ANBD Data Memory Protection Using Integer-Normalized Variables
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data processing systems face challenges in efficiently performing floating-point arithmetic operations due to complexity and high computational overhead, making it impractical for safety-related applications, where the focus is on detecting maximum errors rather than precision.
Innovation Solution
A method that normalizes floating-point variables into integer multiples of a base for ANBD encoding, allowing operations to be performed in integer arithmetic, reducing computational overhead and maintaining flexibility in value range, with encoding and normalization performed during code generation to minimize runtime processing time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If floating-point arithmetic is implemented in ANBD encoding scheme, then precision is improved, but computing time increases by orders of magnitude
Solution Approach 1:
The patent changes the parameter representation from floating-point to fixed-point with integer arithmetic. By representing variables as integers scaled by a fixed point position (e.g., multiplying by 1000 to represent decimal values as integers), the system achieves comparable precision without the computational overhead of floating-point operations in ANBD encoding.
Solution Approach 2:
The patent uses simpler integer arithmetic operations instead of complex floating-point operations. Integer arithmetic is computationally cheaper and faster to execute, sacrificing the dynamic precision range of floating-point but gaining significant speed improvement for safety-critical applications where maximum precision is not always required.
2Adaptability or versatility
If floating-point numbers are processed in ANBD encoding, then value range flexibility is improved, but device complexity increases
Solution Approach 1:
The patent changes from floating-point parameters to fixed-point integer parameters with a predetermined scale factor. This simplifies the data structure and processing logic while maintaining adequate value range flexibility through appropriate scaling (e.g., using integers to represent decimal values by multiplying by a power of 10).
Solution Approach 2:
Instead of representing small fractional values directly as floating-point numbers, the patent inverts the approach by representing them as larger integers with an implicit decimal point position. This inversion simplifies arithmetic operations while preserving the ability to represent a wide range of values.
3Productivity
If uncoded floating-point processing is used, then computing speed is improved, but error detection capability is lost
Solution Approach 1:
The patent segments the data representation into coded variables that include both the functional value and check bits for error detection. By encoding variables according to the ANBD scheme (xc = A*x + Bx + D), the system maintains computing speed through efficient integer arithmetic while adding error detection capabilities through the structured encoding with check bits.
Solution Approach 2:
The patent introduces coded variables as an intermediary between the computational operations and the actual data values. These coded variables carry both the computational value and error detection information, allowing standard integer arithmetic to proceed efficiently while the encoding structure provides automatic error detection without requiring separate verification steps.
Data Source
AI summary
A method for protecting data in a data memory against an undetected change, wherein a functional variable x is encoded via a value, an input constant, an input signature and a timestamp D into a coded variable, where the functional variable is normalized relative to a base to form the integer value from the functional variable.