Joint Variable Length Source Codes and Convolutional Codes for Text

A joint technique to reduce redundancy and provide error control is presented. The proposed technique involves the concatenation of Variable-Length Source Codes (VLSC) and convolutional codes. An algorithm for the design of a joint trellis code is proposed and examples of the design procedure are illustrated with VLSC and convolutional codes of rate ½ and variable constraint length. It is shown that this joint scheme gives a reduction in overall complexity without any loss of error performance compared to that of a disjoint trellis decoder of convolutional code and a VLSC decoder. The proposed joint design is implemented in software and its error performances are compared with those of the separate VSLC and convolutional codes as well as the corresponding uncoded schemes. Simulation results show that a coding gain of up to 6 dB can be achieved from the joint VLSC and (2,1,6) convolutional codes over the uncoded VLSC transmission.