Scaling tape systems to high areal densities results in a loss in signal-to-noise ratio of the read back signal that must be compensated by improvements in the head, media and data channel technologies. State-of-the-art linear tape recording systems employ product error-correction codes (ECC) that encode data in two-dimensional arrays, where each row is an (N1,K1) Reed-Solomon (RS) codeword referred to as a C1 codeword and each column is an (N2,K2) RS codeword referred to as a C2 codeword. In order to ensure the reliable retrieval of data under low-SNR conditions down to around 11 dB, we are developing an error-correction decoding scheme based on a Reed-Solomon product code, which is decoded with an iterative hard-decision decoding scheme that is illustrated in Fig. 1.
Figure 1. Conventional tape ECC decoding scheme shown by black arrows. Iterative decoding shown by red dashed line.
Ask the expert
IBM Research scientist