1978 IEEE TRANSACTIONS ONINSTRUMENTATION AND MEASUREMENT, VOL. 57, NO. 9, SEPTEMBER 2008 Circuit Testing Using the Principles of Self–Nonself Discrimination Cleonilson Protásio de Souza, Francisco Marcos de Assis, and Raimundo Carlos Silvério Freire, Member, IEEE Abstract—As complex very-large-scale integration circuit test- ing using external automatic test equipment is becoming increas- ingly expensive, built-in self-test (BIST) is an alternative technique that can significantly reduce the cost of testing. On the other hand, artificial immune systems have been considered as one of the most promising nature-inspired techniques used for novelty detection systems. One of the main features of such systems is self–nonself discrimination, which enables the body to distinguish any foreign cell from the body’s own cell. In this paper, based on the principles of self–nonself discrimination, an immune-based output response analyzer for BIST is presented. Using the proposed scheme, the evaluation of the circuit under test (CUT) is performed by a set of immune-based detectors that are capable of detecting faulty output responses. Since the number of detectors is proportional to the hardware overhead consumed by the scheme, it is also presented as an optimization algorithm to reduce the number of detectors. Using the reduced set of immune-based detectors, zero aliasing is achieved, and, in addition, the evaluation of the CUT is performed during the test. The experimental results show the effectiveness of the proposed scheme. Index Terms—Artificial immune system, built-in self-test (BIST), output response analyzer (ORA), self–nonself discrimina- tion, zero aliasing. I. I NTRODUCTION B UILT-IN self-test (BIST) is a promising test technique that incorporates additional hardware into integrated circuits to allow them to test their own operations, thereby reducing the dependence on very expensive external automated test equipment [1]. In general, such additional hardware devices are a test pattern generator (TPG), which applies a test pattern sequence to the circuit under test (CUT), and an output response analyzer (ORA), which evaluates the test response sequence, as shown in Fig. 1. Regarding the ORA, its fundamental block is a compressor that compresses the test response sequence into a compact signature [2]. To evaluate whether the CUT is fault free or not, the ORA compares the obtained signature with a predetermined fault-free signature at the end of the test. Due to the loss of information during compression, the CUT might be declared fault free, although it is actually faulty. Such a failure is called aliasing [1]. Aliasing is one of the disadvantages of Manuscript received July 5, 2006; revised January 17, 2008. C. P. de Souza is with the Department of Electrical and Electronic, Federal Center of Technological Education of Maranhão, São Luís 65030-000, Brazil (e-mail: protasio@cefet-ma.br). F. M. de Assis and R. C. S. Freire are with the Department of Electrical Engineering, Federal University of Campina Grande, Campina Grande 58109- 970, Brazil (e-mail: fmarcos@dee.ufcg.edu.br; rcsfreire@dee.ufcg.edu.br). Digital Object Identifier 10.1109/TIM.2008.919002 Fig. 1. BIST environment. The TPG applies a test pattern sequence T = {tn,...,t 2 ,t 1 } in the CUT. The test response sequence R = {rn,...,r 2 ,r 1 } is applied in the ORA that compresses R into a signature that is compared with a predetermined fault-free signature. Such a comparison determines whether the CUT is fault free or not. signature-based ORA. Another disadvantage is that the CUT is only evaluated at the end of the test. The natural systems that may be used as a model for error de- tectors are the biological immune ones, which are very complex systems with several mechanisms for defense against patho- genic organisms. The primary purpose of immune systems is to recognize all cells within the body and categorize those cells as the body’s own cell (self) or foreign cell (nonself). Such a recog- nition process is known as the self–nonself discrimination one. After discrimination, the nonself cells are further categorized to induce an appropriate type of defensive mechanism [3]. With the ability to detect nonself, immune systems seem to be an adequate source of inspiration to the development of algorithms for the early detection of anomalous behavior in systems [4]. Artificial systems coming from immune systems are called artificial immune systems, and they are being considered as one of the most promising nature-inspired techniques used for novelty detection systems [5]. In this paper, an ORA for BIST, called immune ORA, which takes inspiration from the principles of self–nonself discrim- ination of immune system, is presented. Using the proposed immune ORA scheme, the evaluation of the CUT is performed by a set of precomputed immune-based detectors that can detect a faulty circuit output response. Since the larger is the number of detectors and the larger will be the hardware overhead consumed by the scheme, it is also presented as an optimization algorithm to reduce the amount of detectors. Using the reduced set of immune-based detectors, zero alias- ing is achieved, and the evaluation of the CUT is performed during the test. The experimental results show the effectiveness of the proposed scheme. II. NEGATIVE SELECTION ALGORITHM Basically, in biological immune systems, the mechanism for the detection of pathogenic organisms consists of lymphocytes that can be thought as detectors that can recognize pathogens 0018-9456/$25.00 © 2008 IEEE