摘要: This paper proposes a stateless open-digest spam fingerprinting at the packet level (layer 3) based on an algorithm Nilsimsa. Spam emails show several characteristics when viewed gateway level, which are suitable for fingerprinting: (a) content invariance and (b) recipient address dispersion. In this paper, Nilsimsa is adapted to support both fast email class estimation, per-packet basis. Email packets incrementally fingerprinted basis, without need reassembly. detection status tagged last of each email. in turn allows estimation (spam detection) receiving servers more effective handling inbound outbound (relayed) emails. The work presented focuses evaluating accuracy with consideration constraints processing byte streams over network, including reordering, fragmentation, overlapped bytes, different sizes, possibilities random addition attacks. Results that proposed packet-level can detect 100% similarity threshold set between 36 59. method gives 0% false positive true negative, equals performance attained full abstraction 7). shows classifying differentiate non-spam from high confidence viable control implementation middleboxes. Copyright © 2011 John Wiley & Sons, Ltd.