Figure 3 - uploaded by Alexander Charlish
Content may be subject to copyright.
Source publication
The multifunction radar, aided by advances in electronically steered phased array technology, is capable
of supporting numerous, differing and potentially conflicting tasks. However, the full potential of the
radar system is only realised through its ability to automatically manage and configure the finite resource
it has available. This thesis det...
Contexts in source publication
Context 1
... search bars so that the beam centres form a triangle and the potential nulls in the surveillance region are reduced. To reduce the severity of the potential null, the beam pattern can be interlaced [Billam, 1997], so that the beam centres on the next scan are directed at the previous nulls. Using this method the next beam positions are shown in Fig. 3.1 by the dashed line. The triangular search pattern is parameterised by the beam spacing D p which separates each beam by angle θ S , which is taken in terms of the ...
Context 2
... analysed the trades associated with adaptive update rate selection, with similar conclusions. A standard approach [van Keuk and Blackman, 1993] is to select a revisit interval based on the earliest time after the filter angular prediction error, along the major axis of the uncertainty ellipse G, exceeds a fraction of the beamwidth as shown in Fig. 3.2. The fraction is called the track sharpness and denoted v 0 . So, the next revisit time t K+1 is chosen according ...
Context 3
... is demonstrated in Fig. 3.4 where each circles represents a possible state, each dashed line repre- sents a possible action and each solid line represents the action taken. The reward for the whole trajectory is the sum of the individual rewards marked in the ...
Context 4
... aspects which affect performance but are under control such as the task revisit rate. The resource dimensions are the finite resource to be distributed between tasks which is radar time or loading. A utility function is defined which quantifies the satisfaction associated with each point in the quality space. This utility model is demonstrated in Fig. 3.5 where it can be seen that each operational parameter uses a different resource loading and produces a different task utility. Parameters along the "concave majorant" [Hansen et al., 2006], where utility per resource is maximised, are ...
Context 5
... optimum or brick packing methods attempt to form the time line by creating a series of allocation frames of fixed duration. Whilst the previous allocation frame is being executed, the next frame is being calculated. This is represented in Fig. 3.6 for the set of scheduable tasks T A . Given a measure of optimality, an exhaustive search can provide the optimum solution over the time horizon, however, heuristics [Winter and Lupinski, 2006;Winter and Baptiste, 2007] are used to guide the 'packing' of the tasks into the frame. A result of this method is that pop-up tasks which ...
Context 6
... be seen that the minimum track loading is found at 0.21 regardless of the varying manoeuvres and ranges which makes it a useful measure for target tracking control. If track updates are requested at intervals which are determined by a limit of the accuracy in carte- sian coordinates then there is no common minimum loading track sharpness setting. Fig. 4.13(b) shows the loading for bounds on the trace and determinant for varying ranges and manoeuvres. Also, the de- terminant or trace in spherical coordinates removes the dependence on range, but is still dependent on range ...
Context 7
... can exist, that can be distributed over several hosts each running one Java application. Agents exist within a container, with each agent ran in an independent thread. In compliance with FIPA, the agent platform is modelled as containing the agents, an agent management system, a directory facilitator and a message transport system as shown in Fig. 5.3. Only a single agent management system exists on a platform, which handles tasks such as agent creation and individual naming. The directory facilitator provides a yellow page service for agents to advertise services and the message transport system handles the passing of messages between agents in the platform, which could be across ...
Context 8
... architecture of the complete software system is shown in Fig. 5.4. The architecture was designed to allow integration into the radar resource manager architecture shown in Fig. 3.3. The inheritance struc- ture of the agents allowed for different auction mechanisms to be implemented whilst maximising code reuse. It was also important to design each task agent without the knowledge of the task it represents, again to maximise code ...
Context 9
... is first useful to visualise how the valuations submitted to the auction by a task agent representing a track, which determines which task is executed and hence the revisit interval, differ over time for the three variants considered. This is shown in Fig. 6.3(a), Fig. 6.3(b) and Fig. 6.3(c) for RB-EDF, GIF and LQF respectively. In this simulation a single target is tracked using a continuous white noise jerk model as the limiting form of the Singer model with a process noise intensity˜qintensity˜ intensity˜q = 3.3 unless otherwise stated. The target is on the radar boresight at 50km, a unity ...
Context 10
... the bid value submitted to the auction by a track agent in RB-EDF is equal to the time delay based on the execution time specified by the rules. For GIF, which is shown in Fig. 6.3(b), the bid value, which is the mutual information production of the potential measurement, does not increase linearly and is affected by the received SNR which suggests that when using this measure preference is given to 'bright' targets which produce greater information. Fig. 6.3(c) shows LQF where it can be seen that the utility ...
Context 11
... execution time specified by the rules. For GIF, which is shown in Fig. 6.3(b), the bid value, which is the mutual information production of the potential measurement, does not increase linearly and is affected by the received SNR which suggests that when using this measure preference is given to 'bright' targets which produce greater information. Fig. 6.3(c) shows LQF where it can be seen that the utility production of the next measurement increase is also non-linear and is affected by the target process noise intensity˜qintensity˜ intensity˜q in the tracking filter model. This suggests that LQF gives a resource preference to targets performing extreme manoeuvres as they degrade in task ...
Context 12
... scheduling based on information regarding mutual information and task quality through utility. However, it has been found that the extent to which this improves or degrades performance can not be analysed from a single track perspective as it is not clear how the competition for resource in SFPARM will manifest itself into the valuations shown in Fig. 6.3. Therefore to sufficiently analyse and compare the allocation mechanism it is necessary to fully model the competition for resource by simulating numerous competing tracking ...
Context 13
... The distributed, decentralised nature of the mecha- nism provides scalability and allows each task agent to be designed independently. The CDAPS mecha- nism is implemented in the radar resource management testbed as shown in Fig. 5.4. This architecture allows easy integration into a typical radar resource management architecture, shown in Fig. 3.3, by re- placing the task request modules. The CDAPS algorithm selects task parameters from a usable waveform database given the model of the current scenario which includes priority assignment, the usable wave- form database and the current state of the radar task function. The global feedback enables the update of the model of the ...
Citations
... Driven by advanced electronic technologies, such as digital phased array systems, radar is gradually evolving toward multifunctionality. For example, it has the ability to quickly change the waveform between pulses and can adaptively adjust the working state according to the task and external environment [3][4][5][6][7]. ...
... This paper established a radar state transition diagram using six operation modes of a MFR as states, as shown in Figure 12. The states 1 6 S S represent the STT, MTT, TAS, TWS, RWS, and VS modes, with their threat levels defined as 10,8,6,5,3, and 1, respectively. The VS mode had the lowest threat level and was set as the radar's target state. ...
With the advancement of radar technology toward multifunctionality and cognitive capabilities, traditional radar countermeasures are no longer sufficient to meet the demands of countering the advanced multifunctional radar (MFR) systems. Rapid and accurate generation of the optimal jamming strategy is one of the key technologies for efficiently completing radar countermeasures. To enhance the efficiency and accuracy of jamming policy generation, an efficient jamming policy generation method based on multi-timescale ensemble Q-learning (MTEQL) is proposed in this paper. First, the task of generating jamming strategies is framed as a Markov decision process (MDP) by constructing a countermeasure scenario between the jammer and radar, while analyzing the principle radar operation mode transitions. Then, multiple structure-dependent Markov environments are created based on the real-world adversarial interactions between jammers and radars. Q-learning algorithms are executed concurrently in these environments, and their results are merged through an adaptive weighting mechanism that utilizes the Jensen–Shannon divergence (JSD). Ultimately, a low-complexity and near-optimal jamming policy is derived. Simulation results indicate that the proposed method has superior jamming policy generation performance compared with the Q-learning algorithm, in terms of the short jamming decision-making time and low average strategy error rate.
... In this respect, the RRM can use different optimization tools to perform resource allocation. Among them, it is worth mentioning the quality of service resource allocation method (Q-RAM) [4] and the continuous double auction parameter selection (CDAPS) [5], [6]. ...
Multifunction phased array radars (MPARs) exploit the intrinsic flexibility of their active electronically steered array (ESA) to perform, at the same time, a multitude of operations, such as search, tracking, fire control, classification, and communications. This paper aims at addressing the MPAR resource allocation so as to satisfy the quality of service (QoS) demanded by both line of sight (LOS) and reflective intelligent surfaces (RIS)-aided non line of sight (NLOS) search operations along with communications tasks. To this end, the ranges at which the cumulative detection probability and the channel capacity per bandwidth reach a desired value are introduced as task quality metrics for the search and communication functions, respectively. Then, to quantify the satisfaction level of each task, for each of them a bespoke utility function is defined to map the associated quality metric into the corresponding perceived utility. Hence, assigning different priority weights to each task, the resource allocation problem, in terms of radar power aperture (PAP) specification, is formulated as a constrained optimization problem whose solution optimizes the global radar QoS. Several simulations are conducted in scenarios of practical interest to prove the effectiveness of the approach.
... In detail, that means selecting random pairs i, j ∈ N of task indexes until the given number of substitutabilities is achieved and setting I ij = −1. Additionally, we need to define the substitution utility for task T i achieved by executing task T j defined by u i q ij (r j , e) required in (18). Since the tasks are modeled abstract, we assume that u i q ij (r j , e) is calculated by scaling the utility of task T j linearly using a randomly selected factor I s ij while ensuring a maximum utility of 1, i.e. u i q ij (r j , e) := min I s ij u j q j (r j , e) , 1 . ...
... In contrast to the previous section, in this section we calculate the terms u i q i (r i , e) required in (19) as well as u i q ij (r j , e) required in (18) using the tracking performance model described in the following. The tracking performance model is very similar to the one presented in [34]. ...
This paper presents an exact quality of service radar resource management model which considers pairwise dependencies between tasks and is based on mixed-integer programming. This is in contrast to the common approach, which is not able to model any dependencies and hence has to assume independence among the tasks, which is generally an incorrect simplification. We define and implement substitutabilities between radar tasks with regard to their utility. Numerical results demonstrate that in scenarios with substituabilities, the proposed method can achieve a significantly better operational radar performance than a traditional approach, especially in high load situations. Though, a drawback of the added complexity due to considering dependencies is an increased computational runtime in comparison to the common method. Due to implementing dependencies, the method is especially well suited for complex scenarios with many substituabilities among radar tasks that need to be exploited under high task load.
... 3) Adaptive revisit time based on NEES: the accuracy of the covariance is measured through the normalized estimation error squared (NEES) that, for a single sample in the case of constant velocity estimate, should be chi-square distributed with six degrees of freedom [36], [38]. Hence, the average value for the NEES should be six, where the confidence interval for the chi-square distribution is [ξ 2 L , ξ 2 H ]. For instance, the interval [0.87, 16.81] is derived setting the confidence to 98%. As a consequence, the revisit time can be set as the maximum revisit such that the estimated NEES is within a specific interval [ξ 2 L , ξ 2 H ]. It is finally, worth noticing that for all the above mentioned strategies the tuning parameters (viz. ...
The modern battlefield scenario is strongly influenced by the innovative capabilities of the multifunction phased array radars (MPARs) which can perform sequentially or in parallel a plethora of sensing and communication activities. As a matter of fact, the MPAR can functionally cluster its phased array into bespoke sub-apertures implementing different tasks. Accordingly, a portion of the other available resources, e.g., bandwidth, power-aperture product (PAP), and time, is also assigned to each sub-aperture and the grand challenge is the definition of strategies for an optimal scheduling of the tasks to be executed. In this respect, a rule-based algorithm for task scheduling is proposed in this paper. In a nutshell, in each time window, the procedure first allocates the radar tasks (viz. volume search, cued search, update and confirmation tracking) and then utilize the communication (COM) looks so as to fill the empty intra-slot time left by the radar tasks. When there are two concurrent looks, the allocation is performed according to their priorities. Moreover, if the bandwidth and PAP are sufficient, some of them can be also scheduled in parallel. Interesting results in term of bandwidth and time occupancy efficiency are observed from simulations conducted in challenging scenarios comprising also multiple maneuvering targets.
... In this respect, the RRM can use different optimization tools to perform resource allocation. Among them, it is worth mentioning the quality of service resource allocation method (Q-RAM) [4] and the continuous double auction parameter selection (CDAPS) [5], [6]. ...
Multifunction phased array radars (MPARs) exploit the intrinsic flexibility of their active electronically steered array (ESA) to perform, at the same time, a multitude of operations, such as search, tracking, fire control, classification, and communications. This paper aims at addressing the MPAR resource allocation so as to satisfy the quality of service (QoS) demanded by both line of sight (LOS) and non line of sight (NLOS) search operations along with communications tasks. To this end, the ranges at which the cumulative detection probability and the channel capacity per bandwidth reach a desired value are introduced as task quality metrics for the search and communication functions, respectively. Then, to quantify the satisfaction level of each task, for each of them a bespoke utility function is defined to map the associated quality metric into the corresponding perceived utility. Hence, assigning different priority weights to each task, the resource allocation problem, in terms of radar power aperture (PAP) specification, is formulated as a constrained optimization problem whose solution optimizes the global radar QoS. Several simulations are conducted in scenarios of practical interest to prove the effectiveness of the approach.
... It is well known that it is impossible for the jammer to fully understand all the parameters of the radar during the countermeasure. The jamming style and jamming strategy adopted by the jammer are only dependent on the results of its reconnaissance of the target radar [79], [80], [81], [82], [83]. ...
With the increasingly complex electromagnetic environment and the intelligent development of radar, the jammer, as opposed to radar, urgently needs to improve its ability to recognize threat targets and make jamming decisions. In this paper, we first introduce the concepts and systems of cognitive electronic warfare (CEW) and summarize its research status. Through analysis of the existing CEW systems, we propose a CEW model suitable for the cluster confrontation scenarios. Then, for the radar jamming decision-making (RJDM) namely a crucial part of CEW, we discuss the advantages, disadvantages, and applications of the traditional methods and analyze the machine-learning based methods including Markov decision processing, the newest Q-Learning, Deep Q-Learning (DQN), Double Deep Q-Learning (DDQN), A3C algorithms, and their improved algorithms etc. We build radar adversarial models and verify the effectiveness of reinforcement learning (RL) algorithm and the superiority of deep RL by simulating both the underlying Q-Learning and DQN algorithms. Finally, the research trends of CEW are discussed.
... In this section we will briefly describe the approximative solution strategy to the Q-RAM problem proposed in [1], [2]. A notable alternative approach is given in [6]. First, all possible task configurations are generated and evaluated on a per task basis, i.e. their resulting utility and resource requirement are computed. ...
An intelligent radar resource management is an essential building block of any modern radar system. The quality of service based resource allocation model (Q-RAM) provides a framework for profound and quantifiable decision-making but lacks a representation of inter-task dependencies that can e.g. arise for tracking and synchronisation tasks. As a consequence, synchronisation is usually performed in fixed non-optimal patterns. We present an extension of Q-RAM which enables the resource allocation to consider complex inter-task dependencies and can produce adaptive and intelligent synchronisation schemes. The provided experimental results demonstrate a significant improvement over traditional strategies.
... For now on, the utility function we use is decreasing with distance to guarantee a good signal-to-noise ratio (SNR). The choices of the utility function and of the policy of selection can also add priority mechanisms, fuzzy logic or other advanced concepts such as those mentioned in [7]. We use mechanical steering in our work; the utility and policy can be adapted to harness the assets of electronic beamsteering, such as the ability to dynamically switch between different areas. ...
Building a labeled database usually requires extensive work when the labeling is not automated. This paper presents a methodology to build a labeled database of aircraft radar signature (specifically radar cross section) using the OpenSky Network data. This method relies on a radar system steered by the live API to automatically select, track and measure target aircraft according to custom user rules. The OpenSky ADS-B database allows for an efficient processing of the radar data, as it provides ranging and speed information to locate the region of interest inside the radar data. In addition, the ADS-B and Mode-S databases give an estimate of the attitude of the aircraft, which is paramount for signature measurement. We also compare the ADS-B data with the radar data.
... Towards this end, the continuous double auction parameter selection algorithm is proposed in refs. [7,120] which enables the solution from the previous time step to be adapted to the current time step without a complete re-computation of the resource allocation, hence reducing computation for dynamic RRM problem. ...
A multi‐function radar is designed to perform disparate functions, such as surveillance, tracking, fire control, amongst others, within a limited resource (time, frequency, and energy) budget. A radar resource management (RRM) module within a radar system makes decisions on prioritisation, parameter selection, and scheduling of associated tasks. However, optimal RRM algorithms are generally computationally complex and operational radars resort to heuristics. On the other hand, algorithms based on artificial intelligence (AI) have been shown to yield near‐optimal radar resource allocation results at manageable computational complexity. This survey study aims at enabling researchers and practitioners better understand the application of AI in RRM‐related problems by providing a thorough literature review of AI‐based RRM techniques. We first provide background concepts in RRM followed by a brief review of Symbolic‐AI techniques for RRM. We mainly focus on the applications of state‐of‐the‐art machine learning techniques to RRM. We emphasise on the recent findings and their potential within practical RRM scenarios for real‐time resource allocation optimisation. The study concludes with a discussion of open research problems and future research directions in the light of the presented survey.
... In this section we will briefly describe the approximative solution strategy to the Q-RAM problem proposed in [1], [2]. A notable alternative approach is given in [4]. First, all possible task configurations are generated and evaluated on a per task basis, i.e. their resulting utility and resource requirement are computed. ...
An intelligent radar resource management is an essential building block of any modern radar system. The quality of service based resource allocation model (Q-RAM) provides a framework for profound and quantifiable decision making but lacks the flexibility necessary for optimal mitigation strategies in the presence of interference. We define an extension of the Q-RAM based radar resource management framework with an intelligent interference handling capability using various mitigation methods. The approach incorporates virtual time resources and alternative task configurations to compute near-optimal solutions in the presence of interference. The provided experimental results demonstrate a significant improvement over traditional strategies.
