Search

A sequential decision process in which an adaptive radar system repeatedly interacts with a finite-state target channel is studied. The radar is capable of passively sensing the spectrum at regular intervals, which provides side information for the waveform selection process. The radar transmitter uses the sequence of spectrum observations as well as feedback from a collocated receiver to select waveforms which accurately estimate target parameters. It is shown that the waveform selection problem can be effectively addressed using a linear contextual bandit formulation in a manner that is both computationally feasible and sample efficient. Stochastic and adversarial linear contextual bandit models are introduced, allowing the radar to achieve effective performance in broad classes of physical environments. Simulations in a radar-communication coexistence scenario, as well as in an adversarial radar-jammer scenario, demonstrate that the proposed formulation provides a substantial improvement in target detection performance when Thompson Sampling and EXP3 algorithms are used to drive the waveform selection process. Further, it is shown that the harmful impacts of pulse-agile behavior on coherently processed radar data can be mitigated by adopting a time-varying constraint on the radar's waveform catalog., Comment: 16 pages, 9 figures. arXiv admin note: text overlap with arXiv:2010.15698

We investigate three emerging topics essential for the development of cognitive radar (CR) for spectrum sharing: the response time (RT) of the CR, the autonomous regulation of the perception-action cycle (PAC), and the regulation of cognition. The RT measures the latency of all algorithms/hardware and is examined with respect to enabling capabilities of software-defined systems for rapid flexibility and responsiveness. The autonomous regulation of the PAC determines “how fast the CR can interact with the environment” as well as “how fast the CR should interact with the environment.” The regulation of the PAC is explored with respect to pulse-to-pulse waveform agility to coexist successfully with dynamic radio frequency emitters in the ambient electromagnetic environment (EME) and the consequence of modifying the waveform within the coherent processing interval. Finally, the regulation of cognition determines how to select a particular CR technique appropriately for a given dynamic environment. This selection requires a high level, or metadecision process to identify the appropriate CR technique as the EME changes over time and we therefore concentrate discussion on the newly emerging topic for radar called metacognitive radar. The exploration of these three topics include a review of past and current research with discussion of possible future research.

The growing demand for radio frequency (RF) spectrum access poses new challenges for next-generation radar systems. To operate in a crowded electromagnetic environment, radars must coexist with other RF emitters while maintaining system performance. This work evaluates the performance of a spectrum sharing cognitive radar system, which predicts and avoids RF interference (RFI) in real time. The system applies a cognitive perception-action cycle that senses RFI, learns RFI behavior over time, and adapts the radar's frequency band of operation. Through cognition, the system learns a stochastic model describing RF activity. This model allows the cognitive radar to predict RF activity in real time and share the spectrum with emitters, such as communication systems. A set of synthetic and measured interference signals are used to evaluate the performance of this predictive spectrum sharing scheme. This work assesses the impact of RFI on the cognitive radar's range profile with respect to variation in RF environment. The system demonstrates accurate avoidance of deterministic RFI with a degradation in spectrum sharing efficiency as variability over time increases.

In this paper, dynamic non-cooperative coexistence between a cognitive pulsed radar and a nearby communications system is addressed by applying nonlinear value function approximation via deep reinforcement learning (Deep RL) to develop a policy for optimal radar performance. The radar learns to vary the bandwidth and center frequency of its linear frequency modulated (LFM) waveforms to mitigate mutual interference with other systems and improve target detection performance while also maintaining sufficient utilization of the available frequency bands required for a fine range resolution. We demonstrate that our approach, based on the Deep Q-Learning (DQL) algorithm, enhances important radar metrics, including SINR and bandwidth utilization, more effectively than policy iteration or sense-and-avoid (SAA) approaches in a variety of realistic coexistence environments. We also extend the DQL-based approach to incorporate Double Q-learning and a recurrent neural network to form a Double Deep Recurrent Q-Network (DDRQN). We demonstrate the DDRQN results in favorable performance and stability compared to DQL and policy iteration. Finally, we demonstrate the practicality of our proposed approach through a discussion of experiments performed on a software defined radar (SDRadar) prototype system. Our experimental results indicate that the proposed Deep RL approach significantly improves radar detection performance in congested spectral environments when compared to policy iteration and SAA., To Appear in IEEE Transactions on Cognitive Communications and Networking, 2020

In this article, 1 1 Portions of this work were presented at the 2018 IEEE Radar Conference [1] . we examine a cognitive radar that must coexist with communication systems. Specifically, we model a cognitive radar's adaptation to the communication system as a Markov decision process (MDP) and then apply reinforcement learning to solve the resulting optimization problem. More specifically, we learn the environment model and then apply policy iteration to determine the optimal policy. The radar environment consists of a single moving target and a communication system that uses the same bands as the radar. The communication system is modeled using several different transmission behaviors: constant, intermittent, triangular frequency sweep, sawtooth frequency sweep, frequency hop, pseudorandom frequency hop, and direction-dependent interference. We demonstrate how the MDP framework and reinforcement learning can be used to help the radar determine actions (i.e., transmission strategies) to maximize its own performance.