Ubiquitous packet networking and data compression
John Schormans · Electronics Letters · 2005
Recently clock and data synchronisation issues have prompted a rethink of how data should be communicated within embedded microprocessor systems, leading to the development of early standards for serial transmission of data using packets; these feature a symbol for delineating the end of a packet. Bounds are investigated on the potential advantage, in terms of extra data compression, if compression took place in conjunction with packetisation. The Shannon formula for a binary data source is re-derived, showing how the formula varies when compressing relatively small chunks of data into packets. Simulation results suggest that a non-trivial amount of extra compression can be obtained in this way, and that the advantage is greater for data with lower original entropy.