Hi Aaron,

I don't know of any papers on this specific topic. However, I would be concerned about using such a method since you will basically be sending over the wideband output of the modulator containing high-pass shaped noise. So you firstly need a lot of bandwidth, and secondly the channel nonidealities, specifically the finite bandwidth there and in the receiver will likely cause the shaped noise to fold back into the band. This is not considering other possible headaches that may face such as the backcoupling from your Tx to the delta-sigma modulator.

However, there is no other objection in principle. After all, the bitstream from a delta-sigma modulator is quite "random", similar to real communication channels. As long as you can send and detect with a sufficiently high BER, you should be OK.

Regards,

Vivek