Target Track Maintenance and Maneuver Handling with Optimal Revisit Control for Phased Array Radar

Transcription

1 9th International Radar Smposium India - (IRSI - ) Target Trac Maintenance and Maneuver Handling with Optimal Revisit Control for Phased Arra Radar Aparna Rathi, D S Seshagiri, Vive Krishna and Prasanthi E R Electronics and Radar Development Establishment C. V. Raman Nagar Bangalore 6 9 aparna_rathi@rediffmail.com Abstract: The paper presents the challenges and techniques in handling maneuvering target tracing with phased arra radar. The focus is to achieve of optimal revisit control for targets without breaing loc and maintain trac on highl maneuvering targets. The constraint on tracing sstem comes from extended search requirements which results in more time on target and hence increased search scan time. This poses challenges to the target tracing sstem with requirements to handle high g maneuvers and maintain trac on targets which can be as high as. The paper presents approach adopted for maintaining maneuvering tracs with optimal use of radar resources and emphasis to avoid trac loss. I INTRODUCTION The paper is organized as follows. Section A discusses the challenges involved in tracing with narrow azimuth beam and factors affecting choice of revisit to maintain trac on target without breaing loc. Section B discusses the choice of suitable dnamic models for the targets. Section C describes a combination of design approaches adopted for suitable choice of revisit for target tracs based on its dnamics. Section D discusses the results achieved and section E summarizes the performance of the methods adopted. II PARAMETRS AFFECTING TARGET REVISIT The goal of an surveillance sstem is to quicl confirm the presence of the targets in its surveillance volume and maintain trac on them as long as the are in the surveillance coverage. For the phased arra radar sstem which has agile beam, the presence of target can be confirmed quicl b verification process. However, the e here is to manage the radar resources efficientl. The idea is to maintain tracs on targets which are alread detected while quicl confirming presence of new targets entering the radar coverage. The tracing sstem also need to consider that a militar target can exhibit sudden maneuver as high as and the design needs to be robust enough to handle such maneuvers. The radar considered has narrow azimuth beam for accurate tracing and broad beam in elevation to obtain the desired coverage. The narrow beam in azimuth necessitates the proper choice of revisit for the targets, as the target can move awa from angular beam quicl especiall at ranges near to the radar. Relationship between update time and resulting prediction error expressed in terms of variance reduction ratio and other sstem parameters[][].... () and are target maneuver time constant and maneuver standard deviation respectivel. For suitable choice of maneuver time constant appling the above equation to meet prediction error to be within db of radar transmit beam the update time requirement for different g maneuvers at bore sight is shown in figure(a) below for radar beam of.8 degree at bore-sight. in m 6 Sampling time Figure a: Sampling requirements at boresight for azimuth beam of.8 degree If the beam width is reduced further b one third the corresponding sampling required for the target increases as is evident in figure b. in Km.8 deg beam - boresight At boresight.... Sampling Time in Seconds Figure b: Sampling requirements at bore-sight for reduced azimuth beam 7 7 NIMHANS Convention Centre, Bangalore INDIA - December

2 9th International Radar Smposium India - (IRSI - ) For maintaining target tracs, understanding dnamics during maneuver is e for design of tracing sstem. The computation of rate of turn gives indication of angular movement of target and time required to execute different g maneuvers. The figure() below shows rate of turn (w) in deg/s for different g turns Rate of turn w in deg/s Figure : Rate of turn as function of speed The rate of turn (w) depends on the speed and g with which turn is taen. This in turn decides the angular movement of target with respect to radar. Assuming that a target can tae maneuver at an instant of time there is a minimum sampling rate required to maintain the trac on target so that the target does not move out of the beam. Such a constraint on revisit time of target based on target geometr and range helps in trac maintenance. III DESIGN OF TRACKING FILTERS For design of tracing filters the criteria was to arrive at a combination which gives maximum sampling interval and also handle maneuver efficientl. IMM tracs target maneuvers more efficientl[]. The design of IMM tracing filters involves selection of appropriate aircraft models. In the current design we have incorporated a three-model-imm algorithm for tracing maneuvering targets. Three different models are chosen to represent target dnamics: Constant velocit(cv) model to represent straight line level flight, constant acceleration(ca) model to represent changes in target velocit and co-ordinated turn(ct) model to handle g maneuvers. This combination of dnamic models gave an optimal revisit suitable for application. To isolate the errors associated with broader elevation beam a decoupled tracing scheme is adopted with range and azimuth information utilized in IMM filter and elevation incorporated in separate filter. Target Kinematic Models The discrete-time model for a dnamic sstem is given b two relationships the first one describes the dnamics of the sstem X = FX + w + () and, the second one describes the relationship of the state with measurement where, Z = H X + v X is the state vector 7g g g g 6 Speed in m/s () Z is the measurement vector w ~ N(, Q ) is the process noise, with zero-mean and covariance of Q v ~ N(, R ) is the measurement noise, with zeromean and covariance of R The state estimate from the three models at time instant, are denoted b x x& & & x& & & X = x& & X x& = & && x && X = x& & && x && The constant velocit model considers nearl constant velocit for non maneuvering targets. The state transition matrix and process covariance matrix are given b, Similarl for the acceleration model The fixed center constant turn rate maneuver model taes the form of a second-order Marov process, given as a = ω v + w It leads to following discrete time model for each Cartesian () () NIMHANS Convention Centre, Bangalore INDIA - December

3 9th International Radar Smposium India - (IRSI - ) coordinate with state x = [position, velocit, acceleration] [] sinωt ωt sinωt ω ω ω sinωt x T x + = cosω + w ω ω ω sinωt sinωt ω (6) where, the cov(w ) is σ w The turn rate, w is estimated b using the acceleration estimates provided b the model. The turn rate is then approximated as the ratio of latest acceleration and velocit estimates of the target(under the assumption that the velocit and acceleration vectors are orthogonal), i.e a ω = v (7) The w is estimated b using the velocit and acceleration estimates provided b the mixed state estimate of CT model. IV REVISIT CONTROL The revisit control design has to be robust so as to prevent trac loss during target maneuver. Adaptive sampling capabilit due to electronic beam steering provides better tracing performance [] as compared to fixed sampling. Moreover the adaptive sampling optimizes the use of radar resources. To trac maneuvering targets without breaing loc on targets a combination of techniques have been utilized. The techniques are further described below: a) To optimize radar resources an adaptive revisit selection is incorporated based on the predicted values of radar angle innovation standard deviation relative to radar beam width[6].to maintain trac angular deviation of trac is monitored for closed loop target and the error build up is restricted to be within db angular beam width. From the IMM models combine the predicted covariance of each model as / / / / / / (8) / is then converted in to polar coordinates as follows / = C / C (9) with, where, x, are the target coordinates and r is the range of the target. The diagonal elements contain the innovation variances in range and azimuth which can be compared with db radar angular beam. Variance in azimuth(σ a ) can be written as σ a < * BW () BW - Beam width For current application = / b) As tracing filter has inherent lag in model switching the smoothed innovation of CV model is also monitored to eep chec on innovation build up. From the CV model of IMM the trac innovation build up is monitored, the motivation behind that is when a target starts maneuvering the innovation build up for target is maredl visible as is shown at start of maneuver in figure(b) below. (degree) TIME Vs x Figure a) Target with time Trac Innovation(degree) x Figure b) innovation If the innovation reaches the threshold of db radar beam a beam is put within ms so as to capture the target maneuver. c) To tae into consideration the target acceleration and its range from the radar as parameters influencing the target revisit equation() is applied. The appropriate value of maneuver time constant is derived from the acceleration statistics of target. The trac range and acceleration values are derived from IMM filter. selection of the target maneuver time constant is varied based on the acceleration information derived from IMM. Target Acceleration m/s Maneuver time constant s <= < -8 <= < B appropriate manipulation of equation() and putting the constraint of angular variance to be within db of radar beam the equation can be written as T =.*( *Rng/Acc)^. () d) Based on target geometr and range the sampling set is varied. This is to cater for narrow azimuth beam. At ranges near to radar there is requirement of obtaining the target sample before it moves out of radar beam. Hence the constraint on revisit Sampling Sets tried out are: For ranges beond m NIMHANS Convention Centre, Bangalore INDIA - December

4 9th International Radar Smposium India - (IRSI - ) { } For ranges up to m { } e) For robustness in design recover pattern is incorporated in case of target loss. Recover Ccle is initiated to minimize trac loss during maneuver due to absence of detections. In case of detection loss trac beam is scheduled within ms. For two successive detection losses a recover is attempted in angular direction with maximum up to recover ccles for ever target. The tracing ccle proceeds as follows: )Based on target range appropriate sampling set is selected from (d). ) Innovation build up from constant velocit filter is checed, if it is above threshold a quic update < ms is scheduled. ) If innovation build up is within threshold, highest revisit is tried (criteria a). Predicted angular deviation is checed to be within db radar angular beam, if not the next lower interval is tried till the criteria is met. In Parallel the required revisit based on target range and acceleration and geometr is also computed. If this is found to be lower than that given b a, update is scheduled with this. )In case of detection loss recover ccle is initiated A limited search along target bearing is carried out to reacquire the target. in figure(7a) and figure(7b) respectivel. Acceleration Target Figure 6: Target maneuvers g-g (a):b-scope (b):range vs time timestamp Figure: 7a) Acceleration statistics b) Revisit Time Figure(8a) below depicts an helicopter executing multiple maneuvers at ranges less than 6m from radar. Figure(8b) shows corresponding radar trac output in B-scope 7 6 Acceleration Vs timestamp -Sep- x 7 (Km) TimeIntrvl x TgtRng Vs CUBeamPlaTime TIME Vs x 7 -Feb- TimeIntrvl Vs -Sep- V PERFORMANCE EVALUATION The revisit control techniques discussed above were put along with three model IMM estimator and separate height filter and maneuver performance was evaluated with actual targets maneuvering up to 6g. The performance evaluation was also carried out for target maneuvers greater than 6g in simulation mode. Figure() simulates targets maneuver from g to and corresponding acceleration estimates b IMM are depicted in figure(a) and revisit time in figure(b) PPI TgtRng CUBeamPlaTime x 7 Figure 8a: vs time for helicopter Helicopter TIME Vs 8 Acceleration x (a) (b) Figure a: Simulation of target maneuvers from g to a:ppi plot b: range vs time Acceleration Vs timestamp -Sep timestamp x 7 TimeIntrvl(sec) Figure: a) Acceleration statistics b) Revisit Time Figure (6) depicts aircraft executing maneuvers between g and g with corresponding acceleration and revisit depicted... (Km) 9 8 TimeIntrvl Vs -Sep x Figure 8b: B-Scope (Trac maintained b radar) Figure(9) depicts fighter executing high g horizontal maneuver followed b vertical maneuver. The corresponding revisits scheduled b tracer is plotted in Figure(9b). NIMHANS Convention Centre, Bangalore INDIA - December

5 9th International Radar Smposium India - (IRSI - ) Target S. C. Height Figure 9a: Aircraft executing horizontal & Vertical maneuver. Time Vs Update Interval Mr.D.Seshagiri obtained his B.E from Andhra Universit College of Engineering in ECE, M.Tech in Integrated Electronics from IIT Madras. He joined LRDE in Januar as Scientist D. Prior to joining LRDE, he wored in BHPV, Visahapatnam and Wilco International, Hderabad. Currentl he is Scientist 'F' and Project Director of Primar Radar for AEW&C Programme. His areas of interest include Radar Sstem Design, Radar Signal & Data processing and Mechatronics. He is recepient of Scientist of the ear award. Vive Krishna was born in 979 at Ballia, Uttar Pradesh. He completed M. Tech in Electronics Engineering in J.K Institute of Technolog, Allahabad. He wored as a Lecturer at BBS Engineering College, Allahabad and ISDC, Allahabad. He joined Electronics and Radar Development Establishment, Bangalore in 8 as a scientist and has been woring on radar data processing applications. Interval (sec) x 7 Prasanthi E R was born in 98 at Thrissur, Kerala. She graduated in Computer Science and Engineering from Govt. Engineering College, Thrissur in. She is woring in the area of radar data processing in Electronics and Radar Development Establishment since April. Figure 9b: Revisit time VI CONCLUSION Through this paper we have tried to address the issues encountered during tracing of target maneuvers. The emphasis was to avoid trac loss and mae design robust to handle widel different target dnamics. Choice of IMM algorithm consisting of co-ordinated turn model in conjunction with the revisit control techniques discussed has led to successful handling of target maneuvers. REFERENCS [] S. S. Blacman, Multiple Target Tracing with Radar Applications, Norwood, MA: Artech, 986. [] Samuel Blacman, Robert Popoli, Design and Analsis of Modern Tracing Sstems Artec House []Comparative stud of turn rate estimation techniques in IMM estimators for ESA radar tracing, S. Veeraraghavan, Aparna Rathi and M. Justin Sagaaraj Aerospace Conference, 8 IEEE [] X. Rong Li, Vesselin P. Jilov, A surve of maneuvering target tracing-part I: Dnamic models, Proceedings of SPIE Conference on Signal and Data Processing of Small Targets, Orlando, FL, USA, April. []Applications of a phased arra antenna in a multiple maneuvering target environment S. S Blacman, T.J. Broida and M F. Cartier Decision and Control including the Smposium on Adaptive Processes, 98 oth IEEE Conference(Volume:) [6]IMMPDAF for Radar Management and Tracing Benchmar with ECM T Kirubarajan, Y Bar Shalom, W D Blair and G A Watson IEEE Transactions on aerospace and electronic sstems Volume, No october 998 BIO DATA OF AUTHORS Aparna Rathi, received the M.Tech degree in Electronics Design Technolog from Center of Electronics Design Technolog of India, Aurangabad, in 997. She is a scientist in Electronics and Radar Development Establishment, Bangalore since 999. She has wored for development of target tracing and radar data simulation software for different radar applications. NIMHANS Convention Centre, Bangalore INDIA - December