A Suboptimal Embedding Algorithm for Binary Matrix Embedding

A novel suboptimal hiding algorithm for binary data based on iterative searching embedding, ISE, is proposed. In most cases, an ML algorithm is criticized for being extremely sensitive to the dimension (n - m), due to the fact that the operation complexity varies exponentially with (n - m). Rather the complexity exhibits a linear dependence on (n - m) when performing ISE, making it applicable to a long linear code embedding. Given a (n, m, λ_min) linear embedding code, the complexity required by ISE is merely O(μ̅(n-m)) when locating the coset leader e, whilst it is O(2^(n-m)) by the ML algorithm, an unacceptable figure for a large value of (n - m). The much lower complexity in ISE is reached merely at the cost of a small deal of embedding efficiency. Various embedding efficiencies, corresponding to various random codes, are observed.