Multimodal Biometric System Based On Decision Level Fusion Abstract— A Biometric Framework is one of the crucial Pattern Recognition Framework that are utilized for recognizing individuals utilizing distinctive Biometric Characteristics . The Authentication System design using single modality may not fulfill the prerequisite of requesting applications in term of properties, for example, Accuracy, Acceptability and Performances. Due to its limitations , Multimodal Biometrics has been introduced where fusion of the modalities is the bigger challenge. Multimodal Biometric Framework is utilized as productive biometric framework which is a combination of two or more biometric attributes to upgrade the security. In Multimodal Biometrics, Fusion can be performed on different levels. In this paper, Fusion at decision level is implemented where different decision level fusion techniques have been tested on the Iris and Fingerprint samples on standard dataset and performance of the system is measured on the basis of False Acceptance Rate(FAR), False Rejection Rate(FRR) and Recoginition Accuracy. Key words—Multimodal ;Security;Fusion;KNN;Neural 1.Introduction Establishing the individuality of a person is a decisive task in any identity management system. Stand-in representations of personality, for example, passwords and ID cards are not adequate for solid character determination since they can be effortlessly lost, shared, or stolen. Biometric acknowledgment is the investigation of building up the character of a man utilizing his/her anatomical and behavioral characteristics. Usually utilized biometric qualities incorporate unique mark, face, iris, hand geometry, voice, palmprint, transcribed marks. Three components that affected the expanded enthusiasm for the biometric are as per the following: 1) open acknowledgment; 2) new easy to use catch gadgets with wide enhanced abilities; and 3) an expanded scope of uses. Biometrics is a kind of authentication techniques which place confidence in measurable individual and physiological characteristics which will be mechanically verified[3]. Counting on the appliance context, a biometric system could operate either in identification mode or verification mode. Because the level of security breaches and dealings fraud have increased, the necessity of technologies for extremely secure identification and private verification is changing into apparent. Biometric-based[4] solutions are providing the data privacy and confidential transactions. Multi-biometric system uses different i.e. more than one biometric systems for verifying the person identity [5]. This technique exploits the capacities of each individual biometric. These frameworks will have higher exactness attributable to the very reality that they utilize numerous biometric modalities where each individual methodology introduces enough verification to recognize the person. Multimodal biometric systems[6] utilize two or more biometric tests and blend there examination utilizing combination to give a superior choice in indentifying and in the meantime diminishing the FRR and FAR[7]. All uni-modular biometric frameworks are utilized with blend of others to make a multimodal biometrics. For example: a. Iris and Fingerprint b. Face and Palmprint c. Signature and Voice 1.1 Modality Choice In this work, Iris and Fingerprint modalities are combined as both modalities are unique identification of the person and are widely accepted. Although the mixture of multimodal enhances accuracy and security, but still there is an increase in teams of system complexity as the feature extracted from the various sample have increased and also there is increase in cost in terms of acquisition time[8]. Thus the key issue these days, are up to what extent the feature are to be extracted and also to minimise the cost, because the range of options will increase the variability of the intra-personal samples as a result of bigger lag times in between consecutive acquisitions of the sample conjointly will increase[9]. The FAR will increase with the increase in variability of the system. 1.2 Fusion Level Multimodal systems[10] can be combined in numerous different levels as described below: Sensor level: From the different sensors crude information obtained is handled and consolidated with a specific end goal to get new information so that the elements can be extricated from it[11]. For example, in face biometrics, the 3D profundity and 2D surface information (which are gained utilizing two entirely unexpected sensors) are amalgamated to get a 3D surface picture of the face that may then be subjected to highlight extraction and matching[12] Suneet Narula Garg Research Scholar,University Institute Of Engineering And Technology, Punjab University, Chandigarh Renu Vig Professor,University Institute Of Engineering And Technology, Punjab University, Chandigarh Savita Gupta Professor,University Institute Of Engineering And Technology, Punjab University, Chandigarh