VLSI based fuzzy logic controller enabled adaptive interactive multiple model for target tracking
Introduction
Track while scan (TWS) radar and tracking filter operations are described in [1] and [2], respectively. The Interacting Multiple Model (IMM) algorithm [2], [3] based on Kalman filtering (KF) technique has proved useful in tracking maneuvering targets. A simple IMM algorithm consists of velocity model KF and acceleration model KF to accommodate various maneuvering trajectories. Both these models interact with each other and adjust their parameters in a probabilistic manner to provide optimum results. The velocity filter model is ideally suited for predominantly non-maneuvering targets while the acceleration model proves more accurate for maneuvering target trajectory. The model weights and the transition probabilities determine the interaction between the models.
In IMM the transition probabilities is maintained as a constant. Due to this, one of the filters will be always inaccurate in tracking and will contribute to the weighted filter output. This leads to performance degradation of the IMM. Changing the transition probabilities during every scan can reduce this degradation in performance [4]. Such an IMM is referred to as adaptive IMM (AIMM). A fuzzy logic controller (FLC) has been designed for this purpose of providing adaptability [5]. The FLC has error in estimation and change in error as its inputs, and provides suitable transition probabilities (according to which the system model switches between velocity model filter and acceleration model filter) as outputs. Current work is the implementation of this FLC. Since IMM algorithm is used in a real-time tracking application, the VLSI realization of FLC-AIMM is required to meet the critical time specification. At present, there are available hardware chips for the Kalman's algorithm [1] implementation of the velocity and acceleration state model filters. This paper presents the first and preliminary step towards integrating the entire AIMM onto ASIC/FPGAs.
This paper primarily describes the design and implementation of such specific purpose FLC architectures, suitable towards implementation in VLSI. The inputs to the FLC are normalized values of error and change in error calculated from IMM. The outputs from the FLC are the transitional probabilities, required for the IMM to effectively select the filter model that is better suited to the nature of the target. Use of FLC in IMM reduces inaccuracy in spatial estimation, and thereby improves the tracking accuracy of IMM [5].
Section snippets
Principle of fuzzy-based AIMM
Usage of fuzzy logic in tracking maneuvering targets is the topic of the latest research [3], [6], [7]. In the current work, using fuzzy logic an AIMM is developed and logic synthesis of the developed FLC is carried out. The FLC was designed to achieve the timing constraints of TWS radar.
Fig. 1 presents the fuzzy-based AIMM. The range error E(k) and the change in the range error ΔE(k) are the input parameters to fuzzy controller and the transition probabilities p11 and p22 are the output of the
VLSI implementation of the FLC
The need to implement the FLC in VLSI is owing to the computational time constraints in tracking. The transitional probabilities have to be calculated by FLC and made available to the IMM prior to arrival of next measurement from radar as shown in Fig. 1. Moreover, within the radar scan period the tasks, namely signal acquisition and data processing, have to be completed. This gives critical timing constraint in the realization of the FLC. Hence a VLSI implementation of FLC is presented in this
Result and discussion
The hardware description and logic simulation of the FLC have been carried out in VHDL. In order to test this hardware description for accuracy, the simulation is carried out with values provided at input and examining the output generated, which is shown in Fig. 6. Here input error and change in error are maintained as 0.012. This is interpreted as “0000 0011” or “03”H by the FLC module. These inputs fall into the range of ZE in the input fuzzy membership function (please see Fig. 2). From the
Conclusion
A fuzzy logic controller to make IMM as adaptive IMM is designed, tested and realized in VLSI. This FLC has error in estimation and change in error as its inputs, and provides suitable transition probabilities (according to which the system model switches between velocity model filter and acceleration model filter) as outputs. Logic synthesis followed by extensive verification of proposed FLC architecture has been done. The results show that the FLC works as required in all realistically
Uncited references
[9], [10], [11], [12], [13], [14].
Acknowledgements
The author wishes to thank Mr. K. Ganesh Kumar, Mr. Sevvel, Ms. S. Sriprada and Ms. S. Devipriya of MIT for their help in completing the Fuzzy design. He also wishes to thank Mr. S. Varadarajan, Scientist, Radar Data Processing Division, LRDE, Bangalore for checking the fuzzy rules.
Ramesh Chidambaram obtained his Bachelors Degree in Electrical Engineering, with a specialization in Electronics & Instrumentation from the University of Madras (India) in 2002. In 2003 he was awarded the IEEE fellowship award and was an invitee to the 16th International Conference on VLSI design. He has been offered an opportunity to undergo graduate studies leading to a MSc degree in Microelectronics – under the Philips Semiconductors scholarship – at the Delft University of Technology in the
References (14)
- H. Watanabe, K.E. Yount, W.D. Dettloff, AVLSI fuzzy controller with reconfigurable cascadable architecture, IEEE Trans....
Signal Processing—the Model Based Applications
(1986)- C.C.K. Chan, Lee Vika, Leung Henry, Radar tracking for air surveillance in a stressful environment using a Fuzzy gain...
- V. Vaidehi, S. Varadarajan, S. Devipriya, S. Sripradha, C.N. Krishnan, Fuzzy based adaptive interacting multiple model...
- G. Ascia, V. Catania, M. Russo, VLSI hardware architecture for complex fuzzy systems, IEEE Trans. Fuzzy Systems 7 (5)...
- et al.
Interacting multiple model methods in target trackinga survey
IEEE Trans. Aerosp. Electron. Systems
(1998) - et al.
Fuzzy approach to maneuvering target tracking
IEE Proc. Radar Sonar Navig.
(1998)
Cited by (1)
A multi-context processor for real-time concurrent tasks fuzzy reasoning
2013, Journal of Intelligent and Fuzzy Systems
Ramesh Chidambaram obtained his Bachelors Degree in Electrical Engineering, with a specialization in Electronics & Instrumentation from the University of Madras (India) in 2002. In 2003 he was awarded the IEEE fellowship award and was an invitee to the 16th International Conference on VLSI design. He has been offered an opportunity to undergo graduate studies leading to a MSc degree in Microelectronics – under the Philips Semiconductors scholarship – at the Delft University of Technology in the Netherlands. His present interests include System-on-chip design, and design of wireless circuits.