NIS07-1: Efficient transmission of MP3 streams over VPNs

Multimedia applications communicating over a VPN, can leverage the security properties of a VPN with its performance drawbacks by securing only a part of its data. Selecting appropriate packets for security, however, is still an open research problem and novel methods need to be developed that can guarantee the privacy of a conversation even when a relatively large percentage of data is sent in the open. In this paper, we discuss one approach for securing MP3 audio. Here, we capitalize on the fundamental structure of data as defined in the MP3 standard and identify key fields that are necessary for decoding. We conjecture that the privacy of an MP3 stream can be guaranteed, if only these fields are secured. Since these fields are 17 to 32 bytes per MP3 frame, they constitute only 4% to 7.6% of a typical stream sampled/encoded at 44.1 Khz/128 Kbps. We prove this conjecture and demonstrate the performance gains of this approach when streaming over VPNs.