are carried out separately, we propose an analysis-by-synthesis approach that exploits the
properties of source codes to ensure a prescribed level of protection with manageable
computational complexity. To illustrate the power of this methodology, we show that carefully
inducing catastrophic errors in Huffman coded streams is very effective towards confusing a
passive attacker.