Hybrid of artificial immune system and particle swarm optimization-based support vector machine for Radio Frequency Identification-based positioning system

https://doi.org/10.1016/j.cie.2012.10.007Get rights and content

Abstract

This study intends to propose a hybrid of artificial immune system (AIS) and particle swarm optimization (PSO)-based support vector machine (SVM) (HIP–SVM) for optimizing SVM parameters, and applied it to radio frequency identification (RFID)-based positioning system. In order to evaluate HIP–SVM’s capability, six benchmark data sets, Australian, Heart disease, Iris, Ionosphere, Sonar and Vowel, were employed. The computational results showed that HIP–SVM has better performance than AIS-based SVM and PSO-based SVM. HIP–SVM was also applied to classify RSSI for indoor positioning. The experiment results indicated that HIP–SVM can achieve highest accuracy compared to those of AIS–SVM and PSO–SVM. It demonstrated that RFID can be used for storing information and in indoor positioning without additional cost.

Highlights

► Propose a hybrid of AIS and PSO-based SVM (HIP–SVM) for optimizing SVM parameters. ► Apply HIPSVM to RFID-based positioning system. ► The results showed that HIP–SVM has better performance than AIS and PSO.

Introduction

Radio frequency identification (RFID) has the characteristics of contactless, to be reused, durable, the mass storage capacity and multi-read and is suitable to be used in cargo tracking and information collection. More and more applications have been developed for logistics, like warehouse management and in-transit inventory management (Weinstein, 2005). In addition to the applications mentioned above, RFID may also be applied in the indoor positioning system. RFID is one of the recent wireless communication technologies. Due to some physic characteristics, like received signal strength index (RSSI) and the arrival time of radio frequency (RF) between the interrogator and tag, it can be utilized to locate the goods position or picking cart position. After knowing the cart position, once a new order is coming, it can be used to plan the picking rout in order to minimize the picking distance.

On the other hand, support vector machine (SVM) (Vapnik, 1995) has been widely applied in many areas for classification. Thus, this study intends to present a novel SVM based on a hybrid of artificial immune system (AIS) and particle swarm optimization (PSO) (HIP–SVM). The hybrid of AIS and PSO is employed to optimize the parameters combination for SVM. In order to assess HIP–SVM’s capability, six benchmark data sets, Australian, Heart disease, Iris, Ionosphere, Sonar and Vowel, are first employed. Then, HIP–SVM is applied to classify RSSI for indoor positioning.

The rest of paper is organized as follows. Section 2 presents some related literature survey, while the RFID-based position system is proposed in Section 3. Section 4 demonstrates the capability of the proposed HIP–SVM using six benchmark data sets. Section 5 shows the model evaluation results the proposed RFID-based position system through simulation. Finally, concluding remarks are made in Section 6.

Section snippets

Background

This section will briefly introduce radio frequency identification, support vector machine, particle swarm optimization, and artificial immune system.

Methodology

This section will present the proposed hybrid of artificial immune system and particle swarm optimization-based support vector machine (HIP–SVM) for optimizing SVM parameters including penalty constant, C, and parameters of kernel function which can dramatically influence the classification accuracy.

Computational experiences

To examine accuracy and effect of the proposed method, this section uses UCI benchmark data sets for classification and compares the results with those of AIS–SVM and PSO–SVM.

Rfid-based position system

This section establishes RFID experiment environment to receive signal data and create a signal characteristic database. With the current signal characteristic values, the classification methods can achieve indoor positioning function on the condition that the cost is not increased.

From Section 4, the HIP–SVM classification efficiency is better than other two methods. This section uses HIP–SVM to construct classification model and the optimal parameter combination from Taguchi experiment is

Conclusions

This study has proposed a hybrid of artificial immune system and particle swarm optimization for support vector machine. Through the hybrid method, SVM parameters can be determined optimally. Besides, using the proposed HIP–SVM to the RFID-based positioning system, it can provide the highest accuracy compared with those of AIS–SVM and PSO–SVM. Also, the proposed HIP–SVM is more stable than the other two methods due to lower standard deviation. In the future, other AIS can be applied for

Acknowledgement

This study was financially supported by the National Science Council of the Taiwan Government, under Contract Number NSC99-2221-E-011-057-MY3. This support is greatly appreciated.

References (38)

  • M. Clerc

    The swarm and the queen: Towards a deterministic and adaptive particle swarm optimization

    Evolutionary Computation

    (1999)
  • C. Cortes et al.

    Support vector networks

    Machine Learning

    (1995)
  • D. Dasgupta

    Advances in artificial immune systems

    Computational Intelligence Magazine, IEEE

    (2006)
  • de Castro, L. N., & Von Zuben, F. J. (2000). The clonal selection algorithm with engineering applications. In...
  • L.N. de Castro et al.

    Learning and optimization using clonal selection principle

    IEEE Transactions on Evolutionary Computation

    (2002)
  • Eberhart, R., & Kennedy, J. (1995). A new optimizer using particle swarm theory. In Proceedings of the sixth...
  • Eberhart, R., & Shi, Y. (2001). Particle swarm optimization: Developments, applications and resources. In Proceedings...
  • Eberhart, R., & Shi, Y. (2001). Tracking and optimizing dynamic systems with particle swarms. In IEEE international...
  • Gasper, A., & Collard, P. (1999). From GA’s to artificial immune systems: Improving adaptation in time dependent...
  • Cited by (0)

    This manuscript was processed by Area Editor Satish Bukkapatnam.

    View full text