Military communications systems are subject to trauma. Certain channels fail for protracted periods of time. The red noise problem arises when some, but not all, of the channels linking a sender to a receiver become inoperative. The solutions to this problem are called pool/split/restitute processes. P/s/r processes amount to ways to encode digital messages at a sending node so as to make sure that all transmitted information gets through and is decoded correctly at the receiving node whenever at least k out of the n channels linking those two nodes remain operative. P/s/r processes are designed to work even though the sending node has no way to tell which of the channels it is using are inoperative. Ii has been known for a least two years that the encode and the decode operations in a p/s/r process are faster and simpler than those in any but the weakest and most trivial error correcting codes. Moreover the band width expansion is typically smaller in a p/s/r process than in an error correcting code adapted to do the same job. This project is aimed at producing a further orders-of-magnitude improvement in the theory of p/s/r processes. This carries over into a comparable improvement in implementing them.
