the node estimates the radio behavior of targets through passive sensing. Over many target tracks, the network collects the observed behavior of targets and forms clusters of similarly-behaved targets. In this way, the network meta-learns the target class distributions while learning the optimal mode selections for each target class.

Congestion in the frequency spectrum is an ever-growing issue for current and future radar systems due to the increasing number of commercial wireless devices and technologies. Both communications devices and radar continue to demand access to greater swaths of bandwidth, but do not coexist well. The cognitive radio and radar communities have investigated solutions to this problem through the use of dynamic spectrum access (DSA) and spectrum sharing (SS). For radar, the pulse-agility required to effectively share the spectrum with other rapidly changing signals complicates coherent integration due to the variation in the transmitted waveform. In particular, when non-identical pulses are processed with standard Fourier-based range-Doppler (RD) processing for moving target indication, a modulation / distortion effect is induced. The Richardson-Lucy deconvolution algorithm, an image processing technique, is implemented in a software-defined radar (SDRadar) system to remove the undesired modulation from the RD images. The approach is then verified via an over-the-air experiment where the SDRadar must share a radio frequency (RF) band with a communications device and detect a moving target simultaneously.

In phased-array transmitters, array calibration is often used to correct for the magnitude and phase changes of voltage waves from the signal sources to the antennas in the different array elements. Inserting a reconfigurable impedance tuner between the power amplifier and the antenna in each element allows the range to be maximized upon changes in operating frequency or scan angle. However, tuning the impedance causes an undesirable real-time change in the magnitude and phase of the element transmission parameters. Proper assessment of this real-time change is needed to use equalization and impedance tuning to maintain the array pattern. Specifically, the antenna input currents must be monitored, and the antenna input currents can be used with the known antenna patterns to calculate the array transmission pattern. Measurement assessment of the antenna input current for impedance-tuning operations is demonstrated using a dual-directional coupler with a software-defined radio voltage measurement input. This is a practically implemented version of the approach that shows a path forward to system implementation of this “on the fly” transmitter calibration approach for arrays containing reconfigurable circuits, or for use in adjusting traditional array calibrations during operation.

Spectral efficiency and security in wireless applications can be enhanced through directional transmission of multiple signals from a single array aperture. Directional modulation allows the array to transmit multiple, directionally distinct signals out of the same aperture at the same time and frequency. This is accomplished through calculated antenna current excitations to ensure desired directional transmissions. A critical issue in this process involves the power amplifiers in each array element. Due to mutual coupling differences between array elements and the variations in power amplifier nonlinearities, signals leaving the power amplifiers may be distorted and no longer contain the desired information. This can result in bit errors in communication. To correct this issue, real-time impedance tuning and signal equalization are applied to reduce the distortion of the directionally transmitted messages. Signal equalization is corrective feedback technique similar to digital predistortion (DPD). Experimental simulation results are presented using directional modulation in an array of power amplifiers with reconfigurable load impedance tuners. The signal equalization technique applies corrective feedback to adjust the amplifier input signals, compensating for the undesired distortion such that the desired currents will be input to the antennas. Because the impedance tuning and signal equalization techniques may have conflicting effects on the overall distortion, impedance tuning is applied initially, and signal equalization is applied after the impedance tuning operation is complete.

Optimum allocation of bandwidth and carrier frequency in a network of distributed radar nodes is an important non-trivial research problem. In this paper, we propose both model- and deep learning-based joint bandwidth and carrier frequency allocation algorithms for a network consisting of a central coordinator and distributed radar nodes, each operating in a monostatic mode. With an objective of enabling poor performing radar nodes, that observe low target signal-to-noise-interference ratio (SINR) values, benefit from distributed collaboration, we propose model-based max-min approach, in which we maximize the minimum of the SINRs observed by all nodes, under total bandwidth and individual node's range resolution (RR) constraints. This optimization is non-convex, but we solve it efficiently utilizing an explicit relationship between bandwidth and carrier frequencies, and the fact that each node's SINR is a monotonically decreasing function of bandwidth and carrier frequency allocated to the node. We propose two iterative optimization methods that employ successive convex approximation with a) semidefinite programming (SDP) and b) geometric programming (GP) problem formulations. Computer simulations show the performance of the proposed methods under different RR requirements, which significantly outperform the equal bandwidth allocation (EBWA) method and enable poor performing nodes to enhance their individual SINRs significantly. The solutions of this model-based optimization and target locations are then used, respectively, as labels and input, to train a bidirectional long short-term memory (LSTM) network. The trained network can significantly reduce the online run-time complexity of the bandwidth and carrier frequency allocation in distributed radar networks.

In shared spectrum with multiple radio access technologies, wireless standard classification is vital for applications such as dynamic spectrum access (DSA) and wideband spectrum monitoring. However, interfering signals and the presence of unknown classes of signals can diminish classification accuracy. To reduce interference, signals can be isolated in time, frequency, and space, but the isolation process adds distortion that reduces the accuracy of deep learning classifiers. We find that the distortion can be partially mitigated by augmenting the classifier training data with the signal isolation steps. To address unknown signals, we propose an open set hybrid classifier, which combines deep learning and expert feature classifiers to leverage the reliability and explainability of expert feature classifiers and the lower computational complexity of deep learning classifiers. The hybrid classifier reduces the computational complexity by 2 to 7 times on average compared to the expert feature classifiers, while achieving an accuracy of 95% at 15 dB SNR for known signal classes. The hybrid classifier manages to detect unknown classes at nearly 100% accuracy, due to the robustness of the expert feature classifiers.

The concept of metacognition has been proposed and applied to radar system analysis to enhance the classical cognitive radar paradigm. Metacognition allows a cognitive radar to have self-awareness about its cognitive processes. To accurately compare various cognitive processes and select the best under the operational scenario, a performability metric capable of comparing system performance over various scales and units is proposed and analyzed. This correspondence expands previous work on radar operational reliability to provide a metareliability metric for a metacognitive tracking radar. The approach is tested and validated on a radar that tracks a target performing a composite maneuver involving constant velocity, constant turn, and constant acceleration.

Multifunction arrays can allow multiple radar and communication beams, transmitting distinct information streams simultaneously and in different directions at the same operating frequency. Impedance tuning in the array elements allows each element's amplifier to optimize both power and linearity over changes in frequency and array scan angle. This enables clean, highly efficient multi-beam transmissions. Additionally, to ensure different messages are sent in assigned directions, directional modulation must be used. Directional modulation is briefly explained, accompanied with some initially successful simulation test results. Considerations for constructing a directional multifunction system are discussed, in terms of joining array impedance tuning and directional modulation.

In this paper distributed estimation of direction of arrival (DoA) is proposed for a network of radar nodes, in which nodes share their limited information related to their decisions with only their neighboring nodes. The sparsity of target scenario is exploited and distributed DoA estimation is formulated as the estimation of sparse vectors. The neighboring nodes aim to achieve a consensus on their estimation of these sparse vectors. The estimation problem is solved iteratively with alternating direction of method of multipliers (ADMM) method. Each node leverages co-prime array configuration, a type of structured sparse array, to enable direction finding for more sources than the number of node antennas. Numerical simulations show that the proposed distributed method converges within few iterations and provides much improved spatial spectra than local (nondistributed) estimations.

Recent work demonstrated closed-loop results for a real-time cognitive sense-and-notch radar capability via software-defined radio (SDR). The subsequent software-defined radar (SDRadar) generates spectrally shaped random FM (RFM) waveforms on-the-fly containing transmit spectral notches according to fast frequency assessments of other users in the band. Here we demonstrate the final cognitive evaluation step in which the sense-and-notch SDRadar operates in real-time in an open-air setting, performing moving target indication (MTI) processing (except for clutter cancellation) in the presence of a dynamically hopping interferer. This implementation is shown to support pulse repetition frequencies (PRFs) up to 4.4 kHz, meaning new interference-responsive waveforms can be produced at that rate, while achieving a transmit notch depth of 25 dB relative to peak power (greater depth is possible with additional computational resources).