Block Floating Point Compression with Exponent Difference Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing signal compression methods fail to efficiently manage dynamic range and bit allocation for signal samples, leading to suboptimal resource utilization and quality metrics in signal processing applications.
Innovation Solution
The method involves grouping signal samples into blocks, determining a common exponent and allocating bits to mantissas based on this exponent, and encoding these values to form compressed data, with feedback control for maintaining a fixed output data rate or quality metric.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If block floating point representation is used to increase dynamic range, then the number of bits representing signal samples can be reduced, but the complexity of determining common exponents and allocating bits increases
Solution Approach 1:
The patent divides the signal into blocks of samples and processes each block independently with a common exponent. This segmentation allows the system to manage complexity by handling smaller groups at a time while achieving overall compression. Each block gets its own exponent determination and bit allocation, making the complex process manageable and systematic.
Solution Approach 2:
The patent dynamically adjusts the number of bits allocated to mantissas based on the determined exponent value. When the exponent indicates a larger dynamic range is needed, fewer bits are allocated to the mantissa, and vice versa. This parameter change strategy allows flexible adaptation to different signal characteristics while maintaining efficient compression.
2Productivity
If variable bit allocation based on exponent is implemented, then compression efficiency improves, but the difficulty of encoding and decoding increases
Solution Approach 1:
The patent determines the common exponent and performs bit allocation calculations before actually encoding the mantissa values. This preliminary action establishes the encoding scheme in advance, allowing the encoder to systematically apply the determined bit allocation without complex real-time decisions during the encoding process itself.
Solution Approach 2:
The patent implements a feedback mechanism where the determined exponent value directly influences the bit allocation for mantissas, and this allocation information is fed back into the encoding process. The decoder receives both the exponent and the allocated bit information, using it to correctly reconstruct the original signal values with appropriate precision.
3Measurement precision
If common exponent is determined for each block, then dynamic range representation improves, but the processing time and computational resources increase
Solution Approach 1:
By dividing the signal into blocks and determining a common exponent for each block rather than processing the entire signal at once, the patent reduces computational burden. Each block's exponent determination is a smaller, more manageable task that can be performed quickly, and multiple blocks can be processed in parallel or sequentially with minimal overhead.
Solution Approach 2:
The patent makes each block self-contained with its own common exponent and bit allocation scheme. This self-service approach allows blocks to be processed independently without requiring complex global optimization, reducing processing time while maintaining accuracy within each block. The systematic approach of determining exponents and allocating bits for each block automates the process efficiently.
Data Source
AI summary
A method and apparatus for compressing signal samples uses block floating point representations where the number of bits per mantissa is determined by the maximum magnitude sample in the group. The compressor defines groups of signal samples having a fixed number of samples per group. The maximum magnitude sample in the group determines an exponent value corresponding to the number of bits for representing the maximum sample value. The exponent values are encoded to form exponent tokens. Exponent differences between consecutive exponent values may be encoded individually or jointly. The samples in the group are mapped to corresponding mantissas, each mantissa having a number of bits based on the exponent value. Removing LSBs depending on the exponent value produces mantissas having fewer bits. Feedback control monitors the compressed bit rate and/or a quality metric. This abstract does not limit the scope of the invention as described in the claims.