Research On True Random Number Generator Based On Memristors

Random numbers are widely used in deep learning,Monte Carlo simulation,cryptography and other fields.Especially in the field of digital security,random numbers play an important role.In recent years,with the rapid development of the Internet of Things technology,people demand for digital security systems has increased.Traditional digital security systems are vulnerable to external attacks and still have shortcomings in security performance.As the key to digital security technology,the research of random numbers and random number generators has been paid more and more attention.Traditional true random number generators are restricted by entropy sources and often require complex equipment and algorithm assistance to ensure randomness,which greatly increases the complexity and power consumption of the random number generator.Compared with traditional semiconductor devices,memristors have many advantages such as good durability,fast conversion rate,and low power consumption.Random number generators based on memristors benefit from these advantages,which can improve the randomness of data and simplify the circuit.This article mainly studies the design and production of a new type of random number generator based on the memristor.The main research contents are as follows:1.The second chapter mainly conducted the analysis of the three-type memristors,and determined the selection of the memristor.First,I briefly analyzed the basic characteristics and internal mechanism of these 3 devices(Section 2.1).Then I did preliminary tests on these 3 devices.According to the test results,I performed the first screening of the devices,and selected more suitable devices for in-depth test analysis(Section 2.2)).Followed,I tested the random characteristics of the devices and the changes in device performance under irrational states(Section 2.3).Through the test results,comparing and analyzing the characteristics of the three types of devices,I determined the device selection(2.4 Section),and finally built a device model based on the test results to lay a good foundation for the subsequent design and testing(Section2.5).Through comparative tests,it was determined that the Ti N/Ge Se/Ti N memristor best meets the design requirements,and the modified memristive device is determined to be selected.2.The third chapter mainly completed the circuit design and simulation verification of the random number generator.First I determined the circuit design goal and clarified the design direction(section 3.1).Then I completed the overall design of the random number generator and designed a new random number generator based on the memristor(section 3.2).Finally,I selected the device according to the design plan and built a simulation circuit to verify the feasibility of the design(Section 3.3).For the first time,the design introduced the random response delay of the memristor and the random low-resistance state resistance value at the same time,made full use of the random resources of the memristor,further improves the randomness of the circuit,simplified the circuit structure of the random number generator,and strengthened the reliability of the circuit.It improved the rate of random number generation.The circuit added a rate controllable function,so that the random number generator can adapt to more application scenarios.3.The fourth chapter mainly completed the construction and realization of the random number generator circuit based on the circuit design,and completed the test and verification of the circuit performance indicators.First,according to the design goals,the test standards were determined and the test methods were clarified(section 4.1).Then,according to the test needs,the circuit connection method was adjusted,and the test plan was optimized(section 4.2).Then,according to the design plan,the random number generator was completed(Section 4.3).Finally I sort out the test results,compared the design goals and the existing advantages of the program,and determined the superiority of the circuit performance indicators(Section 4.4).The circuit can pass a number of NIST tests without complex data processing,the highest test rate is reached,it has a fault self-check mechanism,the maximum number of test cycles is up to 10~6,and it can achieve flexible control of the random number rate.Under non-ideal conditions,the structure,randomness,reliability,durability,random number rate,and random number rate control of the circuit are significantly better than the existing solutions.