CRC-Based Framing

The concept of CRC-Based Framing was developed in StrataCom, Inc. in order to improve the efficiency of a pre-standard Asynchronous Transfer Mode (ATM) link protocol. This technology was ultimately used in the principal link protocols of ATM itself and was one of the most significant developments of StrataCom. A further advanced version of this was also used as part of the GFP link protocol.

Overview of CRC-Based Framing

This framing method re-uses the header CRC (Cyclic Redundancy Check) which is present in ATM and other similar protocols to provide framing on the link with no additional overhead. The header CRC is needed for other purposes with an ATM system, so reusing it for link framing provided a significant improvement in link efficiency over what would have been otherwise needed.

Invention of CRC-Based Framing

The story behind the invention of CRC-Based Framing is as follows: StrataCom produced the first commercial ATM-like product, the IPX. The IPX used 24 byte packets instead of ATM's 53 byte packets, and the field definitions were slightly different, but the basic idea of using short, fixed length packets was identical. StrataCom's first product had T1 (1.544 Mbps) based links which included a 5 bit header CRC (Cyclic Redundancy Check), similar to ATM's 8 bit header CRC.

T1 is a TDM protocol with 24 byte payloads carried in a 193 bit frame. The first bit of each frame carries one bit out of a special pattern. A receiver finds this special pattern by sequentially looking for the bit position in the receive data where a bit from this pattern shows up every 193rd byte. It was convenient for StrataCom to make the length of one cell equal to the length of one T1 frame because a useful T1 framer Integrated Circuit from Rockwell was on the market. This device found the 193 bit long TDM frame and put out the 24 bytes in a form that could be used effectively.

When it came time to produce a European product, the benefit of using 24 byte frames became a liability. The European E1 format has a 32 byte frame of which 30 bytes could carry data. The development team's first proposal used the HDLC protocol to encapsulate a sequence of 24 byte cells into a byte stream collected from the 30 byte E1 payloads. This was highly inefficient because HDLC has a heavy and data-dependent overhead. This factor may have ruined the economics of product for the european market. In the design review, everyone was lamenting what a bad solution HDLC was for the problem. Brian Holden, who was one of the reviewers present, then uttered the sentence "Why don't you just frame on the CRC?". Everyone in the room went "Oh!" as the lightbulb flashed on. Participating in this design review and uttering this one high-value sentence was Brian's only involvement in this E1 project and the consequent development of CRC-Based Framing.

The project team went on to base the framing of the E1 project on the CRC. A circuit was designed which examined the incoming byte stream emerging from the E1 framer device and found a byte position for which the header CRC value was consistently correct. The developers Rick Enns and Paul O'Hare properly got the patent for the development as Brian Holden did just utter the one sentence.

Standards Use of CRC-Based Framing

The ATM standard for several of link protocols after that point chose CRC-Based Framing for its link protocol. The ITU-T SG15 G.7041 Generic Framing Procedure is also based on an advanced version of CRC-based framing.