Pattern-Recognition Processor Parallel Finite State Machines
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing pattern recognition systems face bottlenecks in processing large volumes of data due to the increasing number of patterns to be identified, leading to delays and high memory consumption when searching for multiple patterns in a data stream.
Innovation Solution
A pattern-recognition processor that uses parallel finite state machines to evaluate multiple search criteria simultaneously, allowing for efficient searching of data streams with reduced memory usage by deconstructing search criteria into smaller terms and utilizing a compiler to format search criteria for native support of regular expressions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple patterns are searched sequentially in a data stream, then each pattern can be detected accurately, but the processing time increases and system performance degrades
Solution Approach 1:
The patent segments the data stream processing into parallel finite state machines, each dedicated to detecting a specific pattern. Instead of sequentially searching for multiple patterns, the system divides the detection task across multiple concurrent state machines that process the data stream simultaneously, thereby reducing overall processing time while maintaining detection accuracy for each pattern
Solution Approach 2:
The patent transitions from sequential pattern matching to parallel processing by introducing a temporal dimension. Multiple finite state machines operate concurrently on the same data stream, effectively adding a parallel processing dimension that allows simultaneous pattern detection without compromising the accuracy of individual pattern matching
2Ease of operation
If copies of the data stream are created for each pattern match, then matching data can be examined, but memory consumption increases significantly
Solution Approach 1:
The patent implements a selective copying mechanism where only the portions of the data stream that match specific patterns are copied and stored, rather than creating copies for all pattern matches. The finite state machines track match positions and selectively replicate only the relevant matching segments, significantly reducing memory consumption while still allowing examination of matching data
Solution Approach 2:
The patent extracts only the essential matching portions of the data stream rather than copying entire segments. The finite state machines identify and extract specifically the data portions that constitute pattern matches, discarding non-matching portions and thereby reducing the total volume of data that needs to be stored in memory
Data Source
AI summary
Disclosed are methods and devices, among which is a device that includes a pattern-recognition processor. The pattern-recognition processor may include a matching-data reporting module, which may have a buffer and a match event table. The buffer may be coupled to a data stream and configured to store at least part of the data stream, and the match event table may be configured to store data indicative of a buffer location corresponding with a start of a search criterion being satisfied.