framing

A wire or radio link carries a continuous stream of bits, one after another, with nothing built in to say where one message ends and the next begins. It is like receiving a long telegram with no spaces or punctuation: thisisallonelongrunofletters. Framing is the trick of inserting markers into that stream so the receiver can tell where each chunk starts and stops. Each chunk, with its markers, is a frame.

There are a few standard ways to do it. One is to put a special flag byte at the start and end of every frame, for example the bit pattern 01111110; whenever the receiver sees that flag, it knows a frame boundary is there. The catch is that the same pattern might appear by accident inside the data, so the sender uses bit stuffing or byte stuffing — it inserts an extra harmless bit or escape byte to break up any data that looks like the flag, and the receiver strips it back out. Another method puts a length count in the header that says exactly how many bytes follow, so the receiver counts them off. Either way, the receiver gains a reliable answer to one question: where does this frame end?

Framing matters because every other link-layer job depends on it. You cannot check a frame for errors, read its address, or acknowledge it if you cannot first tell where it begins and ends. A common misconception is that framing adds reliability — it does not; framing only delineates. Detecting and recovering from errors are separate jobs layered on top of clean framing.