RRAM Filament Comparison for Noise-Robust Random Number Generation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current random number generation technologies, particularly in non-volatile memory devices, face challenges in generating truly random numbers with high density and security, as they are susceptible to read noise and retention issues, limiting their robustness and reliability in secure hardware applications.
Innovation Solution
The use of resistive random-access memory (RRAM) devices with a specific structure featuring serially connected RRAMs and a dielectric layer, where a positive voltage forms filaments and a negative voltage randomly resets one device to a high resistance state, allowing for robust random number generation by comparing resistances and assigning logical values, thereby reducing sensitivity to noise and improving retention.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If software-based pseudo-random number generators are used, then deterministic number generation is achieved, but true randomness is not obtained
Solution Approach 1:
The patent replaces software-based pseudo-random generation with a hardware-based physical system using RRAM devices. The mechanical/electrical resistance states of filaments in the RRAM devices provide true physical randomness, substituting the deterministic software algorithm with an inherent physical random process at the hardware level.
Solution Approach 2:
The patent utilizes changes in resistance parameters of RRAM devices to generate random numbers. By measuring resistance values of filaments in different states (formed or not formed), the system converts physical parameter variations into random number outputs, achieving true randomness through physical parameter changes rather than software determination.
2Reliability
If conventional random number generation methods are used in non-volatile memory, then integration is achieved, but susceptibility to read noise and retention issues limits robustness
Solution Approach 1:
The patent divides the RRAM structure into multiple segments including bottom electrode, insulating layer, middle electrode, and top electrode. This segmentation allows independent control and measurement of different regions, enabling the system to isolate and measure filament resistance without being affected by noise from other parts of the device, thereby improving robustness against read noise.
Solution Approach 2:
The patent introduces a middle electrode as an intermediary element between the bottom and top electrodes. This intermediary structure serves as a reference point for measuring filament resistance and helps isolate the measurement from noise effects, improving the system's ability to withstand read noise and retention issues.
3Reliability
If filaments are formed through insulating layers, then low resistance states are achieved, but high resistance states require random filament disruption
Solution Approach 1:
The patent employs periodic voltage pulses applied to the RRAM devices to control filament formation and disruption. By applying voltage pulses at specific intervals and durations, the system can reliably transition filaments between low resistance (formed) and high resistance (disrupted) states, achieving stable resistance states through periodic electrical action.
Solution Approach 2:
The patent controls resistance state transitions by changing voltage parameters (magnitude, polarity, duration) applied to the RRAM devices. By adjusting these electrical parameters, the system can deterministically control whether filaments form or disrupt, enabling reliable switching between low and high resistance states despite the inherent randomness of individual filament behavior.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enables multiple generations of secure random number patterns, enhancing the robustness of physically unclonable functions (PUFs) and random number generators (RNGs) against noise, ensuring improved retention and security in hardware identification and cryptography.
Implementation Method 1
a first filament that is current conducting and extends through the at least one electrically insulating layer
Implementation Method 2
a voltage sensor configured to sense voltage of the third electrode in order to determine which one of the first filament or the second filament is more resistive
Data Source
AI summary
A random number generator comprising resistive random-access memory (RRAM) devices including: a first electrode; a second electrode; a third electrode located between the first and second electrode; at least one electrically insulating layer separating the first electrode and the second electrode from the third electrode, wherein the at least one electrically insulating layer has a substantially uniform thickness; a first filament that is current conducting and extends through the at least one electrically insulating layer; a second filament is located in the at least one electrically insulating layer and does not extend through the at least one electrically insulating layer; a voltage source configured to apply voltage to at least one of the first electrode and the second electrode; and a voltage sensor configured to sense voltage of the third electrode in order to determine which one of the first filament or the second filament is more resistive.