Data Scrambler for Persistent Memory Using Nested LFSRs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
3D crosspoint memory systems face issues with stuck cells, high raw bit error rates, and vulnerability to data hacking due to repeated writes and lack of robustness in scrambling techniques, especially when AES encryption is not used.
Innovation Solution
A data scrambler incorporating a Linear Feedback Shift Register (LFSR) with additional features like non-linear transformations, balanced mask patterns, and permutations to ensure no stuck cells and resilience against data hacking, using a combination of XOR operations and permutations based on data patterns.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a simple LFSR scrambler is used, then device complexity is reduced, but stuck cells occur and reliability deteriorates
Solution Approach 1:
The scrambler is divided into multiple independent LFSR units (first LFSR and second LFSR) that operate on different data portions. This segmentation allows each LFSR to be simpler while collectively providing comprehensive scrambling coverage that prevents stuck cells across the entire data set.
Solution Approach 2:
Multiple LFSR scramblers are nested within the memory controller architecture, with each LFSR handling specific data portions. The nested structure enables layered scrambling where outer LFSRs process already-scrambled data from inner LFSRs, creating cumulative randomness that prevents stuck cells without requiring any single LFSR to be overly complex.
2Reliability
If AES encryption is used, then security is improved, but device complexity and power consumption increase
Solution Approach 1:
Instead of using expensive, complex AES encryption, the patent employs multiple inexpensive LFSR-based scramblers that provide sufficient security for memory applications. The LFSRs generate pseudo-random sequences that are computationally simple yet effective at preventing data hacking, offering a cost-effective alternative to full cryptographic encryption.
Solution Approach 2:
The patent changes the security approach from cryptographic transformation (AES) to statistical randomness (LFSR sequences). By adjusting the parameters of multiple LFSRs (different polynomial configurations, tap positions, and seed values), the system achieves adequate security through mathematical randomness rather than complex encryption algorithms.
3Productivity
If data is written repeatedly to the same location, then productivity is improved, but stuck cells increase and reliability worsens
Solution Approach 1:
The scrambling operation is performed preliminarily on data before it is written to memory. By applying multiple LFSR scrambling layers in advance, the data is transformed into a pseudo-random pattern that prevents stuck cells during subsequent repeated write operations. This preliminary scrambling ensures that even if the same logical data is written multiple times, the physical bit patterns differ, preventing cell saturation.
4Device complexity
If scrambling is not used, then device complexity is reduced, but vulnerability to data hacking increases
Solution Approach 1:
The patent implements dynamic scrambling where multiple LFSRs with different configurations process data in a coordinated manner. The dynamic interaction between multiple LFSR units creates a more robust scrambling mechanism that is resistant to data hacking attempts, while keeping each individual LFSR relatively simple in structure.
Data Source
AI summary
Techniques and mechanisms to provide one or more substrates, and logic coupled to the one or more substrates. In an embodiment, the logic is to generate a pseudo-random sequence of bits, and to permute one or more bits of binary unscrambled data. In another embodiment, the logic is further to generate scrambled data based on an exclusive-or operation between the pseudo-random sequence of bits and the permuted data.