A noniterative reconstruction algorithm for LFSR PRNG

Pseudo Random Number Generator (PRNG) is widely used in computer software design. The famous Domain Name System (DNS) software - BIND uses Linear Feedback Shift Register (LFSR) based PRNG algorithm to produce randomness of its transaction IDs. In recent years, DNS cache poisoning attack occurs frequently, which exploits BIND PRNG weakness, aiming at forging BIND with fake responses. In this paper, we present detailed analysis of the LFSR PRNG algorithm of BIND 9.2.4, which shows that its PRNG can be reconstructed. An effective noniterative reconstruction algorithm is proposed to fully reconstruct the internal states of the LFSRS. The algorithm is independent of the initial state of LFSR and of specific hardware platform. The experiment shows that our algorithm is more efficient than existing algorithm.