it uses registers, so you'd need some power supply and clock signal to build them. For initialization, you'd need to use a reset. something similar can be done for the receiver.
the receivers attempt to determine what the next input is based on the previous N inputs. In normal cases, after N inputs, the receivers will have seen enough of the pattern. Thus, if the receiver's "failure" output signals are held in reset (or gated off as appropriate) during the first N (or more) cycles, the receivers will have had enough time to synchronize and the false-positives will be eliminated. the alternative is to synchronously reset both transmitter and receiver and initialize them to the same seed value.
This is a little be better, but also makes the reset lines a potentially difficult to route net. the resets can be made easy if the receiver is held in reset for a "long" time, like 2N+ cycles. In that case the transmitter can also be asynchronously reset as long as the seed has certain properties like a long run of 1's. Still, it may be worth it to just synchronize the resets on the transmitters so testing is consistent on every run.