To answer your question, I reached out to Chris Diorio, the chief technology officer at chip manufacturer Impinj. Chris not only oversees the development of his company’s Electronic Product Code (EPC) integrated circuits, but was also closely involved in the creation of the EPC Gen 2 air-interface protocol specification.
Here is Chris’s response: “During the development of the Gen 2 spec, the working group considered two reader-to-tag encoding schemes: Manchester and PIE [pulse-interval encoding]. Manchester had the advantage of 50 percent duty cycle, but a more complicated decoder. PIE had the advantage of simple, counter-based decoding, but a PIE waveform doesn’t have a 50 percent duty cycle (i.e., it isn’t high 50 percent of the time and low 50 percent of the time). Depending on the ratio of 1s and 0s that the reader is transmitting, the amount of time that the waveform is high versus when it is low varies. This variation causes a ‘wander’ in the average value of the waveform that complicates the tag’s job of discriminating when the waveform is high versus when it is low, and thereby complicates the tag’s job of discriminating incoming symbols from noise. The group evaluated the DC wander, showed that it would not be a problem at tag sensitivity levels down to at least -25 dBm, and therefore chose PIE because of its simpler decoder implementation.”
—Mark Roberti, Founder and Editor, RFID Journal
How does one implement Pulse interval encoding in the design? I am unable to understand how does one creates asymmetric patterns for data -0 and data-1?