If you can spare 1 pin and 1 port on the main µC (I am sure you can spare on the communication µC since it has no other job), you can send/receive parallely

That way, even if you are receiving at high baud rates, communication between the 2 processors is much faster(8 times faster), and you have plenty of time...
I would suggest that the communication between the two processors be interrupt-based to maintain data concurrency

If the data cannot be forwarded as fast as it is received you can temorarily store it in buffer in ur RAM... but then u have a problem if the buffer is exceeded when the receiver never catches up.
The buffer program will be interesting to write.. it will be a FIFO, circular queue


BTW What's DPM stand for?

Kundi