Abstract
The bi-orthogonal codes for embedding Side-Information (SI) in data-based blind SLM (BSLM) proposed in Joo et al. (2012) produce better bit error rate (BER) and SI error rate (SIER) performance compared to binary codes. However, the authors do not provide details for code generation; instead, they list some codes with a length of U/2 and a minimum Hamming distance of U/4. The suggested bi-orthogonal code does not work for any value of the maximum iteration number U other than U = 4inverted right perpendicular U/4 inverted left perpendicular. Therefore, this paper proposes two algorithms for generating U codes for any value of U. The proposed methods maintain the normalized minimum Hamming distance between generated codes to 0. 5. However, the second proposed algorithm, which works in the case of U < left perpendicular log(2) U right perpendicular, is also able to only consider codes with a maximum Hamming distance, allowing it to improve SIER performance. Thus, the second proposed algorithm improves SIER performance by up to 1 dB at Eb/No=3dB. Furthermore, the proposed algorithms are able to generate a multiple set of U codes that deliver same performance.