Home WTI User's Guides
cover
1. Introduction
2. Binary Block Read Mode and Structure
2.1. Compressed Binary Block Read Sequence
2.2. Compressed Binary Block Structure
2.2.1. Data Block Structure
2.2.2. Record String Structure
3. Compression Decompression Methods
3.1. Table Decompression
3.1.1. Nibble Positioning
3.1.2. Record String Format
3.1.2.1. Data Decompression Considerations
3.1.3. Table Decompression Example
3.2. Space Compression
4. DLE Stuffing
5. The Cyclic Redundancy Check Option
5.1. Calculation
6. Customer Service
PollCat Series

3.1. Table Decompression

 

The table compression method matches the character to be compressed with a character in the compression table. It then uses the four bit binary offset, from the first character in the table to the character that matches, to replace the original eight bit character. The table below describes the compression table and corresponding offsets.

 

If the character to be compressed does not match any of the characters in the table, an "F" HEX nibble is stored following the actual eight bit, literal character. It therefore takes three four-bit nibbles (or one and a half bytes), to store a character not found in the table.

When decompressing data, the polling device must follow the above procedure in reverse order. A four bit nibble is read from memory and checked in order to determine if it is an "F" nibble. If the four bit nibble is an "F" nibble, this nibble is discarded and the next two nibbles are read as a literal character.

If more than one half of the data is not in the table, it takes one and a half bytes (50% more) to store a character that is not in the table. Therefore, the compressed data would take up more memory than the original data. PollCat compensates for this by not table compressing data when more than one half of the characters are not found in the table. The polling device can determine if data has been table compressed by checking the table compression bit (bit 2) of the mode byte in the record header.

This provision is intended to handle exceptional data such as alpha text strings which may be transmitted as warning messages, or daily block headers, and etc.. Most data typically sent by PBX systems can be table compressed. This is because, the vast majority of characters that occur in transmitted data can usually be found in the compression table.