I have a USB 2.0 low-speed connection design with an unusual constraint. On the host side the D+ and D- lines have a branch, a requirement of the connector and this cannot be changed. On the device side, there is the option of either using just one side of each branch (and creating an unterminated stub on the host side) or connecting both (and forming a loop). This option is indicated by the dashed line.
Which option is better in terms of signal integrity, and how can it be improved. For example, if only the solid line on the device side was used would some kind of termination of the other host branch be helpful?
Thanks for any responses!