Implementation of Enhanced A5/1 Stream Cipher and its Randomness Analysis by NIST Test Suite

Global System for Mobile (GSM) is still widely used cellular standard providing many services like Voice, SMS & Data Service with mobility. To secure information in the GSM cellular network, security is implemented at two levels i.e. initially to authorize the valid users for every location update and then during call, encryption is applied over the information being transmitted on GSM channel to protect from being intercepted and decoded by unauthorized persons. This protection is achieved by converting the original message into an encoded form (cipher form) that appears to be a random stream of symbols. Under "Challenge-Response" mechanism, A3 and A8 algorithms are used to generate triplet (RAND, SRES & K-c) for authentication of any user at GSM network. But the actual information is encrypted using A5 algorithm to generate cipher stream for better protection from interception. This A5 stream cipher has three versions: A5/1, A5/2 and A5/3. A5/1 works on Liner Feedback Shift Registers (LFSRs) with irregular clocking and generates pseudo random binary stream. These three versions of A5 algorithm are being used for encryption of information over GSM since the launching of GSM services and have been cryptographically analyzed by Reverse-Engineering. With passage of time, many hackers & crackers are becoming able to break this encryption identifying some weaknesses of these algorithms and can decrypt the original information. These stream ciphers are facing some weaknesses like poor Liner Complexity (LC) & clocking mechanism (Majority Rule), short clocking period, weak choice of clocking taps and collision problem. Because of such weaknesses, these stream ciphers can be decrypted by intruders. In this papers an attempt has been made to reduce these weaknesses and enhance the security by introducing non-linear combinational generator (NLFSRs), reuse of 32 bits SRES generated by A3 algorithm and finally combining the output stream with last 32 bits of CGI. The randomness analysis of proposed stream cipher is carried out by NIST Statistical Test Suite and it is confirmed by comparison of the randomness parameters results that the randomness of bit-stream produced by the proposed stream cipher has improved significantly hence the enhanced security can be achieved.