Method and apparatus for efficient compression
A compression method and apparatus identifying candidates for compression by selectively fingerprinting shingles or overlapping subsets of an input dataset and creating a set of characteristic input fingerprints based on fingerprint value. In some cases, the characteristic fingerprints are selected based on the relative value of the fingerprints with respect to other fingerprints in the same cluster. Potential matches may be identified and confirmed by comparing the characteristic input fingerprints with fingerprints associated with a history. Advantageously, some examples according to the current invention may be applied to input data such as: data, files, bit streams, byte streams, packet streams and previously encoded, compressed and/or encrypted data. In some cases, the number of fingerprints selected per cluster, the cluster size, the shingle size or shingle configuration, the size or configuration of the overlap between successive shingles, and the configuration for adding entries to a history, may be adaptively configurable.
1. A method for compressing an input dataset comprising the steps of:
selectively fingerprinting the input dataset, the step of selectively fingerprinting comprising the steps of:
fingerprinting shingles of the input dataset wherein shingles comprise overlapping subsets of the input dataset; and,
selecting at least one characteristic input fingerprint per cluster based on the value of the fingerprint, thereby creating a set of characteristic input fingerprints;
identifying a candidate subset of the input dataset that is a match with a target subset of data in a second selectively fingerprinted dataset by searching for fingerprint matches between one or more input fingerprints from the set of characteristic input fingerprints and fingerprints associated with the second selectively fingerprinted dataset; and,
performing a compression operation on the input dataset.
2. The method of claim 1 wherein the step of identifying a candidate subset of the input dataset that is a match further comprises the step of:
identifying fingerprint matches that correspond to exact matches between the candidate subset and the target subset; and,
wherein:
the step of performing a compression operation on the input dataset based on matches comprises the step of performing a compression operation on the input dataset based on exact matches.
3. The method of claim 1 wherein the step of performing a compression operation on the input dataset comprises the step of:
performing a compression operation on the input dataset based on fingerprint matches.
4. The method of claim 1 wherein the step of performing a compression operation on the input dataset comprises the step of:
performing a compression operation on the input dataset based on matches between a subset of the input dataset and a subset of data in a second selectively fingerprinted dataset.
5. The method of claim 1 further comprising the step of extending the size of a match by comparing input dataset elements outside of the candidate subset to dataset elements outside of the target subset in the second selectively fingerprinted dataset.
6. The method of claim 1 further comprising the step of extending the size of a match by comparing fingerprints adjacent to the fingerprint matches.
7. The method of claim 1 wherein the cluster size associated with the step of selectively fingerprinting the input dataset is consistent across the input dataset.
8. The method of claim 1 wherein the cluster size associated with the step of selectively fingerprinting the input dataset changes across the input dataset.
9. The method of claim 1 wherein the second selectively fingerprinted dataset was generated according to the same cluster size as the cluster size associated with the step of selectively fingerprinting the input dataset.
10. The method of claim 1 wherein the second selectively fingerprinted dataset was generated according to a cluster size different from the cluster size associated with the step of selectively fingerprinting the input dataset.
11. The method of claim 1 wherein the step of selecting at least one representative input fingerprint from each cluster comprises the step of:
selecting at least one characteristic input fingerprint per cluster based on the relative value of the fingerprint.
12. The method of claim 1 wherein the step of selecting at least one representative input fingerprint from each cluster comprises selecting the fingerprint with the largest value.
13. The method of claim 1 wherein the step of selecting at least one representative input fingerprint from each cluster comprises selecting the fingerprint with the smallest value.
14. The method of claim 1 wherein the step of selectively fingerprinting comprises creating fingerprints using a hashing function and wherein the fingerprints comprise hash values associated with a hash table.
15. The method of claim 1 wherein the step of selectively fingerprinting comprises creating fingerprints wherein the fingerprint comprises an index into a structure wherein the structure is selected from the list of: associative arrays, tables, trees and databases.
16. The method of claim 1 further comprising the step of adding one or more representative input fingerprints to the second selectively fingerprinted dataset, thereby creating a history and enabling improved compression for subsequent operations.
17. The method of claim 1 wherein a selective fingerprinting parameter may be adaptively configured wherein the selective fingerprinting parameter is selected from the list of: the size of a shingle, the configuration of a shingle, the size of the overlap between successive shingles, the configuration of the overlap between successive shingles, the number of fingerprints selected per cluster and the size of the cluster.
18. A system for compressing an input dataset comprising:
a matcher for selectively fingerprinting the input dataset by:
fingerprinting shingles of the input dataset wherein shingles comprise overlapping subsets of the input dataset;
selecting at least one characteristic input fingerprint per cluster based on the value of the fingerprint, thereby creating a set of characteristic input fingerprints; and,
identifying a candidate subset of the input dataset that is a match with a target subset of data in a second selectively fingerprinted dataset by searching for fingerprint matches between one or more input fingerprints from the set of characteristic input fingerprints and fingerprints associated with the second selectively fingerprinted dataset; and,
a compressor for performing a compression operations on the input dataset, thereby creating a compressed input dataset.
19. The system of claim 18 wherein the matcher selects characteristic input fingerprints per cluster based on the relative value of the fingerprint.
20. The system of claim 18 wherein the system is coupled to an input for receiving an input dataset comprising network traffic.
21. The system of claim 18 wherein the system is coupled to an output for transmitting a compressed input dataset comprising network traffic.
22. The system of claim 18 wherein the matcher is adaptively configurable.