Data Compression Coding with Error Detection Capability
Tang Hong · Telecommunication Engineering · 2003
This paper proposes an arithmetic data compression coding with capability of error detection. The proposed coding uses a modified probability model, in which probability of each source symbol is reduced by factor e(0e1). In compression, the reduced part of the probability model is never used, therefore, decompressing arithmetic will never fall into the unused part if there is no error in the data. But, if error has occurred during transmission or storage, it will be detected after n symbols with probability of [1-(1-e) n]. The main advantages of the proposed coding are that redundancy can be easily adjusted by e and error detecting probability tends to 1 as n becomes infinite. Computer simulations show that more than 99.9% of errors can be detected after 10 symbols for e=0.5, which adds one bit redundancy per symbol.