DEVELOPMENT OF ADVANCED ENCRYPTION STANDARD AND NEURAL DATA SECURITY MODEL FOR PREVENTING E-COMMERCE RELATED CRIMES

Abstract

Growth rate and aspiration for e-Commerce usage is getting increased on daily basis because today’s world is getting and getting more interconnected than ever before. For all its merits, increased connectivity also brings about increase risk in theft, fraud, and abuse of the privilege. There is a need to protect information or messages transmitted from getting to the third party, intruder and or unauthorized person(s). There are numerous encryption and decryption security plans developed to protect information transmitted through networks and especially in a business context as these provide rapid and timely exchange of information. However, the problem of insecurity has been a major challenge in e-Commerce because, if the merchant sites are accessed without adequate security measure in place, the messages can be intercepted, manipulated and/or altered without the original owner of the message being aware. As a result,there is need to encrypt communication or exchange of information between two or more people. Therefore, this work proposes a scheme based on Advanced Encryption Standard (AES) and Neural Data Security Model (NDSM). The AES using Time-based Dynamic Key Generation work by securing customers details submitted into the merchant database and ensures that users are authenticated before gaining access. However, the AES program was written as multithreaded in order to enhance speedy runtime. Neural Data Security Model (NDSM) involves fragmenting or partitioning of sensitive data into High and Low level Confidentiality. The sensitivity of the data determines which category of confidentiality either low or high level the data will fall into. The fragmented data are saved into two different databases, on two different servers and on the same datacenter. The research work was carried out using tools such as PHP,HTML, CSS, MySQL, JavaScript and Apache for it development of the interface and backend while the results obtained were analyzed using Performance Metric (Confusion Matrix, Accuracy and Kappa) and classification of algorithms(Random Forest, Naïve Bayes, Random Tree and J48). Out of the four algorithms, Naive Bayes seems to perform better but with little margin as against other algorithms.Naive Bayes performance result, a total of 105 instances where observed out of which 85.71% where correctly classified while 14.29% were misclassified.With Kappa value of 0.690 shows good performance classification of fraudulent and Non-fraudulent cases of transaction. In summary, it can be concluded that using AES and Neural Data Security Model improves the prevention of e-Commerce related crimes.