multimedia networking
Most of the Internet's traffic — a web page, an email, a downloaded file — is happy to arrive a little late as long as every byte arrives correctly and in order. Now picture a phone call or a live video: the picture and the sound must keep flowing smoothly, and a clip that arrives a second late is useless even if it is perfect. Multimedia networking is the study of carrying this kind of continuous, time-sensitive media — audio and video — across a network that was designed only to do its best, with no promises about timing. It is the art of getting a steady stream of sound and pictures across a road that has no traffic lights.
Media traffic is usually split into three classes that tolerate trouble very differently. Streaming stored media is on-demand video or audio you play from a server (a movie you start watching): a few seconds of startup delay is fine because the content already exists and can be buffered ahead. Streaming live media is a broadcast as it happens (a live sports feed): it can buffer a little, but not minutes. Conversational real-time media is a two-way call (VoIP or video conferencing): here even a small delay is felt as awkward talk-over, so it has the tightest timing budget of all. For each class the engineering question is the same — how much delay, loss, and jitter can it absorb, and what tricks hide the rest?
Why it matters: the original Internet offered only best-effort delivery — it never guaranteed when a packet would arrive or that it would arrive at all. Multimedia networking is the body of techniques that make real-time audio and video usable anyway: timestamps and sequence numbers (RTP), a client-side playout buffer to smooth out timing wobble, loss concealment to paper over missing pieces, and quality-of-service mechanisms in routers for the cases where best effort is not enough. Note the boundary: this field is about moving time-sensitive bits over the network, not about how audio and video are compressed (codecs) — that is a separate topic.
Three apps, three timing budgets. A movie on a streaming service can buffer 10 seconds ahead, so a brief network hiccup is invisible. A live game stream buffers maybe 2 to 5 seconds. A video call can tolerate only about 150 milliseconds of one-way delay before conversation feels unnatural — same network, three very different demands.
The same best-effort network serves stored, live, and conversational media with very different delay tolerances.
A common misconception is that you 'fix' real-time media by adding bandwidth. Bandwidth helps with throughput, but the real enemies of voice and video are delay and jitter — and you cannot buy your way past the speed of light.