Actually it is a continuation of current concerns. We investigated by simulation some of the turbo codes proposed in the IEEE standards. This analysis will allow us to calibrate our own simulation programs and their encoding and decoding methods to verify the veracity of their results.
It is mainly for further study on RSTCs and MNBTCs, in order to find turbo codes that improve performance of the existing ones. The selection of codes will be performed in several iterative steps. In a first step we will select codes with good convergence of the decoding process. This convergence is defined as the average number of iterations required for decoding a block at a given SNR (chosen large enough so that the error rate is below 10e-5, for example). The stop criterion of the iterations, implemented, is also proposed by our research team. For the selected codes in the first step, we will perform several simulations with different interleavers, following a new selection. In this phase of the search we will consider the compatibility between the code and the interleaver. A higher stage of the search will include the analysis of the behavior in fading channels with turbo-codes selected in the second phase. And regarding this subject our research team has already some results. In the last decade, watermarking, and notably, reversible watermarking, appears as an active research topic. By watermarking, data is imperceptibly embedded into digital data. Watermarking is of interest in data security (copyrighting, fingerprinting, authentication) as well as in metadata embedding. In the framework of the project we will investigate the joint coding and watermarking. Instead of first embedding the digital data and then coding the embedded data, the embedding will be performed directly into the coded data. Codes for reversible watermarking will be developed, too. We mention that the mentor has a sound expertise in watermarking.
We intent to develop of new methods/algorithms for decoding. We have already explained several variants of the MAP algorithm applicable for MNBTCs, and we proposed a variant of the LogMAP algorithm called linear. We will investigate the possibilities of application of the algorithm LogMAP linear or its variants to the MNBTCs, hoping to simplify the complexity and also to increase the performance.
Due to the special operational significance of the interleaver in turbo-codes, we will consider adapting the existing methods of interleaving for MNBTCs, and developing new methods of interleaving, with emphasis on interleaving intrasymbol, which will serve to compensate the decreasing of the length of data block for MNBTCs.