Multi-mode dynamic code assignment for data compression

[54] MULTI-MODE DYNAMIC CODE

ASSIGNMENT FOR DATA COMPRESSION

[75] Inventor: Johannes C. Vermeulen, Longmont, Colo.

[73] Assignee: International Business Machines Corporation, Armonk, N.Y.

[21] Appl. No.: 88,771

[22] Filed: Aug. 24,1987

[51] Int. CI.4 H04B 14/06

[52] U.S. a 375/122; 360/40;

375/27; 381/31

[58] Field of Search 375/27, 122; 370/118;

382/43, 56; 358/135, 138, 261; 360/39, 40;

369/59; 381/29, 31

[56] References Cited

U.S. PATENT DOCUMENTS

4,486,784 12/1984 Abraham et al 358/261

4,494,151 1/1985 Liao 358/261

4,527,252 7/1985 Donohue et al 364/900

4,591,829 5/1986 Takeda 340/347 DD

4,608,652 8/1986 Yokokawa et al 382/56

4,633,490 12/1986 Goertzel et al 375/122

4,646,148 2/1987 Lienard et al 358/135

4,648,119 3/1987 Wingfield et al 382/56

Primary Examiner—Benedict V. Safourek
Attorney, Agent, or Firm—Carl M. Wright

[57] ABSTRACT

Bit-serial compression process improved by inclusion of character-repeat (character) mode. During the compression process in a bit serial (bit) mode, using a predictive scheme with code words for each predicted bit, the characters, typically bytes (eight bits), are assembled bit by bit. Each assembled character is compared to the preceding character. When two successive identical characters occur, the process changes to the repeatcharacter mode wherein successive identical characters are signaled in the same manner as correctly predicted bits. When a different character occurs, the process signals in a manner corresponding to an incorrectly predicted bit. Provisions are made for limitations imposed by finite code spaces, for supplying a data bit when recording a code word associated with an exhausted code space or different characted, and for indicating an identical character has occurred when recording a code word associated with an exhausted code space.

2 Claims, 8 Drawing Sheets