Janus general discussion forum

Janus general discussion forum

Untether the JANUS chip length from the centre frequency?

posts: 46
Another of Dale's recommendations was to have the chip length at any given centre frequency (and consequent bandwidth) determined by the channel conditions. In the experimental version 0.3, the chip length is derived from the centre frequency.

The Centre frequency choice is essentially driven by the trade-off between baud rate and range requirements. The chip length should, if similarly driven by channel physics, be determined by the reverberation time of the channel. As a rule of thumb, when the multipath decay time exceeds about half a chip length, the decoder will start to get into trouble.

Since the Centre frequency and chip length are therefore guided by independent properties of the channel and application requirements, shouldn't they be chosen independently? To maintain backward compatibility and provide a non-arbitrary default (to keep things as simple as we can) the idea is to use the current chip length derived from the centre frequency as the default, with an option for a transmitter to use a dyadic multiple of that length if the channel requires it.

Using this approach, a receiver should be able to automatically detect the chip length being used without too much difficulty, and even predict the chip length likely to be used by a transmitter, based on passive analysis of the channel properties. As the channel evolves, nodes could then adaptively switch to the most appropriate chip length to maintain low packet error rates and keep the throughput up with as little channel occupancy as possible.