Bit-stream huffman coding for data compression
View Patent ↗Bit-stream Huffman coding may be used for data compression to quickly and efficiently compress relatively small and large datasets. A symbol used in data compression may not be a byte or 8 bits. Rather it has any number of bits. For a certain dataset, a symbol has a dynamic number of bits according to the data size. A symbol may have less than 8 bits for a small dataset, or more than 8 bits for a large dataset. For rapid processing, a large dataset may be broken into smaller datasets that are fast encoded in parallel. Accordingly, the Huffman encoding inputs from a bit-stream instead of a conventional byte-stream and outputs a bit-stream. In particular, bit-stream Static and Adaptive Huffman codings are presented with extended algorithms. Hardware implementation with parallel Huffman encoders and decoders is also illustrated for fast network data compression.
1. A method of bit-stream Huffman coding for data compression, the method comprising the steps of:
obtaining a dataset to be compressed, the dataset containing a plurality of bytes of data, each byte having a same number of bits;
defining a symbol length based on a length of the dataset, the symbol length defining a length of symbols to be used in a Huffman coding process to compress the data of the dataset, the symbol length being defined without regard to the number of bits in a byte;
converting the dataset to a bit stream;
encoding the bit stream using the Huffman coding process by extracting symbols with the defined symbol length from the bit stream and outputting Huffman codes corresponding to the extracted symbols to generate a compressed dataset.
2. The method of claim 1 , wherein the dataset contains a plurality of bytes of data, each of which is eight bits in length.
3. The method of claim 2 , wherein the symbol length is defined dynamically by assessing the length of the dataset and choosing a symbol length that will approximately 5% overhead.
4. The method of claim 3 , wherein the Huffman coding process is a static Huffman coding process and the overhead represents a storage size of a Huffman dictionary for the Static Huffman coding process.
5. The method of claim 4 , further comprising generating a compression header for the compressed dataset, the compression header including a four bit field identifying the symbol length in bits, and a 1-bit field identifying a Huffman dictionary type.
6. The method of claim 5 , further comprising generating a network protocol data unit having a protocol data unit header and payload portion, and wherein the payload portion carries the compression header and compressed dataset.
7. The method of claim 2 , wherein the Huffman coding process is an adaptive Huffman coding process and the overhead represents the storage size of Not Yet Transmitted codes for the adaptive Huffman coding process.
8. The method of claim 7 , further comprising generating a compression header for the compressed dataset, the compression header including a four bit field identifying the symbol length in bits, and a 1-bit field identifying the Huffman coding process as an adaptive Huffman coding process.
9. The method of claim 8 , further comprising generating a network protocol data unit having a protocol data unit header and payload portion, and wherein the payload portion carries the compression header and compressed dataset.
10. The method of claim 1 , wherein the Huffman coding process is a static bit-stream compression process.
11. The method of claim 1 , wherein the Huffman coding process is an adaptive bit-stream compression process.
12. The method of claim 1 , wherein the symbol length used in the Huffman coding process is longer than a byte, in bits.
13. The method of claim 1 , wherein the symbol length used in the Huffman coding process is shorter than a byte, in bits.
14. The method of claim 1 , further comprising dividing the dataset into a plurality of pieces, and wherein the steps of defining a symbol length, converting the dataset to a bit stream, and encoding the bit stream, is performed individually on each of the pieces such that the compressed dataset includes several individually compressed pieces of the original dataset.
15. The method of claim 14 , wherein a static Huffman coding process is used on the pieces such that each compressed piece of the original dataset includes its own dictionary.
16. The method of claim 1 , wherein the dataset to be compressed is the payload of a protocol data unit.
17. A network element including a bitstream Huffman encoder, the bitstream Huffman encoder comprising:
a plurality of input buffers to store pieces of data to be individually encoded;
a plurality of processors to individually implement Huffman coding processes to encode the pieces of data stored in the input buffers; and
a plurality of output buffers to store encoded pieces of data generated by the processors;
wherein each of the processors encodes a piece of data by:
assessing a length of the piece of data to be encoded;
defining a symbol length based on the length of the data to be encoded, the symbol length defining a length of symbols to be used in the Huffman coding process being used by the processor to compress the piece of data to be encoded, the symbol length being defied without regard to the number of bits in a byte; and
encoding the data to be encoded using the Huffman coding process by extracting symbols with the defined symbol length from the data to be encoded and outputting Huffman codes corresponding to the extracted symbols to generate a compressed dataset.
18. A network element including a bitstream Huffman encoder, the bitstream Huffman encoder comprising:
a deassembler to break an input dataset into the plurality of pieces of data;
a plurality of input buffers to store pieces of data to be individually encoded;
a plurality of processors to individually implement Huffman coding processes to encode the pieces of data stored in the input buffers; and
a plurality of output buffers to store encoded pieces of data generated by the processors; and
an assembler to assemble the individually encoded pieces into a compressed dataset;
wherein each of the processors encodes a piece of data by assessing a length of the data, using the length to determine a symbol length to be used by the Huffman coding process for that particular piece of data.
19. A network element including a bitstream Huffman decoder, the bitstream Huffman decoder comprising:
a plurality of input buffers to store pieces of data that have been previously individually encoded using separate Huffman encoding processes;
a plurality of processors to individually implement Huffman decoding processes to individually decode the pieces of data stored in the input buffers; and
a plurality of output buffers to store decoded pieces of data generated by the processors;
wherein each of the processors decodes a piece of data by determining from a compression header associated with the individually encoded piece of data a symbol length that was used by the Huffman encoding process to encode that particular piece of data, and then implements a Huffman decoding process to output symbols having the determined symbol length.
20. The network element of claim 19 , further comprising a deassembler to break an input dataset into the pieces of data that have been previously individually encoded, and an assembler to assemble the individually decoded pieces into a decoded dataset.