On the Optimality of Secure Network Coding

In network communications, information trans-mission often encounters wiretapping attacks. Secure network coding is introduced to prevent information from being leaked to adversaries. The investigation of performance bounds on the numbers of source symbols and random symbols are two fundamental research problems. For an important case that each wiretap-set with cardinality not larger than r, Cai and Yeung proposed a coding scheme, which is optimal in the senses of maximizing the number of source symbols and at the same time minimizing the number of random symbols. In this letter, we further study achievable lower bound on the number of random key and show that it just depends on the security constraint, and particularly, is independent to the information amount for transmission. This implies that when the number of transmitted source message changes, we can't reduce the number of random key to keep the same security level. We further give an intuitive interpretation on our result. In addition, a similar construction of secure linear network codes is proposed, which achieves this lower bound on the number of random key no matter how much information is transmitted. At last, we also extend our result to imperfect security case.