Pattern Recognition Letters 133 (2020) 341–348 Contents lists available at ScienceDirect Pattern Recognition Letters journal homepage: www.elsevier.com/locate/patrec Genetic algorithm based key sequence generation for cipher system K.B. Sudeepa a,∗ , Ganesh Aithal b , V. Rajinikanth c , Suresh Chandra Satapathy d a Department of Computer Science and Engineering, NMAM Institute of Technology, Nitte, Karnataka, India b Shri Madhwa Vadiraja Institute of Technology and Management, Bantakal, Karnataka, India c Department of Electronics and Instrumentation Engineering, St. Joseph’s College of Engineering, Chennai 600 119 Tamilnadu, India d School of Computer Engineering, Kalinga Institute of Industrial Technology (Deemed to Be University), Bhubaneswar, 751024, Odisha, India a r t i c l e i n f o Article history: Received 23 September 2019 Revised 28 February 2020 Accepted 9 March 2020 Available online 12 March 2020 Keywords: Feedback shift register Genetic algorithm Stream cipher system Pseudo random number generator Cross over Mutation a b s t r a c t Stream cipher system seems to be one of the best alternatives in order to provide confidentiality and security in an on line and high-speed transmission. Cryptography is required to secure the secret infor- mation transmitting over the communication channel. Day by day the importance of security increases due to increase of online transaction processing and e-commerce. As well as encryption / decryption algorithm, randomness characteristics of key sequences prove the strength of the stream cipher. In this work the Linear Feedback Shift Register (LFSR) is used to produce non binary pseudo random key sequence. The length of the sequence has been enhanced by designing hybrid model using LFSR and Ge- netic Algorithm (GA). Achieving the length more than the maximum length of LFSR is the primary inten- tion of this work. The statistical tests is conducted to assess the randomness of key sequence generated from hybrid model. Generated key sequences are used as key sequences in cryptographic applications and results are analyzed. © 2020 Elsevier B.V. All rights reserved. 1. Introduction As present days the demand for the internet access is increas- ing, securing the data from the unauthorised users over the in- ternet is a major concern. It is very much needed to control the accessing of the valuable information by the unauthorised users. Problems of confidentiality and authentications are solved by the cryptographic operations. The confidentiality helps to prevent the access of original data over the internet from unauthorised users, and authentication helps to prevent the unauthorised users from sending the messages. It confirms that the message is sent by the genuine sender. To provide confidentiality, a perfect security system is needed. Shannon [1] defines the perfect security as, it should have individ- ual unique key is used for each data element to encrypt. This is possible only by one-time pad indicating a large sequence of non- repetition of sequence is required. Also, the efficiency of stream cipher system increases as the generated key sequence satisfies the randomness properties to the maximum extent. Especially the length of the sequence plays major role. This paper works on designing hybrid system to generate non binary key sequence using existing LFSR system and GA, such that ∗ Corresponding author. E-mail address: sudeepa@nitte.edu.in (K.B. Sudeepa). it extends length of key sequence which is long enough to have one-time pad for any given system. Before going to the work, it is necessary to discuss a bit of random numbers. The generator of random numbers classified into two cate- gories; (i) True random number and (ii) Pseudo random number. A number generated based on a real-world situation is called true random number. Entropy of real-world situations will be very high, and its prediction will be very difficult [2]. Number generator that does not depend upon the real-world scenario is called Pseudoran- dom Number Generators (PRNGs). Such generator produces num- ber by mathematical functions or by an algorithm that appear ran- dom. [3]. Even though these Pseudorandom Number Generators of binary nature is used more in the application, it will have a drawback of dependency of the randomness for the strength of the cryptography. This can be overcome by introduction of non- binary cryptographic techniques. In this case the strength depends on the algorithm of encryption and the strength of the PRN se- quence generated. Next, we discuss some of the literature based on non-binary oriented pseudo random numbers, which will assist this work. The related work is discussed below. 1.1. Related works Lehrer [4] presented the linear congruential technique to pro- duce pseudo random number where a new number (X n+1 ) ob- https://doi.org/10.1016/j.patrec.2020.03.015 0167-8655/© 2020 Elsevier B.V. All rights reserved.