Machine Learning and Cryptographic Algorithms – Analysis and Design in Ransomware and Vulnerabilities Detection

,


INTRODUCTION
The impact and necessity of information security has increased exponentially over the last few decades as the denial-of-service attacks are increasing, ransomware attacks are increasing, information is being stolen from authenticated data sources, hackers are using more sophisticated and smart methods with help of agile tools for stealing sensitive information.Do small/mid-size/large, government and non-profit

19 April 2020
Machine Learning and Cryptographic Algorithms -Analysis and Design in Ransomware and Vulnerabilities Detection organizations need the security of their system?Yes.
They have sensitive user data, employee data, trading data, customer data and other sensitive confidential information stored in office systems.Do common people need the security of their systems at home?Yes.They may have their taxes files, social security card information, bank account details, private pictures, marketing strategy for their small business and many more private things.Some reports clearly shows the motivation behind ransomware is not just money but in favor of some nation's interest, but overall ransomware has affected a broad spectrum of organizations [1].
The large-scale ransomware attacks have increased 135% since 2015 and this has large impact on the performance of individuals [2].Cryptographic techniques used in the information technology domain to prevent the attacks with the help of complex cryptographic algorithm implementation.
Cryptography is used to encrypt the user or organization data and delete the original data from the user system and eventually ask for ransom to recover the hijacked sensitive data.

II.
What is the problem being solved?
This scientific model is based on the ardent requirement to protect the system and user data from the novel system attacks in the form of ransom and to threat the user identity data.The identification of suspicious pattern recognition through machine learning algorithm to resolve the major issue of the system attack in the form of ransomware and other type of suspicious activities on user system.The mainstream information security systems are mostly engaged in identity and access management, security policies and standardization of the policies but the undermine problem of resolving the system attacks with the help of cognitive and pattern recognition is remain unanswered.
The main system infections vectors are 1.Spam -unsolicited emails where the malware as an attached file 2. Corrupted web pagefiles is hacked by malware hacker, the files offered download as substituted for malware 3. Vulnerabilitieshacker delivers the malware to the host of the users operating system and benefits with these vulnerabilities 4. Phishingfake email and web pages to downloaded to run the malware With these system vectors the ransomware find the proper path the attack the system to gain the system hosted files in exchange with the ransom.Steps involved in the ransomware activity are [1]: Infection is attached vector in the form of email and reached to the user's system and executed by user.C& C server is contact command and control to obtain or store the encryption key.Then the user's data file is encrypted followed by the extortion money in the form of ransomware.
The main task in the ransomware lifecycle is encryption mechanism of the content of the user files, rendering those files unusable unless the user pays ransom to obtain a decryption key from the hacker.
Fast encryption requires a CPU resources, therefore there exists a tradeoff between the longer encryption time and CPU load both will eventually help into ransomware detection [1].
Any form of system hacking in this form can be prevented with the help of implementation of cryptographic algorithm with the machine learning algorithm.Some of the hypothesis of the problems are as describe below.Below table [4]gives the accuracy and

Figure shows the steps
Figure shows the steps involved in the ransomware detection for the standard ransomware detection mechanism.
Figure: Model for ransomware Detection and prevention

FRP-false positive
rate, TRPtrue positive rate With the text CNN method and cryptographic algorithm in place the model needs to be developed for the ransomware detection.Blending of machine learning with the cryptographic algorithm for the ransomware detection and further prevention is challenging work.VII.Conclusions In this paper, proposed machine learning algorithm can be modeled for the novel ransomware detection and the random number decryption techniques for the new model to break the encryption for saving the ransom.The study also suggested for the classification of the infectious files can be differentiated by the machine based on the model it has trained.The main problem has structurally divided into sub problems of the identification of the ransomware problems and the design the cryptographic algorithms based on the machine learning to generate the decryption key for the ransom problem.