Resistive Memory PUF Error Correction for Low-Correlation Identifier Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current resistive-switching memory technologies face challenges in generating data with low cross-correlation among bits, which is essential for unique identification and security applications, and lack effective error correction mechanisms to ensure data integrity.
Innovation Solution
The use of stochastic physical characteristics of resistive switching devices to generate data with low correlation, combined with error correction codes like Hamming code or Reed-Solomon code, to create a physical unclonable function (PUF) that can be used for identifier data and cryptographic keys, and implementing a method to write and correct errors in a resistive memory array.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If resistive-switching memory is used to generate identifier data, then unique identification and security applications are enabled, but cross-correlation among bits occurs reducing data quality
Solution Approach 1:
The patent extracts only the most significant bits from the resistive-switching memory output that exhibit low cross-correlation, discarding bits that show high correlation. This selective extraction maintains data quality for identification while eliminating the harmful correlated portions.
Solution Approach 2:
The patent introduces an intermediary processing stage that measures cross-correlation and selectively applies inversion operations. This intermediary layer transforms the raw memory output into corrected identifier data by mediating between the physical memory characteristics and the required data quality.
2Reliability
If error correction codes are implemented in resistive memory, then data integrity is improved, but device complexity increases
Solution Approach 1:
The patent performs preliminary error correction by measuring cross-correlation before final data usage and selectively inverting bits in advance. This preliminary action prevents errors from propagating without requiring complex real-time correction mechanisms during data processing.
Solution Approach 2:
The system performs self-diagnosis by measuring its own output cross-correlation and self-corrects by selectively inverting correlated bits. This self-service approach enables error correction without external intervention or complex additional hardware.
3Productivity
If formation pulse is applied to program resistive memory bits, then identifier data is generated, but stochastic variations cause bit errors
Solution Approach 1:
The patent implements feedback by measuring the actual output cross-correlation of the formed memory bits and using this measurement to guide selective inversion operations. This feedback loop compensates for stochastic variations introduced during formation pulse application.
Solution Approach 2:
The patent changes the state parameter of specific memory bits by selectively inverting them based on measured cross-correlation. This parameter change corrects errors introduced during formation without re-applying the formation pulse, maintaining productivity while improving precision.
Data Source
AI summary
Leveraging stochastic physical characteristics of resistive switching devices to generate data having very low cross correlation among bits of that data is disclosed. Data generated from stochastic physical characteristics can also be referred to as physical unclonable feature—or function—(PUF) data. Additionally, error correction functions for PUF data generated from resistive switching memory cells are provided. The error correction functions facilitate additional redundancy and longevity of PUF data, among other benefits. Different embodiments include addressing arrangements to incorporate ECC parity bits among generated PUF data bits, even for differential PUF bits respectively defined by multiple memory cells in different portions of a resistive memory array.