In this paper, we analyze a zero-knowledge identification scheme presented in [1], which is based on an average-case hard problem, called distributional matrix representability problem. On the contrary to the soundness property claimed in [1], we show that a simple impersonation attack is feasible.