Numeric Time-Series Compression Using Adaptive Bit Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for compressing time series data, such as seismic data, are inefficient in reducing the volume of data while maintaining lossless compression, particularly when dealing with large files and varying bit lengths.
Innovation Solution
A computer system employing a model with multiple submodels, adaptive filters, and a mixer to predict bit probabilities, combined with arithmetic coding and secondary symbol estimation stages, effectively compresses and decompresses time series data by optimizing bit predictions and weight adjustments for improved compression ratios.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional compression methods (steim1, steim2) are used, then data is compressed into fixed formats, but compression efficiency is limited and cannot adapt to varying bit lengths
Solution Approach 1:
The patent implements dynamic adaptation by allowing the compression system to adjust bit allocation flexibly for each data value based on its magnitude and the statistical properties of the time series data. Unlike fixed-format methods, the system dynamically determines the optimal number of bits (1, 2, 4, 8, 16, or 32 bits) needed for each value, enabling efficient compression while adapting to varying data characteristics.
Solution Approach 2:
The system changes the parameter of bit allocation from fixed to variable. By modifying the number of bits used to represent each data value based on statistical analysis and contextual information, the system achieves higher compression efficiency while maintaining the ability to represent a wide range of values with appropriate precision.
2Measurement precision
If more bits are used per value, then measurement precision is maintained, but data volume increases reducing compression effectiveness
Solution Approach 1:
The patent applies local quality by allocating different numbers of bits to different data values based on their specific requirements. Instead of using a uniform bit allocation, the system analyzes the local characteristics of each value and its context, assigning minimal bits to values that can be accurately represented with fewer bits while allocating more bits only where necessary to maintain precision.
Solution Approach 2:
The system dynamically changes the bit allocation parameter for each data value based on statistical properties and contextual information. This allows the system to maintain measurement precision where needed while reducing data volume by using fewer bits for values that do not require high precision, achieving an optimal balance between these two parameters.
3Productivity
If complex compression algorithms are used, then compression ratio improves, but computational complexity and processing time increase
Solution Approach 1:
The patent performs preliminary statistical analysis of the time series data before compression to identify patterns, ranges, and optimal bit allocation strategies. By pre-processing the data to determine statistical properties and contextual relationships, the system simplifies the subsequent compression process while maintaining high compression ratios, avoiding the need for overly complex real-time algorithms.
Solution Approach 2:
The system uses feedback from statistical analysis and contextual information to dynamically adjust compression parameters. By continuously monitoring data characteristics and adapting the compression strategy based on observed patterns, the system achieves high compression ratios with manageable computational complexity, as the feedback mechanism guides efficient decision-making without requiring exhaustive search algorithms.
Data Source
AI summary
Embodiments described herein relate to compression and decompression of data consisting of a one dimensional time series of floating point numbers. A compressor may comprise a lossless stage and in some embodiments a lossy stage in addition to the lossless stage. The lossy stage quantizes the data by discarding some of the least significant bits as specified by the user. The lossless stage uses a context mixing algorithm with two bit-wise predictive models whose predictions are combined and fed to an arithmetic coder. One model is a direct context model using the most significant bits of prior numeric samples as context. The other model is the output of an adaptive filter, in which the approximate predicted numeric value is used as context to model the actual value. A corresponding decompressor uses the same lossless model with the arithmetic coder replaced by an arithmetic decoder.