Novel design of arithmetic coding for data compression

The paper presents a novel software and hardware design of a universal arithmetic coding algorithm where 256 ASCII codes of different symbols, as a specific example, are included in the alphabet. Essentially, the two coding equations are modified by specifying the code values as the lower end-point value of the coding range and the width of this range. Therefore the procedures of sending output codes, solving the so-called underflow problem, and updating the coding range can be unified and simply controlled by the value of the coding range. As a result, a hardware architecture can be directly designed to implement the algorithm on real-time basis where the single operation of normalisation can be implemented in parallel. In addition, specific design of decoding the compressed output, theoretical analysis and real-time architectures of both encoding and decoding are described. Practical C source codes of main functions and experimental results are also reported